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/widen_conv-3.ll

136 lines
5.5 KiB
LLVM
Raw Normal View History

[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse2 | FileCheck %s --check-prefix=X86 --check-prefix=X86-SSE2
; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=X86 --check-prefix=X86-SSE42
; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-SSE2
; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-SSE42
Update tests to use FileCheck llvm-svn: 84282
2009-10-17 06:09:05 +08:00
Added some basic test cases for r61209 llvm-svn: 61210
2008-12-19 04:05:58 +08:00
; sign to float v2i16 to v2f32
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
define void @convert_v2i16_to_v2f32(<2 x float>* %dst.addr, <2 x i16> %src) nounwind {
[X86] Type legalize v2f32 stores by widening to v4f32, casting to v2f64, extracting f64 and storing. Summary: This is similar to what D52528 did for loads. It should match what generic type legalization does in 64-bit mode where it uses a v2i64 cast and an i64 store. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53173 llvm-svn: 344470
2018-10-14 11:36:27 +08:00
; X86-LABEL: convert_v2i16_to_v2f32:
; X86: # %bb.0: # %entry
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: psllq $48, %xmm0
; X86-NEXT: psrad $16, %xmm0
; X86-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
; X86-NEXT: cvtdq2ps %xmm0, %xmm0
; X86-NEXT: movlps %xmm0, (%eax)
; X86-NEXT: retl
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
;
; X64-LABEL: convert_v2i16_to_v2f32:
[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
; X64: # %bb.0: # %entry
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X64-NEXT: psllq $48, %xmm0
; X64-NEXT: psrad $16, %xmm0
; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
; X64-NEXT: cvtdq2ps %xmm0, %xmm0
; X64-NEXT: movlps %xmm0, (%rdi)
; X64-NEXT: retq
Added some basic test cases for r61209 llvm-svn: 61210
2008-12-19 04:05:58 +08:00
entry:
%val = sitofp <2 x i16> %src to <2 x float>
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
store <2 x float> %val, <2 x float>* %dst.addr, align 4
Added some basic test cases for r61209 llvm-svn: 61210
2008-12-19 04:05:58 +08:00
ret void
}
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; sign to float v3i8 to v3f32
define void @convert_v3i8_to_v3f32(<3 x float>* %dst.addr, <3 x i8>* %src.addr) nounwind {
; X86-SSE2-LABEL: convert_v3i8_to_v3f32:
[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
; X86-SSE2: # %bb.0: # %entry
[X86][SSE] Reapplied: Simplify vector LOAD + EXTEND on pre-SSE41 hardware Improve vector extension of vectors on hardware without dedicated VSEXT/VZEXT instructions. We already convert these to SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG but can further improve this by using the legalizer instead of prematurely splitting into legal vectors in the combine as this only properly helps for lowering to VSEXT/VZEXT. Removes a lot of unnecessary any_extend + mask pattern - (Fix for PR25718). Reapplied with a fix for PR26953 (missing vector widening legalization). Differential Revision: http://reviews.llvm.org/D17932 llvm-svn: 264062
2016-03-23 00:22:08 +08:00
; X86-SSE2-NEXT: pushl %ebp
; X86-SSE2-NEXT: movl %esp, %ebp
; X86-SSE2-NEXT: pushl %esi
; X86-SSE2-NEXT: andl $-16, %esp
; X86-SSE2-NEXT: subl $32, %esp
; X86-SSE2-NEXT: movl 8(%ebp), %eax
; X86-SSE2-NEXT: movl 12(%ebp), %ecx
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
; X86-SSE2-NEXT: movzwl (%ecx), %edx
; X86-SSE2-NEXT: movd %edx, %xmm0
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X86-SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
[X86][SSE] Reapplied: Simplify vector LOAD + EXTEND on pre-SSE41 hardware Improve vector extension of vectors on hardware without dedicated VSEXT/VZEXT instructions. We already convert these to SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG but can further improve this by using the legalizer instead of prematurely splitting into legal vectors in the combine as this only properly helps for lowering to VSEXT/VZEXT. Removes a lot of unnecessary any_extend + mask pattern - (Fix for PR25718). Reapplied with a fix for PR26953 (missing vector widening legalization). Differential Revision: http://reviews.llvm.org/D17932 llvm-svn: 264062
2016-03-23 00:22:08 +08:00
; X86-SSE2-NEXT: movdqa %xmm0, (%esp)
[DAGCombiner][X86][Mips] Enable combineShuffleOfScalars to run between vector op legalization and DAG legalization. Fix bad one use check in combineShuffleOfScalars It's possible for vector op legalization to generate a shuffle. If that happens we should give a chance for DAG combine to combine that with a build_vector input. I also fixed a bug in combineShuffleOfScalars that was considering the number of uses on a undef input to a shuffle. We don't care how many times undef is used. Differential Revision: https://reviews.llvm.org/D54283 llvm-svn: 346530
2018-11-10 02:04:34 +08:00
; X86-SSE2-NEXT: movl (%esp), %edx
; X86-SSE2-NEXT: movl {{[0-9]+}}(%esp), %esi
; X86-SSE2-NEXT: shll $8, %esi
[X86][SSE] Reapplied: Simplify vector LOAD + EXTEND on pre-SSE41 hardware Improve vector extension of vectors on hardware without dedicated VSEXT/VZEXT instructions. We already convert these to SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG but can further improve this by using the legalizer instead of prematurely splitting into legal vectors in the combine as this only properly helps for lowering to VSEXT/VZEXT. Removes a lot of unnecessary any_extend + mask pattern - (Fix for PR25718). Reapplied with a fix for PR26953 (missing vector widening legalization). Differential Revision: http://reviews.llvm.org/D17932 llvm-svn: 264062
2016-03-23 00:22:08 +08:00
; X86-SSE2-NEXT: shll $8, %edx
[DAGCombiner][X86][Mips] Enable combineShuffleOfScalars to run between vector op legalization and DAG legalization. Fix bad one use check in combineShuffleOfScalars It's possible for vector op legalization to generate a shuffle. If that happens we should give a chance for DAG combine to combine that with a build_vector input. I also fixed a bug in combineShuffleOfScalars that was considering the number of uses on a undef input to a shuffle. We don't care how many times undef is used. Differential Revision: https://reviews.llvm.org/D54283 llvm-svn: 346530
2018-11-10 02:04:34 +08:00
; X86-SSE2-NEXT: movd %edx, %xmm0
; X86-SSE2-NEXT: pinsrw $1, %esi, %xmm0
; X86-SSE2-NEXT: movzbl 2(%ecx), %ecx
; X86-SSE2-NEXT: shll $8, %ecx
; X86-SSE2-NEXT: pinsrw $2, %ecx, %xmm0
[X86][SSE] Reapplied: Simplify vector LOAD + EXTEND on pre-SSE41 hardware Improve vector extension of vectors on hardware without dedicated VSEXT/VZEXT instructions. We already convert these to SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG but can further improve this by using the legalizer instead of prematurely splitting into legal vectors in the combine as this only properly helps for lowering to VSEXT/VZEXT. Removes a lot of unnecessary any_extend + mask pattern - (Fix for PR25718). Reapplied with a fix for PR26953 (missing vector widening legalization). Differential Revision: http://reviews.llvm.org/D17932 llvm-svn: 264062
2016-03-23 00:22:08 +08:00
; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X86-SSE2-NEXT: psrad $24, %xmm0
; X86-SSE2-NEXT: cvtdq2ps %xmm0, %xmm0
; X86-SSE2-NEXT: movss %xmm0, (%eax)
; X86-SSE2-NEXT: movaps %xmm0, %xmm1
[X86] Prefer unpckhpd over movhlps in isel for fake unary cases In r337348, I changed lowering to prefer X86ISD::UNPCKL/UNPCKH opcodes over MOVLHPS/MOVHLPS for v2f64 {0,0} and {1,1} shuffles when we have SSE2. This enabled the removal of a bunch of weirdly bitcasted isel patterns in r337349. To avoid changing the tests I placed a gross hack in isel to still emit movhlps instructions for fake unary unpckh nodes. A similar hack was not needed for unpckl and movlhps because we do execution domain switching for those. But unpckh and movhlps have swapped operand order. This patch removes the hack. This is a code size increase since unpckhpd requires a 0x66 prefix and movhlps does not. But if that's a big concern we should be using movhlps for all unpckhpd opcodes and let commuteInstruction turnit into unpckhpd when its an advantage. Differential Revision: https://reviews.llvm.org/D49499 llvm-svn: 341973
2018-09-12 01:57:27 +08:00
; X86-SSE2-NEXT: unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X86-SSE2-NEXT: movss %xmm1, 8(%eax)
; X86-SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[1,1,2,3]
; X86-SSE2-NEXT: movss %xmm0, 4(%eax)
[X86][SSE] Reapplied: Simplify vector LOAD + EXTEND on pre-SSE41 hardware Improve vector extension of vectors on hardware without dedicated VSEXT/VZEXT instructions. We already convert these to SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG but can further improve this by using the legalizer instead of prematurely splitting into legal vectors in the combine as this only properly helps for lowering to VSEXT/VZEXT. Removes a lot of unnecessary any_extend + mask pattern - (Fix for PR25718). Reapplied with a fix for PR26953 (missing vector widening legalization). Differential Revision: http://reviews.llvm.org/D17932 llvm-svn: 264062
2016-03-23 00:22:08 +08:00
; X86-SSE2-NEXT: leal -4(%ebp), %esp
; X86-SSE2-NEXT: popl %esi
; X86-SSE2-NEXT: popl %ebp
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X86-SSE2-NEXT: retl
;
; X86-SSE42-LABEL: convert_v3i8_to_v3f32:
[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
; X86-SSE42: # %bb.0: # %entry
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X86-SSE42-NEXT: pushl %eax
; X86-SSE42-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-SSE42-NEXT: movl {{[0-9]+}}(%esp), %ecx
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
; X86-SSE42-NEXT: movzbl 2(%ecx), %edx
; X86-SSE42-NEXT: movzwl (%ecx), %ecx
; X86-SSE42-NEXT: movd %ecx, %xmm0
; X86-SSE42-NEXT: pmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
; X86-SSE42-NEXT: pinsrd $2, %edx, %xmm0
; X86-SSE42-NEXT: pslld $24, %xmm0
; X86-SSE42-NEXT: psrad $24, %xmm0
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X86-SSE42-NEXT: cvtdq2ps %xmm0, %xmm0
; X86-SSE42-NEXT: extractps $2, %xmm0, 8(%eax)
; X86-SSE42-NEXT: extractps $1, %xmm0, 4(%eax)
; X86-SSE42-NEXT: movss %xmm0, (%eax)
; X86-SSE42-NEXT: popl %eax
; X86-SSE42-NEXT: retl
;
; X64-SSE2-LABEL: convert_v3i8_to_v3f32:
[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
; X64-SSE2: # %bb.0: # %entry
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
; X64-SSE2-NEXT: movzwl (%rsi), %eax
[X86][SSE2] Fix asm string for movq (Move Quadword) instruction. Replace "mov{d|q}" with "movq". Differential Revision: https://reviews.llvm.org/D32220 llvm-svn: 301386
2017-04-26 15:08:44 +08:00
; X64-SSE2-NEXT: movq %rax, %xmm0
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X64-SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; X64-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
[DAGCombiner] Set the right SDLoc on a newly-created zextload (1/N) Setting the right SDLoc on a newly-created zextload fixes a line table bug which resulted in non-linear stepping behavior. Several backend tests contained CHECK lines which relied on the IROrder inherited from the wrong SDLoc. This patch breaks that dependence where feasbile and regenerates test cases where not. In some cases, changing a node's IROrder may alter register allocation and spill behavior. This can affect performance. I have chosen not to prevent this by applying a "known good" IROrder to SDLocs, as this may hide a more general bug in the scheduler, or cause regressions on other test inputs. rdar://33755881, Part of: llvm.org/PR37262 Differential Revision: https://reviews.llvm.org/D45995 llvm-svn: 331300
2018-05-02 03:26:15 +08:00
; X64-SSE2-NEXT: movdqa %xmm0, -{{[0-9]+}}(%rsp)
[DAGCombiner][X86][Mips] Enable combineShuffleOfScalars to run between vector op legalization and DAG legalization. Fix bad one use check in combineShuffleOfScalars It's possible for vector op legalization to generate a shuffle. If that happens we should give a chance for DAG combine to combine that with a build_vector input. I also fixed a bug in combineShuffleOfScalars that was considering the number of uses on a undef input to a shuffle. We don't care how many times undef is used. Differential Revision: https://reviews.llvm.org/D54283 llvm-svn: 346530
2018-11-10 02:04:34 +08:00
; X64-SSE2-NEXT: movl -{{[0-9]+}}(%rsp), %eax
[DAGCombiner] Set the right SDLoc on a newly-created zextload (1/N) Setting the right SDLoc on a newly-created zextload fixes a line table bug which resulted in non-linear stepping behavior. Several backend tests contained CHECK lines which relied on the IROrder inherited from the wrong SDLoc. This patch breaks that dependence where feasbile and regenerates test cases where not. In some cases, changing a node's IROrder may alter register allocation and spill behavior. This can affect performance. I have chosen not to prevent this by applying a "known good" IROrder to SDLocs, as this may hide a more general bug in the scheduler, or cause regressions on other test inputs. rdar://33755881, Part of: llvm.org/PR37262 Differential Revision: https://reviews.llvm.org/D45995 llvm-svn: 331300
2018-05-02 03:26:15 +08:00
; X64-SSE2-NEXT: movl -{{[0-9]+}}(%rsp), %ecx
; X64-SSE2-NEXT: shll $8, %ecx
[DAGCombiner][X86][Mips] Enable combineShuffleOfScalars to run between vector op legalization and DAG legalization. Fix bad one use check in combineShuffleOfScalars It's possible for vector op legalization to generate a shuffle. If that happens we should give a chance for DAG combine to combine that with a build_vector input. I also fixed a bug in combineShuffleOfScalars that was considering the number of uses on a undef input to a shuffle. We don't care how many times undef is used. Differential Revision: https://reviews.llvm.org/D54283 llvm-svn: 346530
2018-11-10 02:04:34 +08:00
; X64-SSE2-NEXT: shll $8, %eax
; X64-SSE2-NEXT: movd %eax, %xmm0
; X64-SSE2-NEXT: pinsrw $1, %ecx, %xmm0
; X64-SSE2-NEXT: movzbl 2(%rsi), %eax
; X64-SSE2-NEXT: shll $8, %eax
; X64-SSE2-NEXT: pinsrw $2, %eax, %xmm0
[X86][SSE] Reapplied: Simplify vector LOAD + EXTEND on pre-SSE41 hardware Improve vector extension of vectors on hardware without dedicated VSEXT/VZEXT instructions. We already convert these to SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG but can further improve this by using the legalizer instead of prematurely splitting into legal vectors in the combine as this only properly helps for lowering to VSEXT/VZEXT. Removes a lot of unnecessary any_extend + mask pattern - (Fix for PR25718). Reapplied with a fix for PR26953 (missing vector widening legalization). Differential Revision: http://reviews.llvm.org/D17932 llvm-svn: 264062
2016-03-23 00:22:08 +08:00
; X64-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X64-SSE2-NEXT: psrad $24, %xmm0
; X64-SSE2-NEXT: cvtdq2ps %xmm0, %xmm0
; X64-SSE2-NEXT: movlps %xmm0, (%rdi)
[X86][SSE] Avoid specifying unused arguments in SHUFPD lowering As discussed on PR26491, we are missing the opportunity to make use of the smaller MOVHLPS instruction because we set both arguments of a SHUFPD when using it to lower a single input shuffle. This patch sets the lowered argument to UNDEF if that shuffle element is undefined. This in turn makes it easier for target shuffle combining to decode UNDEF shuffle elements, allowing combines to MOVHLPS to occur. A fix to match against MOVHPD stores was necessary as well. This builds on the improved MOVLHPS/MOVHLPS lowering and memory folding support added in D16956 Adding similar support for SHUFPS will have to wait until have better support for target combining of binary shuffles. Differential Revision: https://reviews.llvm.org/D23027 llvm-svn: 279430
2016-08-22 20:56:54 +08:00
; X64-SSE2-NEXT: movhlps {{.*#+}} xmm0 = xmm0[1,1]
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X64-SSE2-NEXT: movss %xmm0, 8(%rdi)
; X64-SSE2-NEXT: retq
;
; X64-SSE42-LABEL: convert_v3i8_to_v3f32:
[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
; X64-SSE42: # %bb.0: # %entry
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
; X64-SSE42-NEXT: movzbl 2(%rsi), %eax
; X64-SSE42-NEXT: movzwl (%rsi), %ecx
[X86][SSE2] Fix asm string for movq (Move Quadword) instruction. Replace "mov{d|q}" with "movq". Differential Revision: https://reviews.llvm.org/D32220 llvm-svn: 301386
2017-04-26 15:08:44 +08:00
; X64-SSE42-NEXT: movq %rcx, %xmm0
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
; X64-SSE42-NEXT: pmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
; X64-SSE42-NEXT: pinsrd $2, %eax, %xmm0
; X64-SSE42-NEXT: pslld $24, %xmm0
; X64-SSE42-NEXT: psrad $24, %xmm0
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
; X64-SSE42-NEXT: cvtdq2ps %xmm0, %xmm0
; X64-SSE42-NEXT: extractps $2, %xmm0, 8(%rdi)
; X64-SSE42-NEXT: movlps %xmm0, (%rdi)
; X64-SSE42-NEXT: retq
entry:
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
%load = load <3 x i8>, <3 x i8>* %src.addr, align 1
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
%cvt = sitofp <3 x i8> %load to <3 x float>
[X86] Reduced alignment of widened vector load/stores to better match PR26953 cases llvm-svn: 263649
2016-03-17 02:32:44 +08:00
store <3 x float> %cvt, <3 x float>* %dst.addr, align 4
[X86] Regenerated + extended widened vector conversion tests - Ensure we test X86 + X64 - sitopfp / uitofp requires testing for SSE2 and SSE42 as well (part of the fix for PR26953) llvm-svn: 263640
2016-03-16 23:33:43 +08:00
ret void
}