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/X86/fold-load-unops.ll

216 lines
6.7 KiB
LLVM
Raw Normal View History

Make utils/update_llc_test_checks.py note that the assertions are autogenerated. Also update existing test cases which appear to be generated by it and weren't modified (other than addition of the header) by rerunning it. llvm-svn: 253917
2015-11-24 05:33:58 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; RUN: llc -mtriple=x86_64-unknown-unknown -mattr=+sse2 < %s | FileCheck %s --check-prefix=SSE
; RUN: llc -mtriple=x86_64-unknown-unknown -mattr=+avx < %s | FileCheck %s --check-prefix=AVX
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; RUN: llc -mtriple=x86_64-unknown-unknown -mattr=+avx512f < %s | FileCheck %s --check-prefix=AVX
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; Verify we fold loads into unary sse intrinsics only when optimizing for size
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
define float @rcpss(float* %a) {
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-LABEL: rcpss:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; SSE-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: rcpss %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: rcpss:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; AVX-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vrcpss %xmm0, %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; AVX-NEXT: retq
[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
%ld = load float, float* %a
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
%ins = insertelement <4 x float> undef, float %ld, i32 0
%res = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %ins)
%ext = extractelement <4 x float> %res, i32 0
ret float %ext
}
define float @rsqrtss(float* %a) {
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-LABEL: rsqrtss:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; SSE-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: rsqrtss %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: rsqrtss:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; AVX-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vrsqrtss %xmm0, %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; AVX-NEXT: retq
[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
%ld = load float, float* %a
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
%ins = insertelement <4 x float> undef, float %ld, i32 0
%res = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %ins)
%ext = extractelement <4 x float> %res, i32 0
ret float %ext
}
define float @sqrtss(float* %a) {
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-LABEL: sqrtss:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; SSE-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: sqrtss %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: sqrtss:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; AVX-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vsqrtss %xmm0, %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; AVX-NEXT: retq
[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
%ld = load float, float* %a
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
%ins = insertelement <4 x float> undef, float %ld, i32 0
%res = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %ins)
%ext = extractelement <4 x float> %res, i32 0
ret float %ext
}
define double @sqrtsd(double* %a) {
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-LABEL: sqrtsd:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: sqrtsd %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: sqrtsd:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Regenerate fold-load-unops.ll and add and avx512f command line. llvm-svn: 318014
2017-11-13 13:25:21 +08:00
; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vsqrtsd %xmm0, %xmm0, %xmm0
add SSE run to check non-AVX codegen llvm-svn: 235809
2015-04-26 04:41:51 +08:00
; AVX-NEXT: retq
[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
%ld = load double, double* %a
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
%ins = insertelement <2 x double> undef, double %ld, i32 0
%res = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %ins)
%ext = extractelement <2 x double> %res, i32 0
ret double %ext
}
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
define float @rcpss_size(float* %a) optsize {
; SSE-LABEL: rcpss_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: rcpss (%rdi), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: rcpss_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vrcpss (%rdi), %xmm0, %xmm0
; AVX-NEXT: retq
%ld = load float, float* %a
%ins = insertelement <4 x float> undef, float %ld, i32 0
%res = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %ins)
%ext = extractelement <4 x float> %res, i32 0
ret float %ext
}
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
define <4 x float> @rcpss_full_size(<4 x float>* %a) optsize {
; SSE-LABEL: rcpss_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; SSE-NEXT: rcpss (%rdi), %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: rcpss_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; AVX-NEXT: vrcpss (%rdi), %xmm0, %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; AVX-NEXT: retq
%ld = load <4 x float>, <4 x float>* %a
%res = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %ld)
ret <4 x float> %res
}
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
define float @rsqrtss_size(float* %a) optsize {
; SSE-LABEL: rsqrtss_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: rsqrtss (%rdi), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: rsqrtss_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vrsqrtss (%rdi), %xmm0, %xmm0
; AVX-NEXT: retq
%ld = load float, float* %a
%ins = insertelement <4 x float> undef, float %ld, i32 0
%res = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %ins)
%ext = extractelement <4 x float> %res, i32 0
ret float %ext
}
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
define <4 x float> @rsqrtss_full_size(<4 x float>* %a) optsize {
; SSE-LABEL: rsqrtss_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; SSE-NEXT: rsqrtss (%rdi), %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: rsqrtss_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; AVX-NEXT: vrsqrtss (%rdi), %xmm0, %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; AVX-NEXT: retq
%ld = load <4 x float>, <4 x float>* %a
%res = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %ld)
ret <4 x float> %res
}
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
define float @sqrtss_size(float* %a) optsize{
; SSE-LABEL: sqrtss_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: sqrtss (%rdi), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: sqrtss_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vsqrtss (%rdi), %xmm0, %xmm0
; AVX-NEXT: retq
%ld = load float, float* %a
%ins = insertelement <4 x float> undef, float %ld, i32 0
%res = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %ins)
%ext = extractelement <4 x float> %res, i32 0
ret float %ext
}
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
define <4 x float> @sqrtss_full_size(<4 x float>* %a) optsize{
; SSE-LABEL: sqrtss_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; SSE-NEXT: sqrtss (%rdi), %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: sqrtss_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; AVX-NEXT: vsqrtss (%rdi), %xmm0, %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; AVX-NEXT: retq
%ld = load <4 x float>, <4 x float>* %a
%res = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %ld)
ret <4 x float> %res
}
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
define double @sqrtsd_size(double* %a) optsize {
; SSE-LABEL: sqrtsd_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; SSE-NEXT: sqrtsd (%rdi), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: sqrtsd_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Avoid folding scalar loads into unary sse intrinsics Not folding these cases tends to avoid partial register updates: sqrtss (%eax), %xmm0 Has a partial update of %xmm0, while movss (%eax), %xmm0 sqrtss %xmm0, %xmm0 Has a clobber of the high lanes immediately before the partial update, avoiding a potential stall. Given this, we only want to fold when optimizing for size. This is consistent with the patterns we already have for some of the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl() Differential Revision: http://reviews.llvm.org/D15741 llvm-svn: 256671
2015-12-31 17:45:16 +08:00
; AVX-NEXT: vsqrtsd (%rdi), %xmm0, %xmm0
; AVX-NEXT: retq
%ld = load double, double* %a
%ins = insertelement <2 x double> undef, double %ld, i32 0
%res = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %ins)
%ext = extractelement <2 x double> %res, i32 0
ret double %ext
}
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
define <2 x double> @sqrtsd_full_size(<2 x double>* %a) optsize {
; SSE-LABEL: sqrtsd_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; SSE-NEXT: sqrtsd (%rdi), %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: sqrtsd_full_size:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86] Fix SQRTSS/SQRTSD/RCPSS/RCPSD intrinsics to use sse_load_f32/sse_load_f64 to increase load folding opportunities. llvm-svn: 318016
2017-11-13 13:25:24 +08:00
; AVX-NEXT: vsqrtsd (%rdi), %xmm0, %xmm0
[X86] Add tests for full vector loads to fold-load-unops.ll. We should be able to fold a full vector load into a scalar intrinsic. Since it's legal to narrow a load. llvm-svn: 318015
2017-11-13 13:25:23 +08:00
; AVX-NEXT: retq
%ld = load <2 x double>, <2 x double>* %a
%res = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %ld)
ret <2 x double> %res
}
add X86 load folding tests for unary math ops X86 load folding is fragile; eg, the tests here don't work without AVX even though they should. This is because we have a mix of tablegen patterns that have been added over time, and we have a load folding table used by the peephole optimizer that has to be kept in sync with the ever-changing ISA and tablegen defs. llvm-svn: 229870
2015-02-20 00:59:11 +08:00
declare <4 x float> @llvm.x86.sse.rcp.ss(<4 x float>) nounwind readnone
declare <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float>) nounwind readnone
declare <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float>) nounwind readnone
declare <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double>) nounwind readnone