maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/AArch64/arm64-vmul.ll

2037 lines
82 KiB
LLVM
Raw Normal View History

AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
; RUN: llc -asm-verbose=false < %s -march=arm64 -aarch64-neon-syntax=apple | FileCheck %s
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <8 x i16> @smull8h(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK-LABEL: smull8h:
;CHECK: smull.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i8>, <8 x i8>* %A
%tmp2 = load <8 x i8>, <8 x i8>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <8 x i16> @llvm.aarch64.neon.smull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp3
}
define <4 x i32> @smull4s(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: smull4s:
;CHECK: smull.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp3
}
define <2 x i64> @smull2d(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: smull2d:
;CHECK: smull.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <8 x i16> @llvm.aarch64.neon.smull.v8i16(<8 x i8>, <8 x i8>) nounwind readnone
declare <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32>, <2 x i32>) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <8 x i16> @umull8h(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK-LABEL: umull8h:
;CHECK: umull.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i8>, <8 x i8>* %A
%tmp2 = load <8 x i8>, <8 x i8>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <8 x i16> @llvm.aarch64.neon.umull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp3
}
define <4 x i32> @umull4s(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: umull4s:
;CHECK: umull.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp3
}
define <2 x i64> @umull2d(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: umull2d:
;CHECK: umull.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <8 x i16> @llvm.aarch64.neon.umull.v8i16(<8 x i8>, <8 x i8>) nounwind readnone
declare <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32>, <2 x i32>) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i32> @sqdmull4s(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmull4s:
;CHECK: sqdmull.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp3
}
define <2 x i64> @sqdmull2d(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmull2d:
;CHECK: sqdmull.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp3
}
define <4 x i32> @sqdmull2_4s(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmull2_4s:
;CHECK: sqdmull2.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <8 x i16>, <8 x i16>* %A
%load2 = load <8 x i16>, <8 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <8 x i16> %load1, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%tmp2 = shufflevector <8 x i16> %load2, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp3
}
define <2 x i64> @sqdmull2_2d(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmull2_2d:
;CHECK: sqdmull2.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <4 x i32>, <4 x i32>* %A
%load2 = load <4 x i32>, <4 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <4 x i32> %load1, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%tmp2 = shufflevector <4 x i32> %load2, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32>, <2 x i32>) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <8 x i16> @pmull8h(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK-LABEL: pmull8h:
;CHECK: pmull.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i8>, <8 x i8>* %A
%tmp2 = load <8 x i8>, <8 x i8>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <8 x i16> @llvm.aarch64.neon.pmull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <8 x i16> @llvm.aarch64.neon.pmull.v8i16(<8 x i8>, <8 x i8>) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i16> @sqdmulh_4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmulh_4h:
;CHECK: sqdmulh.4h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i16> @llvm.aarch64.neon.sqdmulh.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i16> %tmp3
}
define <8 x i16> @sqdmulh_8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmulh_8h:
;CHECK: sqdmulh.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i16>, <8 x i16>* %A
%tmp2 = load <8 x i16>, <8 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <8 x i16> @llvm.aarch64.neon.sqdmulh.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp3
}
define <2 x i32> @sqdmulh_2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmulh_2s:
;CHECK: sqdmulh.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x i32> @llvm.aarch64.neon.sqdmulh.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i32> %tmp3
}
define <4 x i32> @sqdmulh_4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmulh_4s:
;CHECK: sqdmulh.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i32>, <4 x i32>* %A
%tmp2 = load <4 x i32>, <4 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i32> @llvm.aarch64.neon.sqdmulh.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp3
}
define i32 @sqdmulh_1s(i32* %A, i32* %B) nounwind {
;CHECK-LABEL: sqdmulh_1s:
;CHECK: sqdmulh s0, {{s[0-9]+}}, {{s[0-9]+}}
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load i32, i32* %A
%tmp2 = load i32, i32* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call i32 @llvm.aarch64.neon.sqdmulh.i32(i32 %tmp1, i32 %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i32 %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <4 x i16> @llvm.aarch64.neon.sqdmulh.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <8 x i16> @llvm.aarch64.neon.sqdmulh.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <2 x i32> @llvm.aarch64.neon.sqdmulh.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
declare <4 x i32> @llvm.aarch64.neon.sqdmulh.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
declare i32 @llvm.aarch64.neon.sqdmulh.i32(i32, i32) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i16> @sqrdmulh_4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_4h:
;CHECK: sqrdmulh.4h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i16> @llvm.aarch64.neon.sqrdmulh.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i16> %tmp3
}
define <8 x i16> @sqrdmulh_8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_8h:
;CHECK: sqrdmulh.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i16>, <8 x i16>* %A
%tmp2 = load <8 x i16>, <8 x i16>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <8 x i16> @llvm.aarch64.neon.sqrdmulh.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp3
}
define <2 x i32> @sqrdmulh_2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_2s:
;CHECK: sqrdmulh.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x i32> @llvm.aarch64.neon.sqrdmulh.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i32> %tmp3
}
define <4 x i32> @sqrdmulh_4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_4s:
;CHECK: sqrdmulh.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i32>, <4 x i32>* %A
%tmp2 = load <4 x i32>, <4 x i32>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x i32> @llvm.aarch64.neon.sqrdmulh.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp3
}
define i32 @sqrdmulh_1s(i32* %A, i32* %B) nounwind {
;CHECK-LABEL: sqrdmulh_1s:
;CHECK: sqrdmulh s0, {{s[0-9]+}}, {{s[0-9]+}}
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load i32, i32* %A
%tmp2 = load i32, i32* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call i32 @llvm.aarch64.neon.sqrdmulh.i32(i32 %tmp1, i32 %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i32 %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <4 x i16> @llvm.aarch64.neon.sqrdmulh.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <8 x i16> @llvm.aarch64.neon.sqrdmulh.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <2 x i32> @llvm.aarch64.neon.sqrdmulh.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
declare <4 x i32> @llvm.aarch64.neon.sqrdmulh.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
declare i32 @llvm.aarch64.neon.sqrdmulh.i32(i32, i32) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <2 x float> @fmulx_2s(<2 x float>* %A, <2 x float>* %B) nounwind {
;CHECK-LABEL: fmulx_2s:
;CHECK: fmulx.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x float>, <2 x float>* %A
%tmp2 = load <2 x float>, <2 x float>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x float> @llvm.aarch64.neon.fmulx.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x float> %tmp3
}
define <4 x float> @fmulx_4s(<4 x float>* %A, <4 x float>* %B) nounwind {
;CHECK-LABEL: fmulx_4s:
;CHECK: fmulx.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x float>, <4 x float>* %A
%tmp2 = load <4 x float>, <4 x float>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <4 x float> @llvm.aarch64.neon.fmulx.v4f32(<4 x float> %tmp1, <4 x float> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x float> %tmp3
}
define <2 x double> @fmulx_2d(<2 x double>* %A, <2 x double>* %B) nounwind {
;CHECK-LABEL: fmulx_2d:
;CHECK: fmulx.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x double>, <2 x double>* %A
%tmp2 = load <2 x double>, <2 x double>* %B
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp3 = call <2 x double> @llvm.aarch64.neon.fmulx.v2f64(<2 x double> %tmp1, <2 x double> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x double> %tmp3
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <2 x float> @llvm.aarch64.neon.fmulx.v2f32(<2 x float>, <2 x float>) nounwind readnone
declare <4 x float> @llvm.aarch64.neon.fmulx.v4f32(<4 x float>, <4 x float>) nounwind readnone
declare <2 x double> @llvm.aarch64.neon.fmulx.v2f64(<2 x double>, <2 x double>) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i32> @smlal4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: smlal4s:
;CHECK: smlal.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = add <4 x i32> %tmp3, %tmp4
ret <4 x i32> %tmp5
}
define <2 x i64> @smlal2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: smlal2d:
;CHECK: smlal.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = add <2 x i64> %tmp3, %tmp4
ret <2 x i64> %tmp5
}
define <4 x i32> @smlsl4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: smlsl4s:
;CHECK: smlsl.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = sub <4 x i32> %tmp3, %tmp4
ret <4 x i32> %tmp5
}
define <2 x i64> @smlsl2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: smlsl2d:
;CHECK: smlsl.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = sub <2 x i64> %tmp3, %tmp4
ret <2 x i64> %tmp5
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <4 x i32> @llvm.aarch64.neon.sqadd.v4i32(<4 x i32>, <4 x i32>)
declare <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64>, <2 x i64>)
declare <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32>, <4 x i32>)
declare <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64>, <2 x i64>)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i32> @sqdmlal4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlal4s:
;CHECK: sqdmlal.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqadd.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp5
}
define <2 x i64> @sqdmlal2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlal2d:
;CHECK: sqdmlal.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp5
}
define <4 x i32> @sqdmlal2_4s(<8 x i16>* %A, <8 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlal2_4s:
;CHECK: sqdmlal2.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <8 x i16>, <8 x i16>* %A
%load2 = load <8 x i16>, <8 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <8 x i16> %load1, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%tmp2 = shufflevector <8 x i16> %load2, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqadd.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp5
}
define <2 x i64> @sqdmlal2_2d(<4 x i32>* %A, <4 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlal2_2d:
;CHECK: sqdmlal2.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <4 x i32>, <4 x i32>* %A
%load2 = load <4 x i32>, <4 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <4 x i32> %load1, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%tmp2 = shufflevector <4 x i32> %load2, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp5
}
define <4 x i32> @sqdmlsl4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlsl4s:
;CHECK: sqdmlsl.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp5
}
define <2 x i64> @sqdmlsl2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlsl2d:
;CHECK: sqdmlsl.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp5
}
define <4 x i32> @sqdmlsl2_4s(<8 x i16>* %A, <8 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlsl2_4s:
;CHECK: sqdmlsl2.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <8 x i16>, <8 x i16>* %A
%load2 = load <8 x i16>, <8 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <8 x i16> %load1, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%tmp2 = shufflevector <8 x i16> %load2, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp5
}
define <2 x i64> @sqdmlsl2_2d(<4 x i32>* %A, <4 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlsl2_2d:
;CHECK: sqdmlsl2.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <4 x i32>, <4 x i32>* %A
%load2 = load <4 x i32>, <4 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <4 x i32> %load1, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%tmp2 = shufflevector <4 x i32> %load2, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp5
}
define <4 x i32> @umlal4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: umlal4s:
;CHECK: umlal.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = add <4 x i32> %tmp3, %tmp4
ret <4 x i32> %tmp5
}
define <2 x i64> @umlal2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: umlal2d:
;CHECK: umlal.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = add <2 x i64> %tmp3, %tmp4
ret <2 x i64> %tmp5
}
define <4 x i32> @umlsl4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: umlsl4s:
;CHECK: umlsl.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = sub <4 x i32> %tmp3, %tmp4
ret <4 x i32> %tmp5
}
define <2 x i64> @umlsl2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: umlsl2d:
;CHECK: umlsl.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp5 = sub <2 x i64> %tmp3, %tmp4
ret <2 x i64> %tmp5
}
define <2 x float> @fmla_2s(<2 x float>* %A, <2 x float>* %B, <2 x float>* %C) nounwind {
;CHECK-LABEL: fmla_2s:
;CHECK: fmla.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x float>, <2 x float>* %A
%tmp2 = load <2 x float>, <2 x float>* %B
%tmp3 = load <2 x float>, <2 x float>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = call <2 x float> @llvm.fma.v2f32(<2 x float> %tmp1, <2 x float> %tmp2, <2 x float> %tmp3)
ret <2 x float> %tmp4
}
define <4 x float> @fmla_4s(<4 x float>* %A, <4 x float>* %B, <4 x float>* %C) nounwind {
;CHECK-LABEL: fmla_4s:
;CHECK: fmla.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x float>, <4 x float>* %A
%tmp2 = load <4 x float>, <4 x float>* %B
%tmp3 = load <4 x float>, <4 x float>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = call <4 x float> @llvm.fma.v4f32(<4 x float> %tmp1, <4 x float> %tmp2, <4 x float> %tmp3)
ret <4 x float> %tmp4
}
define <2 x double> @fmla_2d(<2 x double>* %A, <2 x double>* %B, <2 x double>* %C) nounwind {
;CHECK-LABEL: fmla_2d:
;CHECK: fmla.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x double>, <2 x double>* %A
%tmp2 = load <2 x double>, <2 x double>* %B
%tmp3 = load <2 x double>, <2 x double>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = call <2 x double> @llvm.fma.v2f64(<2 x double> %tmp1, <2 x double> %tmp2, <2 x double> %tmp3)
ret <2 x double> %tmp4
}
declare <2 x float> @llvm.fma.v2f32(<2 x float>, <2 x float>, <2 x float>) nounwind readnone
declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>) nounwind readnone
declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) nounwind readnone
define <2 x float> @fmls_2s(<2 x float>* %A, <2 x float>* %B, <2 x float>* %C) nounwind {
;CHECK-LABEL: fmls_2s:
;CHECK: fmls.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x float>, <2 x float>* %A
%tmp2 = load <2 x float>, <2 x float>* %B
%tmp3 = load <2 x float>, <2 x float>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = fsub <2 x float> <float -0.0, float -0.0>, %tmp2
%tmp5 = call <2 x float> @llvm.fma.v2f32(<2 x float> %tmp1, <2 x float> %tmp4, <2 x float> %tmp3)
ret <2 x float> %tmp5
}
define <4 x float> @fmls_4s(<4 x float>* %A, <4 x float>* %B, <4 x float>* %C) nounwind {
;CHECK-LABEL: fmls_4s:
;CHECK: fmls.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x float>, <4 x float>* %A
%tmp2 = load <4 x float>, <4 x float>* %B
%tmp3 = load <4 x float>, <4 x float>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %tmp2
%tmp5 = call <4 x float> @llvm.fma.v4f32(<4 x float> %tmp1, <4 x float> %tmp4, <4 x float> %tmp3)
ret <4 x float> %tmp5
}
define <2 x double> @fmls_2d(<2 x double>* %A, <2 x double>* %B, <2 x double>* %C) nounwind {
;CHECK-LABEL: fmls_2d:
;CHECK: fmls.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x double>, <2 x double>* %A
%tmp2 = load <2 x double>, <2 x double>* %B
%tmp3 = load <2 x double>, <2 x double>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = fsub <2 x double> <double -0.0, double -0.0>, %tmp2
%tmp5 = call <2 x double> @llvm.fma.v2f64(<2 x double> %tmp1, <2 x double> %tmp4, <2 x double> %tmp3)
ret <2 x double> %tmp5
}
define <2 x float> @fmls_commuted_neg_2s(<2 x float>* %A, <2 x float>* %B, <2 x float>* %C) nounwind {
;CHECK-LABEL: fmls_commuted_neg_2s:
;CHECK: fmls.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x float>, <2 x float>* %A
%tmp2 = load <2 x float>, <2 x float>* %B
%tmp3 = load <2 x float>, <2 x float>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = fsub <2 x float> <float -0.0, float -0.0>, %tmp2
%tmp5 = call <2 x float> @llvm.fma.v2f32(<2 x float> %tmp4, <2 x float> %tmp1, <2 x float> %tmp3)
ret <2 x float> %tmp5
}
define <4 x float> @fmls_commuted_neg_4s(<4 x float>* %A, <4 x float>* %B, <4 x float>* %C) nounwind {
;CHECK-LABEL: fmls_commuted_neg_4s:
;CHECK: fmls.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x float>, <4 x float>* %A
%tmp2 = load <4 x float>, <4 x float>* %B
%tmp3 = load <4 x float>, <4 x float>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %tmp2
%tmp5 = call <4 x float> @llvm.fma.v4f32(<4 x float> %tmp4, <4 x float> %tmp1, <4 x float> %tmp3)
ret <4 x float> %tmp5
}
define <2 x double> @fmls_commuted_neg_2d(<2 x double>* %A, <2 x double>* %B, <2 x double>* %C) nounwind {
;CHECK-LABEL: fmls_commuted_neg_2d:
;CHECK: fmls.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x double>, <2 x double>* %A
%tmp2 = load <2 x double>, <2 x double>* %B
%tmp3 = load <2 x double>, <2 x double>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = fsub <2 x double> <double -0.0, double -0.0>, %tmp2
%tmp5 = call <2 x double> @llvm.fma.v2f64(<2 x double> %tmp4, <2 x double> %tmp1, <2 x double> %tmp3)
ret <2 x double> %tmp5
}
define <2 x float> @fmls_indexed_2s(<2 x float> %a, <2 x float> %b, <2 x float> %c) nounwind readnone ssp {
;CHECK-LABEL: fmls_indexed_2s:
;CHECK: fmls.2s
entry:
%0 = fsub <2 x float> <float -0.000000e+00, float -0.000000e+00>, %c
%lane = shufflevector <2 x float> %b, <2 x float> undef, <2 x i32> zeroinitializer
%fmls1 = tail call <2 x float> @llvm.fma.v2f32(<2 x float> %0, <2 x float> %lane, <2 x float> %a)
ret <2 x float> %fmls1
}
define <4 x float> @fmls_indexed_4s(<4 x float> %a, <4 x float> %b, <4 x float> %c) nounwind readnone ssp {
;CHECK-LABEL: fmls_indexed_4s:
;CHECK: fmls.4s
entry:
%0 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %c
%lane = shufflevector <4 x float> %b, <4 x float> undef, <4 x i32> zeroinitializer
%fmls1 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %0, <4 x float> %lane, <4 x float> %a)
ret <4 x float> %fmls1
}
define <2 x double> @fmls_indexed_2d(<2 x double> %a, <2 x double> %b, <2 x double> %c) nounwind readnone ssp {
;CHECK-LABEL: fmls_indexed_2d:
;CHECK: fmls.2d
entry:
%0 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %c
%lane = shufflevector <2 x double> %b, <2 x double> undef, <2 x i32> zeroinitializer
%fmls1 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %0, <2 x double> %lane, <2 x double> %a)
ret <2 x double> %fmls1
}
define <2 x float> @fmla_indexed_scalar_2s(<2 x float> %a, <2 x float> %b, float %c) nounwind readnone ssp {
entry:
; CHECK-LABEL: fmla_indexed_scalar_2s:
; CHECK-NEXT: fmla.2s
; CHECK-NEXT: ret
%v1 = insertelement <2 x float> undef, float %c, i32 0
%v2 = insertelement <2 x float> %v1, float %c, i32 1
%fmla1 = tail call <2 x float> @llvm.fma.v2f32(<2 x float> %v1, <2 x float> %b, <2 x float> %a) nounwind
ret <2 x float> %fmla1
}
define <4 x float> @fmla_indexed_scalar_4s(<4 x float> %a, <4 x float> %b, float %c) nounwind readnone ssp {
entry:
; CHECK-LABEL: fmla_indexed_scalar_4s:
; CHECK-NEXT: fmla.4s
; CHECK-NEXT: ret
%v1 = insertelement <4 x float> undef, float %c, i32 0
%v2 = insertelement <4 x float> %v1, float %c, i32 1
%v3 = insertelement <4 x float> %v2, float %c, i32 2
%v4 = insertelement <4 x float> %v3, float %c, i32 3
%fmla1 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %v4, <4 x float> %b, <4 x float> %a) nounwind
ret <4 x float> %fmla1
}
define <2 x double> @fmla_indexed_scalar_2d(<2 x double> %a, <2 x double> %b, double %c) nounwind readnone ssp {
; CHECK-LABEL: fmla_indexed_scalar_2d:
; CHECK-NEXT: fmla.2d
; CHECK-NEXT: ret
entry:
%v1 = insertelement <2 x double> undef, double %c, i32 0
%v2 = insertelement <2 x double> %v1, double %c, i32 1
%fmla1 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %v2, <2 x double> %b, <2 x double> %a) nounwind
ret <2 x double> %fmla1
}
define <4 x i16> @mul_4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: mul_4h:
;CHECK-NOT: dup
;CHECK: mul.4h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
%tmp4 = mul <4 x i16> %tmp1, %tmp3
ret <4 x i16> %tmp4
}
define <8 x i16> @mul_8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: mul_8h:
;CHECK-NOT: dup
;CHECK: mul.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i16>, <8 x i16>* %A
%tmp2 = load <8 x i16>, <8 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> %tmp2, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%tmp4 = mul <8 x i16> %tmp1, %tmp3
ret <8 x i16> %tmp4
}
define <2 x i32> @mul_2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: mul_2s:
;CHECK-NOT: dup
;CHECK: mul.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
%tmp4 = mul <2 x i32> %tmp1, %tmp3
ret <2 x i32> %tmp4
}
define <4 x i32> @mul_4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: mul_4s:
;CHECK-NOT: dup
;CHECK: mul.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i32>, <4 x i32>* %A
%tmp2 = load <4 x i32>, <4 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i32> %tmp2, <4 x i32> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
%tmp4 = mul <4 x i32> %tmp1, %tmp3
ret <4 x i32> %tmp4
}
define <2 x i64> @mul_2d(<2 x i64> %A, <2 x i64> %B) nounwind {
; CHECK-LABEL: mul_2d:
; CHECK: mul
; CHECK: mul
%tmp1 = mul <2 x i64> %A, %B
ret <2 x i64> %tmp1
}
define <2 x float> @fmul_lane_2s(<2 x float>* %A, <2 x float>* %B) nounwind {
;CHECK-LABEL: fmul_lane_2s:
;CHECK-NOT: dup
;CHECK: fmul.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x float>, <2 x float>* %A
%tmp2 = load <2 x float>, <2 x float>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x float> %tmp2, <2 x float> %tmp2, <2 x i32> <i32 1, i32 1>
%tmp4 = fmul <2 x float> %tmp1, %tmp3
ret <2 x float> %tmp4
}
define <4 x float> @fmul_lane_4s(<4 x float>* %A, <4 x float>* %B) nounwind {
;CHECK-LABEL: fmul_lane_4s:
;CHECK-NOT: dup
;CHECK: fmul.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x float>, <4 x float>* %A
%tmp2 = load <4 x float>, <4 x float>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x float> %tmp2, <4 x float> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
%tmp4 = fmul <4 x float> %tmp1, %tmp3
ret <4 x float> %tmp4
}
define <2 x double> @fmul_lane_2d(<2 x double>* %A, <2 x double>* %B) nounwind {
;CHECK-LABEL: fmul_lane_2d:
;CHECK-NOT: dup
;CHECK: fmul.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x double>, <2 x double>* %A
%tmp2 = load <2 x double>, <2 x double>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x double> %tmp2, <2 x double> %tmp2, <2 x i32> <i32 1, i32 1>
%tmp4 = fmul <2 x double> %tmp1, %tmp3
ret <2 x double> %tmp4
}
define float @fmul_lane_s(float %A, <4 x float> %vec) nounwind {
;CHECK-LABEL: fmul_lane_s:
;CHECK-NOT: dup
;CHECK: fmul.s s0, s0, v1[3]
%B = extractelement <4 x float> %vec, i32 3
%res = fmul float %A, %B
ret float %res
}
define double @fmul_lane_d(double %A, <2 x double> %vec) nounwind {
;CHECK-LABEL: fmul_lane_d:
;CHECK-NOT: dup
;CHECK: fmul.d d0, d0, v1[1]
%B = extractelement <2 x double> %vec, i32 1
%res = fmul double %A, %B
ret double %res
}
define <2 x float> @fmulx_lane_2s(<2 x float>* %A, <2 x float>* %B) nounwind {
;CHECK-LABEL: fmulx_lane_2s:
;CHECK-NOT: dup
;CHECK: fmulx.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x float>, <2 x float>* %A
%tmp2 = load <2 x float>, <2 x float>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x float> %tmp2, <2 x float> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x float> @llvm.aarch64.neon.fmulx.v2f32(<2 x float> %tmp1, <2 x float> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x float> %tmp4
}
define <4 x float> @fmulx_lane_4s(<4 x float>* %A, <4 x float>* %B) nounwind {
;CHECK-LABEL: fmulx_lane_4s:
;CHECK-NOT: dup
;CHECK: fmulx.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x float>, <4 x float>* %A
%tmp2 = load <4 x float>, <4 x float>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x float> %tmp2, <4 x float> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x float> @llvm.aarch64.neon.fmulx.v4f32(<4 x float> %tmp1, <4 x float> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x float> %tmp4
}
define <2 x double> @fmulx_lane_2d(<2 x double>* %A, <2 x double>* %B) nounwind {
;CHECK-LABEL: fmulx_lane_2d:
;CHECK-NOT: dup
;CHECK: fmulx.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x double>, <2 x double>* %A
%tmp2 = load <2 x double>, <2 x double>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x double> %tmp2, <2 x double> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x double> @llvm.aarch64.neon.fmulx.v2f64(<2 x double> %tmp1, <2 x double> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x double> %tmp4
}
define <4 x i16> @sqdmulh_lane_4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmulh_lane_4h:
;CHECK-NOT: dup
;CHECK: sqdmulh.4h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i16> @llvm.aarch64.neon.sqdmulh.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i16> %tmp4
}
define <8 x i16> @sqdmulh_lane_8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmulh_lane_8h:
;CHECK-NOT: dup
;CHECK: sqdmulh.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i16>, <8 x i16>* %A
%tmp2 = load <8 x i16>, <8 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> %tmp2, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <8 x i16> @llvm.aarch64.neon.sqdmulh.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp4
}
define <2 x i32> @sqdmulh_lane_2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmulh_lane_2s:
;CHECK-NOT: dup
;CHECK: sqdmulh.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i32> @llvm.aarch64.neon.sqdmulh.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i32> %tmp4
}
define <4 x i32> @sqdmulh_lane_4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmulh_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmulh.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i32>, <4 x i32>* %A
%tmp2 = load <4 x i32>, <4 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i32> %tmp2, <4 x i32> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmulh.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp4
}
define i32 @sqdmulh_lane_1s(i32 %A, <4 x i32> %B) nounwind {
;CHECK-LABEL: sqdmulh_lane_1s:
;CHECK-NOT: dup
;CHECK: sqdmulh.s s0, {{s[0-9]+}}, {{v[0-9]+}}[1]
%tmp1 = extractelement <4 x i32> %B, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp2 = call i32 @llvm.aarch64.neon.sqdmulh.i32(i32 %A, i32 %tmp1)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i32 %tmp2
}
define <4 x i16> @sqrdmulh_lane_4h(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_lane_4h:
;CHECK-NOT: dup
;CHECK: sqrdmulh.4h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i16> @llvm.aarch64.neon.sqrdmulh.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i16> %tmp4
}
define <8 x i16> @sqrdmulh_lane_8h(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_lane_8h:
;CHECK-NOT: dup
;CHECK: sqrdmulh.8h
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <8 x i16>, <8 x i16>* %A
%tmp2 = load <8 x i16>, <8 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> %tmp2, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <8 x i16> @llvm.aarch64.neon.sqrdmulh.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %tmp4
}
define <2 x i32> @sqrdmulh_lane_2s(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_lane_2s:
;CHECK-NOT: dup
;CHECK: sqrdmulh.2s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i32> @llvm.aarch64.neon.sqrdmulh.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i32> %tmp4
}
define <4 x i32> @sqrdmulh_lane_4s(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: sqrdmulh_lane_4s:
;CHECK-NOT: dup
;CHECK: sqrdmulh.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i32>, <4 x i32>* %A
%tmp2 = load <4 x i32>, <4 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i32> %tmp2, <4 x i32> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqrdmulh.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp4
}
define i32 @sqrdmulh_lane_1s(i32 %A, <4 x i32> %B) nounwind {
;CHECK-LABEL: sqrdmulh_lane_1s:
;CHECK-NOT: dup
;CHECK: sqrdmulh.s s0, {{s[0-9]+}}, {{v[0-9]+}}[1]
%tmp1 = extractelement <4 x i32> %B, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp2 = call i32 @llvm.aarch64.neon.sqrdmulh.i32(i32 %A, i32 %tmp1)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i32 %tmp2
}
define <4 x i32> @sqdmull_lane_4s(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmull_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmull.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp4
}
define <2 x i64> @sqdmull_lane_2d(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmull_lane_2d:
;CHECK-NOT: dup
;CHECK: sqdmull.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp4
}
define <4 x i32> @sqdmull2_lane_4s(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: sqdmull2_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmull2.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <8 x i16>, <8 x i16>* %A
%load2 = load <8 x i16>, <8 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <8 x i16> %load1, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%tmp2 = shufflevector <8 x i16> %load2, <8 x i16> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp4
}
define <2 x i64> @sqdmull2_lane_2d(<4 x i32>* %A, <4 x i32>* %B) nounwind {
;CHECK-LABEL: sqdmull2_lane_2d:
;CHECK-NOT: dup
;CHECK: sqdmull2.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <4 x i32>, <4 x i32>* %A
%load2 = load <4 x i32>, <4 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <4 x i32> %load1, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%tmp2 = shufflevector <4 x i32> %load2, <4 x i32> undef, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp4
}
define <4 x i32> @umull_lane_4s(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: umull_lane_4s:
;CHECK-NOT: dup
;CHECK: umull.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp4
}
define <2 x i64> @umull_lane_2d(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: umull_lane_2d:
;CHECK-NOT: dup
;CHECK: umull.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp4
}
define <4 x i32> @smull_lane_4s(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK-LABEL: smull_lane_4s:
;CHECK-NOT: dup
;CHECK: smull.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp4
}
define <2 x i64> @smull_lane_2d(<2 x i32>* %A, <2 x i32>* %B) nounwind {
;CHECK-LABEL: smull_lane_2d:
;CHECK-NOT: dup
;CHECK: smull.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp4
}
define <4 x i32> @smlal_lane_4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: smlal_lane_4s:
;CHECK-NOT: dup
;CHECK: smlal.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = add <4 x i32> %tmp3, %tmp5
ret <4 x i32> %tmp6
}
define <2 x i64> @smlal_lane_2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: smlal_lane_2d:
;CHECK-NOT: dup
;CHECK: smlal.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = add <2 x i64> %tmp3, %tmp5
ret <2 x i64> %tmp6
}
define <4 x i32> @sqdmlal_lane_4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlal_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmlal.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp4)
%tmp6 = call <4 x i32> @llvm.aarch64.neon.sqadd.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp6
}
define <2 x i64> @sqdmlal_lane_2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlal_lane_2d:
;CHECK-NOT: dup
;CHECK: sqdmlal.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp4)
%tmp6 = call <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp6
}
define <4 x i32> @sqdmlal2_lane_4s(<8 x i16>* %A, <8 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlal2_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmlal2.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <8 x i16>, <8 x i16>* %A
%load2 = load <8 x i16>, <8 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <8 x i16> %load1, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%tmp2 = shufflevector <8 x i16> %load2, <8 x i16> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
%tmp6 = call <4 x i32> @llvm.aarch64.neon.sqadd.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp6
}
define <2 x i64> @sqdmlal2_lane_2d(<4 x i32>* %A, <4 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlal2_lane_2d:
;CHECK-NOT: dup
;CHECK: sqdmlal2.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <4 x i32>, <4 x i32>* %A
%load2 = load <4 x i32>, <4 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <4 x i32> %load1, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%tmp2 = shufflevector <4 x i32> %load2, <4 x i32> undef, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
%tmp6 = call <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp6
}
define i32 @sqdmlal_lane_1s(i32 %A, i16 %B, <4 x i16> %C) nounwind {
;CHECK-LABEL: sqdmlal_lane_1s:
;CHECK: sqdmlal.4s
%lhs = insertelement <4 x i16> undef, i16 %B, i32 0
%rhs = shufflevector <4 x i16> %C, <4 x i16> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%prod.vec = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %lhs, <4 x i16> %rhs)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%prod = extractelement <4 x i32> %prod.vec, i32 0
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = call i32 @llvm.aarch64.neon.sqadd.i32(i32 %A, i32 %prod)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i32 %res
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare i32 @llvm.aarch64.neon.sqadd.i32(i32, i32)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define i32 @sqdmlsl_lane_1s(i32 %A, i16 %B, <4 x i16> %C) nounwind {
;CHECK-LABEL: sqdmlsl_lane_1s:
;CHECK: sqdmlsl.4s
%lhs = insertelement <4 x i16> undef, i16 %B, i32 0
%rhs = shufflevector <4 x i16> %C, <4 x i16> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%prod.vec = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %lhs, <4 x i16> %rhs)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%prod = extractelement <4 x i32> %prod.vec, i32 0
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = call i32 @llvm.aarch64.neon.sqsub.i32(i32 %A, i32 %prod)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i32 %res
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare i32 @llvm.aarch64.neon.sqsub.i32(i32, i32)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define i64 @sqdmlal_lane_1d(i64 %A, i32 %B, <2 x i32> %C) nounwind {
;CHECK-LABEL: sqdmlal_lane_1d:
;CHECK: sqdmlal.s
%rhs = extractelement <2 x i32> %C, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%prod = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %B, i32 %rhs)
%res = call i64 @llvm.aarch64.neon.sqadd.i64(i64 %A, i64 %prod)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i64 %res
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare i64 @llvm.aarch64.neon.sqdmulls.scalar(i32, i32)
declare i64 @llvm.aarch64.neon.sqadd.i64(i64, i64)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define i64 @sqdmlsl_lane_1d(i64 %A, i32 %B, <2 x i32> %C) nounwind {
;CHECK-LABEL: sqdmlsl_lane_1d:
;CHECK: sqdmlsl.s
%rhs = extractelement <2 x i32> %C, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%prod = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %B, i32 %rhs)
%res = call i64 @llvm.aarch64.neon.sqsub.i64(i64 %A, i64 %prod)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret i64 %res
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare i64 @llvm.aarch64.neon.sqsub.i64(i64, i64)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i32> @umlal_lane_4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: umlal_lane_4s:
;CHECK-NOT: dup
;CHECK: umlal.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = add <4 x i32> %tmp3, %tmp5
ret <4 x i32> %tmp6
}
define <2 x i64> @umlal_lane_2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: umlal_lane_2d:
;CHECK-NOT: dup
;CHECK: umlal.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = add <2 x i64> %tmp3, %tmp5
ret <2 x i64> %tmp6
}
define <4 x i32> @smlsl_lane_4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: smlsl_lane_4s:
;CHECK-NOT: dup
;CHECK: smlsl.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = sub <4 x i32> %tmp3, %tmp5
ret <4 x i32> %tmp6
}
define <2 x i64> @smlsl_lane_2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: smlsl_lane_2d:
;CHECK-NOT: dup
;CHECK: smlsl.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = sub <2 x i64> %tmp3, %tmp5
ret <2 x i64> %tmp6
}
define <4 x i32> @sqdmlsl_lane_4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlsl_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmlsl.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp4)
%tmp6 = call <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp6
}
define <2 x i64> @sqdmlsl_lane_2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlsl_lane_2d:
;CHECK-NOT: dup
;CHECK: sqdmlsl.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp4)
%tmp6 = call <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp6
}
define <4 x i32> @sqdmlsl2_lane_4s(<8 x i16>* %A, <8 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: sqdmlsl2_lane_4s:
;CHECK-NOT: dup
;CHECK: sqdmlsl2.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <8 x i16>, <8 x i16>* %A
%load2 = load <8 x i16>, <8 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <8 x i16> %load1, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%tmp2 = shufflevector <8 x i16> %load2, <8 x i16> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
%tmp6 = call <4 x i32> @llvm.aarch64.neon.sqsub.v4i32(<4 x i32> %tmp3, <4 x i32> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %tmp6
}
define <2 x i64> @sqdmlsl2_lane_2d(<4 x i32>* %A, <4 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: sqdmlsl2_lane_2d:
;CHECK-NOT: dup
;CHECK: sqdmlsl2.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%load1 = load <4 x i32>, <4 x i32>* %A
%load2 = load <4 x i32>, <4 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp1 = shufflevector <4 x i32> %load1, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%tmp2 = shufflevector <4 x i32> %load2, <4 x i32> undef, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
%tmp6 = call <2 x i64> @llvm.aarch64.neon.sqsub.v2i64(<2 x i64> %tmp3, <2 x i64> %tmp5)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %tmp6
}
define <4 x i32> @umlsl_lane_4s(<4 x i16>* %A, <4 x i16>* %B, <4 x i32>* %C) nounwind {
;CHECK-LABEL: umlsl_lane_4s:
;CHECK-NOT: dup
;CHECK: umlsl.4s
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <4 x i16>, <4 x i16>* %A
%tmp2 = load <4 x i16>, <4 x i16>* %B
%tmp3 = load <4 x i32>, <4 x i32>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <4 x i16> %tmp2, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = sub <4 x i32> %tmp3, %tmp5
ret <4 x i32> %tmp6
}
define <2 x i64> @umlsl_lane_2d(<2 x i32>* %A, <2 x i32>* %B, <2 x i64>* %C) nounwind {
;CHECK-LABEL: umlsl_lane_2d:
;CHECK-NOT: dup
;CHECK: umlsl.2d
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%tmp1 = load <2 x i32>, <2 x i32>* %A
%tmp2 = load <2 x i32>, <2 x i32>* %B
%tmp3 = load <2 x i64>, <2 x i64>* %C
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp4 = shufflevector <2 x i32> %tmp2, <2 x i32> %tmp2, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp5 = call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp4)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%tmp6 = sub <2 x i64> %tmp3, %tmp5
ret <2 x i64> %tmp6
}
; Scalar FMULX
define float @fmulxs(float %a, float %b) nounwind {
; CHECK-LABEL: fmulxs:
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: fmulx s0, s0, s1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%fmulx.i = tail call float @llvm.aarch64.neon.fmulx.f32(float %a, float %b) nounwind
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: ret
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret float %fmulx.i
}
define double @fmulxd(double %a, double %b) nounwind {
; CHECK-LABEL: fmulxd:
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: fmulx d0, d0, d1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%fmulx.i = tail call double @llvm.aarch64.neon.fmulx.f64(double %a, double %b) nounwind
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: ret
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret double %fmulx.i
}
define float @fmulxs_lane(float %a, <4 x float> %vec) nounwind {
; CHECK-LABEL: fmulxs_lane:
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: fmulx.s s0, s0, v1[3]
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%b = extractelement <4 x float> %vec, i32 3
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%fmulx.i = tail call float @llvm.aarch64.neon.fmulx.f32(float %a, float %b) nounwind
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: ret
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret float %fmulx.i
}
define double @fmulxd_lane(double %a, <2 x double> %vec) nounwind {
; CHECK-LABEL: fmulxd_lane:
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: fmulx d0, d0, v1[1]
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%b = extractelement <2 x double> %vec, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%fmulx.i = tail call double @llvm.aarch64.neon.fmulx.f64(double %a, double %b) nounwind
Revert part of "Fix broken FileCheck prefixes" This reverts part of commit r209538. llvm-svn: 209544
2014-05-24 03:33:49 +08:00
; CHECKNEXT: ret
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret double %fmulx.i
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare double @llvm.aarch64.neon.fmulx.f64(double, double) nounwind readnone
declare float @llvm.aarch64.neon.fmulx.f32(float, float) nounwind readnone
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <8 x i16> @smull2_8h_simple(<16 x i8> %a, <16 x i8> %b) nounwind {
; CHECK-LABEL: smull2_8h_simple:
; CHECK-NEXT: smull2.8h v0, v0, v1
; CHECK-NEXT: ret
%1 = shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
%2 = shufflevector <16 x i8> %b, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%3 = tail call <8 x i16> @llvm.aarch64.neon.smull.v8i16(<8 x i8> %1, <8 x i8> %2) #2
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %3
}
define <8 x i16> @foo0(<16 x i8> %a, <16 x i8> %b) nounwind {
; CHECK-LABEL: foo0:
; CHECK: smull2.8h v0, v0, v1
%tmp = bitcast <16 x i8> %a to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <8 x i8>
%tmp2 = bitcast <16 x i8> %b to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <8 x i8>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull.i.i = tail call <8 x i16> @llvm.aarch64.neon.smull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %vmull.i.i
}
define <4 x i32> @foo1(<8 x i16> %a, <8 x i16> %b) nounwind {
; CHECK-LABEL: foo1:
; CHECK: smull2.4s v0, v0, v1
%tmp = bitcast <8 x i16> %a to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <4 x i16>
%tmp2 = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <4 x i16>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i.i
}
define <2 x i64> @foo2(<4 x i32> %a, <4 x i32> %b) nounwind {
; CHECK-LABEL: foo2:
; CHECK: smull2.2d v0, v0, v1
%tmp = bitcast <4 x i32> %a to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <2 x i32>
%tmp2 = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <2 x i32>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i.i
}
define <8 x i16> @foo3(<16 x i8> %a, <16 x i8> %b) nounwind {
; CHECK-LABEL: foo3:
; CHECK: umull2.8h v0, v0, v1
%tmp = bitcast <16 x i8> %a to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <8 x i8>
%tmp2 = bitcast <16 x i8> %b to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <8 x i8>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull.i.i = tail call <8 x i16> @llvm.aarch64.neon.umull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %vmull.i.i
}
define <4 x i32> @foo4(<8 x i16> %a, <8 x i16> %b) nounwind {
; CHECK-LABEL: foo4:
; CHECK: umull2.4s v0, v0, v1
%tmp = bitcast <8 x i16> %a to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <4 x i16>
%tmp2 = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <4 x i16>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i.i
}
define <2 x i64> @foo5(<4 x i32> %a, <4 x i32> %b) nounwind {
; CHECK-LABEL: foo5:
; CHECK: umull2.2d v0, v0, v1
%tmp = bitcast <4 x i32> %a to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <2 x i32>
%tmp2 = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <2 x i32>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i.i
}
define <4 x i32> @foo6(<4 x i32> %a, <8 x i16> %b, <4 x i16> %c) nounwind readnone optsize ssp {
; CHECK-LABEL: foo6:
; CHECK-NEXT: smull2.4s v0, v1, v2[1]
; CHECK-NEXT: ret
entry:
%0 = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i to <4 x i16>
%shuffle = shufflevector <4 x i16> %c, <4 x i16> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %1, <4 x i16> %shuffle) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i
}
define <2 x i64> @foo7(<2 x i64> %a, <4 x i32> %b, <2 x i32> %c) nounwind readnone optsize ssp {
; CHECK-LABEL: foo7:
; CHECK-NEXT: smull2.2d v0, v1, v2[1]
; CHECK-NEXT: ret
entry:
%0 = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i to <2 x i32>
%shuffle = shufflevector <2 x i32> %c, <2 x i32> undef, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %1, <2 x i32> %shuffle) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i
}
define <4 x i32> @foo8(<4 x i32> %a, <8 x i16> %b, <4 x i16> %c) nounwind readnone optsize ssp {
; CHECK-LABEL: foo8:
; CHECK-NEXT: umull2.4s v0, v1, v2[1]
; CHECK-NEXT: ret
entry:
%0 = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i to <4 x i16>
%shuffle = shufflevector <4 x i16> %c, <4 x i16> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %1, <4 x i16> %shuffle) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i
}
define <2 x i64> @foo9(<2 x i64> %a, <4 x i32> %b, <2 x i32> %c) nounwind readnone optsize ssp {
; CHECK-LABEL: foo9:
; CHECK-NEXT: umull2.2d v0, v1, v2[1]
; CHECK-NEXT: ret
entry:
%0 = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i to <2 x i32>
%shuffle = shufflevector <2 x i32> %c, <2 x i32> undef, <2 x i32> <i32 1, i32 1>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %1, <2 x i32> %shuffle) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i
}
define <8 x i16> @bar0(<8 x i16> %a, <16 x i8> %b, <16 x i8> %c) nounwind {
; CHECK-LABEL: bar0:
; CHECK: smlal2.8h v0, v1, v2
; CHECK-NEXT: ret
%tmp = bitcast <16 x i8> %b to <2 x i64>
%shuffle.i.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i.i to <8 x i8>
%tmp2 = bitcast <16 x i8> %c to <2 x i64>
%shuffle.i3.i.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i.i to <8 x i8>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull.i.i.i = tail call <8 x i16> @llvm.aarch64.neon.smull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add.i = add <8 x i16> %vmull.i.i.i, %a
ret <8 x i16> %add.i
}
define <4 x i32> @bar1(<4 x i32> %a, <8 x i16> %b, <8 x i16> %c) nounwind {
; CHECK-LABEL: bar1:
; CHECK: smlal2.4s v0, v1, v2
; CHECK-NEXT: ret
%tmp = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i.i to <4 x i16>
%tmp2 = bitcast <8 x i16> %c to <2 x i64>
%shuffle.i3.i.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i.i to <4 x i16>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i.i = tail call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add.i = add <4 x i32> %vmull2.i.i.i, %a
ret <4 x i32> %add.i
}
define <2 x i64> @bar2(<2 x i64> %a, <4 x i32> %b, <4 x i32> %c) nounwind {
; CHECK-LABEL: bar2:
; CHECK: smlal2.2d v0, v1, v2
; CHECK-NEXT: ret
%tmp = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i.i to <2 x i32>
%tmp2 = bitcast <4 x i32> %c to <2 x i64>
%shuffle.i3.i.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i.i to <2 x i32>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i.i = tail call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add.i = add <2 x i64> %vmull2.i.i.i, %a
ret <2 x i64> %add.i
}
define <8 x i16> @bar3(<8 x i16> %a, <16 x i8> %b, <16 x i8> %c) nounwind {
; CHECK-LABEL: bar3:
; CHECK: umlal2.8h v0, v1, v2
; CHECK-NEXT: ret
%tmp = bitcast <16 x i8> %b to <2 x i64>
%shuffle.i.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i.i to <8 x i8>
%tmp2 = bitcast <16 x i8> %c to <2 x i64>
%shuffle.i3.i.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i.i to <8 x i8>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull.i.i.i = tail call <8 x i16> @llvm.aarch64.neon.umull.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add.i = add <8 x i16> %vmull.i.i.i, %a
ret <8 x i16> %add.i
}
define <4 x i32> @bar4(<4 x i32> %a, <8 x i16> %b, <8 x i16> %c) nounwind {
; CHECK-LABEL: bar4:
; CHECK: umlal2.4s v0, v1, v2
; CHECK-NEXT: ret
%tmp = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i.i to <4 x i16>
%tmp2 = bitcast <8 x i16> %c to <2 x i64>
%shuffle.i3.i.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i.i to <4 x i16>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i.i = tail call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add.i = add <4 x i32> %vmull2.i.i.i, %a
ret <4 x i32> %add.i
}
define <2 x i64> @bar5(<2 x i64> %a, <4 x i32> %b, <4 x i32> %c) nounwind {
; CHECK-LABEL: bar5:
; CHECK: umlal2.2d v0, v1, v2
; CHECK-NEXT: ret
%tmp = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i.i to <2 x i32>
%tmp2 = bitcast <4 x i32> %c to <2 x i64>
%shuffle.i3.i.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i.i to <2 x i32>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i.i = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add.i = add <2 x i64> %vmull2.i.i.i, %a
ret <2 x i64> %add.i
}
define <4 x i32> @mlal2_1(<4 x i32> %a, <8 x i16> %b, <4 x i16> %c) nounwind {
; CHECK-LABEL: mlal2_1:
; CHECK: smlal2.4s v0, v1, v2[3]
; CHECK-NEXT: ret
%shuffle = shufflevector <4 x i16> %c, <4 x i16> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
%tmp = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <4 x i16>
%tmp2 = bitcast <8 x i16> %shuffle to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <4 x i16>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add = add <4 x i32> %vmull2.i.i, %a
ret <4 x i32> %add
}
define <2 x i64> @mlal2_2(<2 x i64> %a, <4 x i32> %b, <2 x i32> %c) nounwind {
; CHECK-LABEL: mlal2_2:
; CHECK: smlal2.2d v0, v1, v2[1]
; CHECK-NEXT: ret
%shuffle = shufflevector <2 x i32> %c, <2 x i32> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
%tmp = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <2 x i32>
%tmp2 = bitcast <4 x i32> %shuffle to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <2 x i32>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add = add <2 x i64> %vmull2.i.i, %a
ret <2 x i64> %add
}
define <4 x i32> @mlal2_4(<4 x i32> %a, <8 x i16> %b, <4 x i16> %c) nounwind {
; CHECK-LABEL: mlal2_4:
; CHECK: umlal2.4s v0, v1, v2[2]
; CHECK-NEXT: ret
%shuffle = shufflevector <4 x i16> %c, <4 x i16> undef, <8 x i32> <i32 2, i32 2, i32 2, i32 2, i32 2, i32 2, i32 2, i32 2>
%tmp = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <4 x i16>
%tmp2 = bitcast <8 x i16> %shuffle to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <4 x i16>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add = add <4 x i32> %vmull2.i.i, %a
ret <4 x i32> %add
}
define <2 x i64> @mlal2_5(<2 x i64> %a, <4 x i32> %b, <2 x i32> %c) nounwind {
; CHECK-LABEL: mlal2_5:
; CHECK: umlal2.2d v0, v1, v2[0]
; CHECK-NEXT: ret
%shuffle = shufflevector <2 x i32> %c, <2 x i32> undef, <4 x i32> zeroinitializer
%tmp = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %tmp, <2 x i64> undef, <1 x i32> <i32 1>
%tmp1 = bitcast <1 x i64> %shuffle.i.i to <2 x i32>
%tmp2 = bitcast <4 x i32> %shuffle to <2 x i64>
%shuffle.i3.i = shufflevector <2 x i64> %tmp2, <2 x i64> undef, <1 x i32> <i32 1>
%tmp3 = bitcast <1 x i64> %shuffle.i3.i to <2 x i32>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp3) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%add = add <2 x i64> %vmull2.i.i, %a
ret <2 x i64> %add
}
; rdar://12328502
define <2 x double> @vmulq_n_f64(<2 x double> %x, double %y) nounwind readnone ssp {
entry:
; CHECK-LABEL: vmulq_n_f64:
; CHECK-NOT: dup.2d
; CHECK: fmul.2d v0, v0, v1[0]
%vecinit.i = insertelement <2 x double> undef, double %y, i32 0
%vecinit1.i = insertelement <2 x double> %vecinit.i, double %y, i32 1
%mul.i = fmul <2 x double> %vecinit1.i, %x
ret <2 x double> %mul.i
}
define <4 x float> @vmulq_n_f32(<4 x float> %x, float %y) nounwind readnone ssp {
entry:
; CHECK-LABEL: vmulq_n_f32:
; CHECK-NOT: dup.4s
; CHECK: fmul.4s v0, v0, v1[0]
%vecinit.i = insertelement <4 x float> undef, float %y, i32 0
%vecinit1.i = insertelement <4 x float> %vecinit.i, float %y, i32 1
%vecinit2.i = insertelement <4 x float> %vecinit1.i, float %y, i32 2
%vecinit3.i = insertelement <4 x float> %vecinit2.i, float %y, i32 3
%mul.i = fmul <4 x float> %vecinit3.i, %x
ret <4 x float> %mul.i
}
define <2 x float> @vmul_n_f32(<2 x float> %x, float %y) nounwind readnone ssp {
entry:
; CHECK-LABEL: vmul_n_f32:
; CHECK-NOT: dup.2s
; CHECK: fmul.2s v0, v0, v1[0]
%vecinit.i = insertelement <2 x float> undef, float %y, i32 0
%vecinit1.i = insertelement <2 x float> %vecinit.i, float %y, i32 1
%mul.i = fmul <2 x float> %vecinit1.i, %x
ret <2 x float> %mul.i
}
define <4 x i16> @vmla_laneq_s16_test(<4 x i16> %a, <4 x i16> %b, <8 x i16> %c) nounwind readnone ssp {
entry:
; CHECK: vmla_laneq_s16_test
; CHECK-NOT: ext
; CHECK: mla.4h v0, v1, v2[6]
; CHECK-NEXT: ret
%shuffle = shufflevector <8 x i16> %c, <8 x i16> undef, <4 x i32> <i32 6, i32 6, i32 6, i32 6>
%mul = mul <4 x i16> %shuffle, %b
%add = add <4 x i16> %mul, %a
ret <4 x i16> %add
}
define <2 x i32> @vmla_laneq_s32_test(<2 x i32> %a, <2 x i32> %b, <4 x i32> %c) nounwind readnone ssp {
entry:
; CHECK: vmla_laneq_s32_test
; CHECK-NOT: ext
; CHECK: mla.2s v0, v1, v2[3]
; CHECK-NEXT: ret
%shuffle = shufflevector <4 x i32> %c, <4 x i32> undef, <2 x i32> <i32 3, i32 3>
%mul = mul <2 x i32> %shuffle, %b
%add = add <2 x i32> %mul, %a
ret <2 x i32> %add
}
ARM64: add 128-bit MLA operations to the custom selection code. Without this change, the llvm_unreachable kicked in. The code pattern being spotted is rather non-canonical for 128-bit MLAs, but it can happen and there's no point in generating sub-optimal code for it just because it looks odd. Should fix PR19332. llvm-svn: 205615
2014-04-04 17:03:02 +08:00
define <8 x i16> @not_really_vmlaq_laneq_s16_test(<8 x i16> %a, <8 x i16> %b, <8 x i16> %c) nounwind readnone ssp {
entry:
; CHECK: not_really_vmlaq_laneq_s16_test
; CHECK-NOT: ext
; CHECK: mla.8h v0, v1, v2[5]
; CHECK-NEXT: ret
%shuffle1 = shufflevector <8 x i16> %c, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%shuffle2 = shufflevector <4 x i16> %shuffle1, <4 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%mul = mul <8 x i16> %shuffle2, %b
%add = add <8 x i16> %mul, %a
ret <8 x i16> %add
}
define <4 x i32> @not_really_vmlaq_laneq_s32_test(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) nounwind readnone ssp {
entry:
; CHECK: not_really_vmlaq_laneq_s32_test
; CHECK-NOT: ext
; CHECK: mla.4s v0, v1, v2[3]
; CHECK-NEXT: ret
%shuffle1 = shufflevector <4 x i32> %c, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%shuffle2 = shufflevector <2 x i32> %shuffle1, <2 x i32> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
%mul = mul <4 x i32> %shuffle2, %b
%add = add <4 x i32> %mul, %a
ret <4 x i32> %add
}
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
define <4 x i32> @vmull_laneq_s16_test(<4 x i16> %a, <8 x i16> %b) nounwind readnone ssp {
entry:
; CHECK: vmull_laneq_s16_test
; CHECK-NOT: ext
; CHECK: smull.4s v0, v0, v1[6]
; CHECK-NEXT: ret
%shuffle = shufflevector <8 x i16> %b, <8 x i16> undef, <4 x i32> <i32 6, i32 6, i32 6, i32 6>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %a, <4 x i16> %shuffle) #2
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i
}
define <2 x i64> @vmull_laneq_s32_test(<2 x i32> %a, <4 x i32> %b) nounwind readnone ssp {
entry:
; CHECK: vmull_laneq_s32_test
; CHECK-NOT: ext
; CHECK: smull.2d v0, v0, v1[2]
; CHECK-NEXT: ret
%shuffle = shufflevector <4 x i32> %b, <4 x i32> undef, <2 x i32> <i32 2, i32 2>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %a, <2 x i32> %shuffle) #2
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i
}
define <4 x i32> @vmull_laneq_u16_test(<4 x i16> %a, <8 x i16> %b) nounwind readnone ssp {
entry:
; CHECK: vmull_laneq_u16_test
; CHECK-NOT: ext
; CHECK: umull.4s v0, v0, v1[6]
; CHECK-NEXT: ret
%shuffle = shufflevector <8 x i16> %b, <8 x i16> undef, <4 x i32> <i32 6, i32 6, i32 6, i32 6>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %a, <4 x i16> %shuffle) #2
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i
}
define <2 x i64> @vmull_laneq_u32_test(<2 x i32> %a, <4 x i32> %b) nounwind readnone ssp {
entry:
; CHECK: vmull_laneq_u32_test
; CHECK-NOT: ext
; CHECK: umull.2d v0, v0, v1[2]
; CHECK-NEXT: ret
%shuffle = shufflevector <4 x i32> %b, <4 x i32> undef, <2 x i32> <i32 2, i32 2>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %a, <2 x i32> %shuffle) #2
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i
}
define <4 x i32> @vmull_high_n_s16_test(<4 x i32> %a, <8 x i16> %b, <4 x i16> %c, i32 %d) nounwind readnone optsize ssp {
entry:
; CHECK: vmull_high_n_s16_test
; CHECK-NOT: ext
; CHECK: smull2.4s
; CHECK-NEXT: ret
%conv = trunc i32 %d to i16
%0 = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i.i to <4 x i16>
%vecinit.i = insertelement <4 x i16> undef, i16 %conv, i32 0
%vecinit1.i = insertelement <4 x i16> %vecinit.i, i16 %conv, i32 1
%vecinit2.i = insertelement <4 x i16> %vecinit1.i, i16 %conv, i32 2
%vecinit3.i = insertelement <4 x i16> %vecinit2.i, i16 %conv, i32 3
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <4 x i32> @llvm.aarch64.neon.smull.v4i32(<4 x i16> %1, <4 x i16> %vecinit3.i) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i.i
}
define <2 x i64> @vmull_high_n_s32_test(<2 x i64> %a, <4 x i32> %b, <2 x i32> %c, i32 %d) nounwind readnone optsize ssp {
entry:
; CHECK: vmull_high_n_s32_test
; CHECK-NOT: ext
; CHECK: smull2.2d
; CHECK-NEXT: ret
%0 = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i.i to <2 x i32>
%vecinit.i = insertelement <2 x i32> undef, i32 %d, i32 0
%vecinit1.i = insertelement <2 x i32> %vecinit.i, i32 %d, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <2 x i64> @llvm.aarch64.neon.smull.v2i64(<2 x i32> %1, <2 x i32> %vecinit1.i) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i.i
}
define <4 x i32> @vmull_high_n_u16_test(<4 x i32> %a, <8 x i16> %b, <4 x i16> %c, i32 %d) nounwind readnone optsize ssp {
entry:
; CHECK: vmull_high_n_u16_test
; CHECK-NOT: ext
; CHECK: umull2.4s
; CHECK-NEXT: ret
%conv = trunc i32 %d to i16
%0 = bitcast <8 x i16> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i.i to <4 x i16>
%vecinit.i = insertelement <4 x i16> undef, i16 %conv, i32 0
%vecinit1.i = insertelement <4 x i16> %vecinit.i, i16 %conv, i32 1
%vecinit2.i = insertelement <4 x i16> %vecinit1.i, i16 %conv, i32 2
%vecinit3.i = insertelement <4 x i16> %vecinit2.i, i16 %conv, i32 3
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <4 x i32> @llvm.aarch64.neon.umull.v4i32(<4 x i16> %1, <4 x i16> %vecinit3.i) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <4 x i32> %vmull2.i.i
}
define <2 x i64> @vmull_high_n_u32_test(<2 x i64> %a, <4 x i32> %b, <2 x i32> %c, i32 %d) nounwind readnone optsize ssp {
entry:
; CHECK: vmull_high_n_u32_test
; CHECK-NOT: ext
; CHECK: umull2.2d
; CHECK-NEXT: ret
%0 = bitcast <4 x i32> %b to <2 x i64>
%shuffle.i.i = shufflevector <2 x i64> %0, <2 x i64> undef, <1 x i32> <i32 1>
%1 = bitcast <1 x i64> %shuffle.i.i to <2 x i32>
%vecinit.i = insertelement <2 x i32> undef, i32 %d, i32 0
%vecinit1.i = insertelement <2 x i32> %vecinit.i, i32 %d, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%vmull2.i.i = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %1, <2 x i32> %vecinit1.i) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %vmull2.i.i
}
define <4 x i32> @vmul_built_dup_test(<4 x i32> %a, <4 x i32> %b) {
; CHECK-LABEL: vmul_built_dup_test:
; CHECK-NOT: ins
; CHECK-NOT: dup
; CHECK: mul.4s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[1]
%vget_lane = extractelement <4 x i32> %b, i32 1
%vecinit.i = insertelement <4 x i32> undef, i32 %vget_lane, i32 0
%vecinit1.i = insertelement <4 x i32> %vecinit.i, i32 %vget_lane, i32 1
%vecinit2.i = insertelement <4 x i32> %vecinit1.i, i32 %vget_lane, i32 2
%vecinit3.i = insertelement <4 x i32> %vecinit2.i, i32 %vget_lane, i32 3
%prod = mul <4 x i32> %a, %vecinit3.i
ret <4 x i32> %prod
}
define <4 x i16> @vmul_built_dup_fromsmall_test(<4 x i16> %a, <4 x i16> %b) {
; CHECK-LABEL: vmul_built_dup_fromsmall_test:
; CHECK-NOT: ins
; CHECK-NOT: dup
; CHECK: mul.4h {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[3]
%vget_lane = extractelement <4 x i16> %b, i32 3
%vecinit.i = insertelement <4 x i16> undef, i16 %vget_lane, i32 0
%vecinit1.i = insertelement <4 x i16> %vecinit.i, i16 %vget_lane, i32 1
%vecinit2.i = insertelement <4 x i16> %vecinit1.i, i16 %vget_lane, i32 2
%vecinit3.i = insertelement <4 x i16> %vecinit2.i, i16 %vget_lane, i32 3
%prod = mul <4 x i16> %a, %vecinit3.i
ret <4 x i16> %prod
}
define <8 x i16> @vmulq_built_dup_fromsmall_test(<8 x i16> %a, <4 x i16> %b) {
; CHECK-LABEL: vmulq_built_dup_fromsmall_test:
; CHECK-NOT: ins
; CHECK-NOT: dup
; CHECK: mul.8h {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
%vget_lane = extractelement <4 x i16> %b, i32 0
%vecinit.i = insertelement <8 x i16> undef, i16 %vget_lane, i32 0
%vecinit1.i = insertelement <8 x i16> %vecinit.i, i16 %vget_lane, i32 1
%vecinit2.i = insertelement <8 x i16> %vecinit1.i, i16 %vget_lane, i32 2
%vecinit3.i = insertelement <8 x i16> %vecinit2.i, i16 %vget_lane, i32 3
%vecinit4.i = insertelement <8 x i16> %vecinit3.i, i16 %vget_lane, i32 4
%vecinit5.i = insertelement <8 x i16> %vecinit4.i, i16 %vget_lane, i32 5
%vecinit6.i = insertelement <8 x i16> %vecinit5.i, i16 %vget_lane, i32 6
%vecinit7.i = insertelement <8 x i16> %vecinit6.i, i16 %vget_lane, i32 7
%prod = mul <8 x i16> %a, %vecinit7.i
ret <8 x i16> %prod
}
define <2 x i64> @mull_from_two_extracts(<4 x i32> %lhs, <4 x i32> %rhs) {
; CHECK-LABEL: mull_from_two_extracts:
; CHECK-NOT: ext
; CHECK: sqdmull2.2d
%lhs.high = shufflevector <4 x i32> %lhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%rhs.high = shufflevector <4 x i32> %rhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %lhs.high, <2 x i32> %rhs.high) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %res
}
define <2 x i64> @mlal_from_two_extracts(<2 x i64> %accum, <4 x i32> %lhs, <4 x i32> %rhs) {
; CHECK-LABEL: mlal_from_two_extracts:
; CHECK-NOT: ext
; CHECK: sqdmlal2.2d
%lhs.high = shufflevector <4 x i32> %lhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%rhs.high = shufflevector <4 x i32> %rhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %lhs.high, <2 x i32> %rhs.high) nounwind
%sum = call <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64> %accum, <2 x i64> %res)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %sum
}
define <2 x i64> @mull_from_extract_dup(<4 x i32> %lhs, i32 %rhs) {
; CHECK-LABEL: mull_from_extract_dup:
; CHECK-NOT: ext
; CHECK: sqdmull2.2d
%rhsvec.tmp = insertelement <2 x i32> undef, i32 %rhs, i32 0
%rhsvec = insertelement <2 x i32> %rhsvec.tmp, i32 %rhs, i32 1
%lhs.high = shufflevector <4 x i32> %lhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %lhs.high, <2 x i32> %rhsvec) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %res
}
define <8 x i16> @pmull_from_extract_dup(<16 x i8> %lhs, i8 %rhs) {
; CHECK-LABEL: pmull_from_extract_dup:
; CHECK-NOT: ext
; CHECK: pmull2.8h
%rhsvec.0 = insertelement <8 x i8> undef, i8 %rhs, i32 0
%rhsvec = shufflevector <8 x i8> %rhsvec.0, <8 x i8> undef, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
%lhs.high = shufflevector <16 x i8> %lhs, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <8 x i16> @llvm.aarch64.neon.pmull.v8i16(<8 x i8> %lhs.high, <8 x i8> %rhsvec) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %res
}
define <8 x i16> @pmull_from_extract_duplane(<16 x i8> %lhs, <8 x i8> %rhs) {
; CHECK-LABEL: pmull_from_extract_duplane:
; CHECK-NOT: ext
; CHECK: pmull2.8h
%lhs.high = shufflevector <16 x i8> %lhs, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
%rhs.high = shufflevector <8 x i8> %rhs, <8 x i8> undef, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <8 x i16> @llvm.aarch64.neon.pmull.v8i16(<8 x i8> %lhs.high, <8 x i8> %rhs.high) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <8 x i16> %res
}
define <2 x i64> @sqdmull_from_extract_duplane(<4 x i32> %lhs, <4 x i32> %rhs) {
; CHECK-LABEL: sqdmull_from_extract_duplane:
; CHECK-NOT: ext
; CHECK: sqdmull2.2d
%lhs.high = shufflevector <4 x i32> %lhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%rhs.high = shufflevector <4 x i32> %rhs, <4 x i32> undef, <2 x i32> <i32 0, i32 0>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %lhs.high, <2 x i32> %rhs.high) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %res
}
define <2 x i64> @sqdmlal_from_extract_duplane(<2 x i64> %accum, <4 x i32> %lhs, <4 x i32> %rhs) {
; CHECK-LABEL: sqdmlal_from_extract_duplane:
; CHECK-NOT: ext
; CHECK: sqdmlal2.2d
%lhs.high = shufflevector <4 x i32> %lhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%rhs.high = shufflevector <4 x i32> %rhs, <4 x i32> undef, <2 x i32> <i32 0, i32 0>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <2 x i64> @llvm.aarch64.neon.sqdmull.v2i64(<2 x i32> %lhs.high, <2 x i32> %rhs.high) nounwind
%sum = call <2 x i64> @llvm.aarch64.neon.sqadd.v2i64(<2 x i64> %accum, <2 x i64> %res)
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
ret <2 x i64> %sum
}
define <2 x i64> @umlal_from_extract_duplane(<2 x i64> %accum, <4 x i32> %lhs, <4 x i32> %rhs) {
; CHECK-LABEL: umlal_from_extract_duplane:
; CHECK-NOT: ext
; CHECK: umlal2.2d
%lhs.high = shufflevector <4 x i32> %lhs, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
%rhs.high = shufflevector <4 x i32> %rhs, <4 x i32> undef, <2 x i32> <i32 0, i32 0>
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%res = tail call <2 x i64> @llvm.aarch64.neon.umull.v2i64(<2 x i32> %lhs.high, <2 x i32> %rhs.high) nounwind
ARM64: initial backend import This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. llvm-svn: 205090
2014-03-29 18:18:08 +08:00
%sum = add <2 x i64> %accum, %res
ret <2 x i64> %sum
}
define float @scalar_fmla_from_extract_v4f32(float %accum, float %lhs, <4 x float> %rvec) {
; CHECK-LABEL: scalar_fmla_from_extract_v4f32:
; CHECK: fmla.s s0, s1, v2[3]
%rhs = extractelement <4 x float> %rvec, i32 3
%res = call float @llvm.fma.f32(float %lhs, float %rhs, float %accum)
ret float %res
}
define float @scalar_fmla_from_extract_v2f32(float %accum, float %lhs, <2 x float> %rvec) {
; CHECK-LABEL: scalar_fmla_from_extract_v2f32:
; CHECK: fmla.s s0, s1, v2[1]
%rhs = extractelement <2 x float> %rvec, i32 1
%res = call float @llvm.fma.f32(float %lhs, float %rhs, float %accum)
ret float %res
}
define float @scalar_fmls_from_extract_v4f32(float %accum, float %lhs, <4 x float> %rvec) {
; CHECK-LABEL: scalar_fmls_from_extract_v4f32:
; CHECK: fmls.s s0, s1, v2[3]
%rhs.scal = extractelement <4 x float> %rvec, i32 3
%rhs = fsub float -0.0, %rhs.scal
%res = call float @llvm.fma.f32(float %lhs, float %rhs, float %accum)
ret float %res
}
define float @scalar_fmls_from_extract_v2f32(float %accum, float %lhs, <2 x float> %rvec) {
; CHECK-LABEL: scalar_fmls_from_extract_v2f32:
; CHECK: fmls.s s0, s1, v2[1]
%rhs.scal = extractelement <2 x float> %rvec, i32 1
%rhs = fsub float -0.0, %rhs.scal
%res = call float @llvm.fma.f32(float %lhs, float %rhs, float %accum)
ret float %res
}
declare float @llvm.fma.f32(float, float, float)
define double @scalar_fmla_from_extract_v2f64(double %accum, double %lhs, <2 x double> %rvec) {
; CHECK-LABEL: scalar_fmla_from_extract_v2f64:
; CHECK: fmla.d d0, d1, v2[1]
%rhs = extractelement <2 x double> %rvec, i32 1
%res = call double @llvm.fma.f64(double %lhs, double %rhs, double %accum)
ret double %res
}
define double @scalar_fmls_from_extract_v2f64(double %accum, double %lhs, <2 x double> %rvec) {
; CHECK-LABEL: scalar_fmls_from_extract_v2f64:
; CHECK: fmls.d d0, d1, v2[1]
%rhs.scal = extractelement <2 x double> %rvec, i32 1
%rhs = fsub double -0.0, %rhs.scal
%res = call double @llvm.fma.f64(double %lhs, double %rhs, double %accum)
ret double %res
}
declare double @llvm.fma.f64(double, double, double)
define <2 x float> @fmls_with_fneg_before_extract_v2f32(<2 x float> %accum, <2 x float> %lhs, <4 x float> %rhs) {
; CHECK-LABEL: fmls_with_fneg_before_extract_v2f32:
; CHECK: fmls.2s v0, v1, v2[3]
%rhs_neg = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %rhs
%splat = shufflevector <4 x float> %rhs_neg, <4 x float> undef, <2 x i32> <i32 3, i32 3>
%res = call <2 x float> @llvm.fma.v2f32(<2 x float> %lhs, <2 x float> %splat, <2 x float> %accum)
ret <2 x float> %res
}
define <2 x float> @fmls_with_fneg_before_extract_v2f32_1(<2 x float> %accum, <2 x float> %lhs, <2 x float> %rhs) {
; CHECK-LABEL: fmls_with_fneg_before_extract_v2f32_1:
; CHECK: fmls.2s v0, v1, v2[1]
%rhs_neg = fsub <2 x float> <float -0.0, float -0.0>, %rhs
%splat = shufflevector <2 x float> %rhs_neg, <2 x float> undef, <2 x i32> <i32 1, i32 1>
%res = call <2 x float> @llvm.fma.v2f32(<2 x float> %lhs, <2 x float> %splat, <2 x float> %accum)
ret <2 x float> %res
}
define <4 x float> @fmls_with_fneg_before_extract_v4f32(<4 x float> %accum, <4 x float> %lhs, <4 x float> %rhs) {
; CHECK-LABEL: fmls_with_fneg_before_extract_v4f32:
; CHECK: fmls.4s v0, v1, v2[3]
%rhs_neg = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %rhs
%splat = shufflevector <4 x float> %rhs_neg, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
%res = call <4 x float> @llvm.fma.v4f32(<4 x float> %lhs, <4 x float> %splat, <4 x float> %accum)
ret <4 x float> %res
}
define <4 x float> @fmls_with_fneg_before_extract_v4f32_1(<4 x float> %accum, <4 x float> %lhs, <2 x float> %rhs) {
; CHECK-LABEL: fmls_with_fneg_before_extract_v4f32_1:
; CHECK: fmls.4s v0, v1, v2[1]
%rhs_neg = fsub <2 x float> <float -0.0, float -0.0>, %rhs
%splat = shufflevector <2 x float> %rhs_neg, <2 x float> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
%res = call <4 x float> @llvm.fma.v4f32(<4 x float> %lhs, <4 x float> %splat, <4 x float> %accum)
ret <4 x float> %res
}
define <2 x double> @fmls_with_fneg_before_extract_v2f64(<2 x double> %accum, <2 x double> %lhs, <2 x double> %rhs) {
; CHECK-LABEL: fmls_with_fneg_before_extract_v2f64:
; CHECK: fmls.2d v0, v1, v2[1]
%rhs_neg = fsub <2 x double> <double -0.0, double -0.0>, %rhs
%splat = shufflevector <2 x double> %rhs_neg, <2 x double> undef, <2 x i32> <i32 1, i32 1>
%res = call <2 x double> @llvm.fma.v2f64(<2 x double> %lhs, <2 x double> %splat, <2 x double> %accum)
ret <2 x double> %res
}
define <1 x double> @test_fmul_v1f64(<1 x double> %L, <1 x double> %R) nounwind {
; CHECK-LABEL: test_fmul_v1f64:
; CHECK: fmul
%prod = fmul <1 x double> %L, %R
ret <1 x double> %prod
}
define <1 x double> @test_fdiv_v1f64(<1 x double> %L, <1 x double> %R) nounwind {
; CHECK-LABEL: test_fdiv_v1f64:
; CHECK-LABEL: fdiv
%prod = fdiv <1 x double> %L, %R
ret <1 x double> %prod
}
ARM64: add patterns for scalar sqdmlal & sqdmlsl. llvm-svn: 205207
2014-03-31 23:46:38 +08:00
define i64 @sqdmlal_d(i32 %A, i32 %B, i64 %C) nounwind {
;CHECK-LABEL: sqdmlal_d:
;CHECK: sqdmlal
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %A, i32 %B)
%tmp5 = call i64 @llvm.aarch64.neon.sqadd.i64(i64 %C, i64 %tmp4)
ARM64: add patterns for scalar sqdmlal & sqdmlsl. llvm-svn: 205207
2014-03-31 23:46:38 +08:00
ret i64 %tmp5
}
define i64 @sqdmlsl_d(i32 %A, i32 %B, i64 %C) nounwind {
;CHECK-LABEL: sqdmlsl_d:
;CHECK: sqdmlsl
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%tmp4 = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %A, i32 %B)
%tmp5 = call i64 @llvm.aarch64.neon.sqsub.i64(i64 %C, i64 %tmp4)
ARM64: add patterns for scalar sqdmlal & sqdmlsl. llvm-svn: 205207
2014-03-31 23:46:38 +08:00
ret i64 %tmp5
}
ARM64: add intrinsic for pmull (p64 x p64 = p128) operations. llvm-svn: 205302
2014-04-01 20:22:37 +08:00
define <16 x i8> @test_pmull_64(i64 %l, i64 %r) nounwind {
; CHECK-LABEL: test_pmull_64:
; CHECK: pmull.1q
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%val = call <16 x i8> @llvm.aarch64.neon.pmull64(i64 %l, i64 %r)
ARM64: add intrinsic for pmull (p64 x p64 = p128) operations. llvm-svn: 205302
2014-04-01 20:22:37 +08:00
ret <16 x i8> %val
}
define <16 x i8> @test_pmull_high_64(<2 x i64> %l, <2 x i64> %r) nounwind {
; CHECK-LABEL: test_pmull_high_64:
; CHECK: pmull2.1q
%l_hi = extractelement <2 x i64> %l, i32 1
%r_hi = extractelement <2 x i64> %r, i32 1
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
%val = call <16 x i8> @llvm.aarch64.neon.pmull64(i64 %l_hi, i64 %r_hi)
ARM64: add intrinsic for pmull (p64 x p64 = p128) operations. llvm-svn: 205302
2014-04-01 20:22:37 +08:00
ret <16 x i8> %val
}
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 20:50:23 +08:00
declare <16 x i8> @llvm.aarch64.neon.pmull64(i64, i64)
ARM64: scalarize v1i64 mul operation This is the second part of fixing PR19367. llvm-svn: 205836
2014-04-09 15:07:02 +08:00
define <1 x i64> @test_mul_v1i64(<1 x i64> %lhs, <1 x i64> %rhs) nounwind {
; CHECK-LABEL: test_mul_v1i64:
; CHECK: mul
%prod = mul <1 x i64> %lhs, %rhs
ret <1 x i64> %prod
}