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/avx-vperm2x128.ll

698 lines
29 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
[X86] Add isel patterns to be able to fold loads into VPERM2F128 even when the load is on the first input to the SDNode. We just need to toggle bits 1 and 5 of the immediate and swap the sources. The peephole pass could trigger commuting/folding for this later, but its easy enough to fix in isel. Disable the peephole pass on the main vperm2x128 test so we know we're doing this through isel. llvm-svn: 313455
2017-09-16 17:16:48 +08:00
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx -disable-peephole | FileCheck %s --check-prefix=ALL --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 -disable-peephole | FileCheck %s --check-prefix=ALL --check-prefix=AVX2
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_45670123(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v8f32_45670123:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_45670123:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
; AVX2-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
ret <8 x float> %shuffle
}
[X86][AVX] Added extra memory folding tests for D19228 llvm-svn: 266662
2016-04-19 03:48:16 +08:00
define <8 x float> @shuffle_v8f32_45670123_mem(<8 x float>* %pa, <8 x float>* %pb) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v8f32_45670123_mem:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3,0,1]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_45670123_mem:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = mem[2,3,0,1]
; AVX2-NEXT: retq
[X86][AVX] Added extra memory folding tests for D19228 llvm-svn: 266662
2016-04-19 03:48:16 +08:00
entry:
%a = load <8 x float>, <8 x float>* %pa
%b = load <8 x float>, <8 x float>* %pb
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_0123cdef(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_0123cdef:
[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
; ALL: # %bb.0: # %entry
[x86] Enable the new vector shuffle lowering by default. Update the entire regression test suite for the new shuffles. Remove most of the old testing which was devoted to the old shuffle lowering path and is no longer relevant really. Also remove a few other random tests that only really exercised shuffles and only incidently or without any interesting aspects to them. Benchmarking that I have done shows a few small regressions with this on LNT, zero measurable regressions on real, large applications, and for several benchmarks where the loop vectorizer fires in the hot path it shows 5% to 40% improvements for SSE2 and SSE3 code running on Sandy Bridge machines. Running on AMD machines shows even more dramatic improvements. When using newer ISA vector extensions the gains are much more modest, but the code is still better on the whole. There are a few regressions being tracked (PR21137, PR21138, PR21139) but by and large this is expected to be a win for x86 generated code performance. It is also more correct than the code it replaces. I have fuzz tested this extensively with ISA extensions up through AVX2 and found no crashes or miscompiles (yet...). The old lowering had a few miscompiles and crashers after a somewhat smaller amount of fuzz testing. There is one significant area where the new code path lags behind and that is in AVX-512 support. However, there was *extremely little* support for that already and so this isn't a significant step backwards and the new framework will probably make it easier to implement lowering that uses the full power of AVX-512's table-based shuffle+blend (IMO). Many thanks to Quentin, Andrea, Robert, and others for benchmarking assistance. Thanks to Adam and others for help with AVX-512. Thanks to Hal, Eric, and *many* others for answering my incessant questions about how the backend actually works. =] I will leave the old code path in the tree until the 3 PRs above are at least resolved to folks' satisfaction. Then I will rip it (and 1000s of lines of code) out. =] I don't expect this flag to stay around for very long. It may not survive next week. llvm-svn: 219046
2014-10-04 11:52:55 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; ALL-NEXT: retq
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 12, i32 13, i32 14, i32 15>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_01230123(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
[X86][AVX2] Prefer VPERMQ/VPERMPD over VINSERTI128/VINSERTF128 for unary shuffles Using VPERMQ/VPERMPD allows memory folding of the (repeated) input where VINSERTI128/VINSERTF128 can not. Differential Revision: http://reviews.llvm.org/D19228 llvm-svn: 266728
2016-04-19 20:26:40 +08:00
; AVX1-LABEL: shuffle_v8f32_01230123:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX2] Prefer VPERMQ/VPERMPD over VINSERTI128/VINSERTF128 for unary shuffles Using VPERMQ/VPERMPD allows memory folding of the (repeated) input where VINSERTI128/VINSERTF128 can not. Differential Revision: http://reviews.llvm.org/D19228 llvm-svn: 266728
2016-04-19 20:26:40 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_01230123:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,0,1]
[X86][AVX2] Prefer VPERMQ/VPERMPD over VINSERTI128/VINSERTF128 for unary shuffles Using VPERMQ/VPERMPD allows memory folding of the (repeated) input where VINSERTI128/VINSERTF128 can not. Differential Revision: http://reviews.llvm.org/D19228 llvm-svn: 266728
2016-04-19 20:26:40 +08:00
; AVX2-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <8 x float> %shuffle
}
[X86][AVX] Added extra memory folding tests for D19228 llvm-svn: 266662
2016-04-19 03:48:16 +08:00
define <8 x float> @shuffle_v8f32_01230123_mem(<8 x float>* %pa, <8 x float>* %pb) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v8f32_01230123_mem:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_01230123_mem:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX2-NEXT: retq
[X86][AVX] Added extra memory folding tests for D19228 llvm-svn: 266662
2016-04-19 03:48:16 +08:00
entry:
%a = load <8 x float>, <8 x float>* %pa
%b = load <8 x float>, <8 x float>* %pb
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_45674567(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v8f32_45674567:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_45674567:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX2-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7>
[X86][AVX] Added an additional vperm2f128 memory folding test llvm-svn: 275413
2016-07-14 21:40:53 +08:00
ret <8 x float> %shuffle
}
define <8 x float> @shuffle_v8f32_45674567_mem(<8 x float>* %pa, <8 x float>* %pb) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v8f32_45674567_mem:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_45674567_mem:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = mem[2,3,2,3]
; AVX2-NEXT: retq
[X86][AVX] Added an additional vperm2f128 memory folding test llvm-svn: 275413
2016-07-14 21:40:53 +08:00
entry:
%a = load <8 x float>, <8 x float>* %pa
%b = load <8 x float>, <8 x float>* %pb
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7>
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 test to make it quicker to review missed one the first time round... llvm-svn: 266623
2016-04-19 00:08:19 +08:00
define <32 x i8> @shuffle_v32i8_2323(<32 x i8> %a, <32 x i8> %b) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v32i8_2323:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v32i8_2323:
[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
; AVX2: # %bb.0: # %entry
[X86] Add VPERMPD/VPERMQ and VPERMPS/VPERMD to the execution domain fixing table. llvm-svn: 313610
2017-09-19 12:39:55 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
ret <32 x i8> %shuffle
}
[X86][AVX] Renamed vperm2f128 test to make it quicker to review missed one the first time round... llvm-svn: 266623
2016-04-19 00:08:19 +08:00
define <32 x i8> @shuffle_v32i8_2323_domain(<32 x i8> %a, <32 x i8> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: shuffle_v32i8_2323_domain:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Renamed vperm2f128 test to make it quicker to review missed one the first time round... llvm-svn: 266623
2016-04-19 00:08:19 +08:00
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
; AVX1-NEXT: vpsubb %xmm1, %xmm0, %xmm0
[X86][AVX] Renamed vperm2f128 test to make it quicker to review missed one the first time round... llvm-svn: 266623
2016-04-19 00:08:19 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v32i8_2323_domain:
[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
; AVX2: # %bb.0: # %entry
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX2-NEXT: vpcmpeqd %ymm1, %ymm1, %ymm1
; AVX2-NEXT: vpsubb %ymm1, %ymm0, %ymm0
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
[X86][AVX] Renamed vperm2f128 test to make it quicker to review missed one the first time round... llvm-svn: 266623
2016-04-19 00:08:19 +08:00
; AVX2-NEXT: retq
entry:
; add forces execution domain
%a2 = add <32 x i8> %a, <i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1>
%shuffle = shufflevector <32 x i8> %a2, <32 x i8> %b, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
ret <32 x i8> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x i64> @shuffle_v4i64_6701(<4 x i64> %a, <4 x i64> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v4i64_6701:
[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
; ALL: # %bb.0: # %entry
[x86] Teach the new vector shuffle lowering how to lower 128-bit shuffles using AVX and AVX2 instructions. This fixes PR21138, one of the few remaining regressions impacting benchmarks from the new vector shuffle lowering. You may note that it "regresses" many of the vperm2x128 test cases -- these were actually "improved" by the naive lowering that the new shuffle lowering previously did. This regression gave me fits. I had this patch ready-to-go about an hour after flipping the switch but wasn't sure how to have the best of both worlds here and thought the correct solution might be a completely different approach to lowering these vector shuffles. I'm now convinced this is the correct lowering and the missed optimizations shown in vperm2x128 are actually due to missing target-independent DAG combines. I've even written most of the needed DAG combine and will submit it shortly, but this part is ready and should help some real-world benchmarks out. llvm-svn: 219079
2014-10-05 19:41:36 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[0,1]
; ALL-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
ret <4 x i64> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x i64> @shuffle_v4i64_6701_domain(<4 x i64> %a, <4 x i64> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: shuffle_v4i64_6701_domain:
[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
; AVX1: # %bb.0: # %entry
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX1-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpsubq %xmm2, %xmm0, %xmm0
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[0,1]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX1-NEXT: retq
;
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX2-LABEL: shuffle_v4i64_6701_domain:
[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
; AVX2: # %bb.0: # %entry
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX2-NEXT: vpcmpeqd %ymm2, %ymm2, %ymm2
; AVX2-NEXT: vpsubq %ymm2, %ymm0, %ymm0
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[0,1]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX2-NEXT: retq
entry:
; add forces execution domain
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
%a2 = add <4 x i64> %a, <i64 1, i64 1, i64 1, i64 1>
%shuffle = shufflevector <4 x i64> %a2, <4 x i64> %b, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
ret <4 x i64> %shuffle
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x i32> @shuffle_v8i32_u5u7cdef(<8 x i32> %a, <8 x i32> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: shuffle_v8i32_u5u7cdef:
[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
; AVX1: # %bb.0: # %entry
[x86, dag] Teach the DAG combiner to prune inputs toa vector_shuffle that are unused. This allows the combiner to delete math feeding shuffles where the math isn't actually necessary. This improves some of the vperm2x128 tests that regressed when the vector shuffle lowering started actually generating vperm instructions rather than forcibly decomposing them. Sadly, this isn't enough to get this *really* right because we still form a completely unnecessary permutation. To fix that, we also need to fold shuffles which just rearrange concatenated or inserted subvectors. llvm-svn: 219086
2014-10-06 03:14:34 +08:00
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX1-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpsubd %xmm2, %xmm0, %xmm0
[x86, dag] Teach the DAG combiner to prune inputs toa vector_shuffle that are unused. This allows the combiner to delete math feeding shuffles where the math isn't actually necessary. This improves some of the vperm2x128 tests that regressed when the vector shuffle lowering started actually generating vperm instructions rather than forcibly decomposing them. Sadly, this isn't enough to get this *really* right because we still form a completely unnecessary permutation. To fix that, we also need to fold shuffles which just rearrange concatenated or inserted subvectors. llvm-svn: 219086
2014-10-06 03:14:34 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
[x86] Teach the new vector shuffle lowering how to lower 128-bit shuffles using AVX and AVX2 instructions. This fixes PR21138, one of the few remaining regressions impacting benchmarks from the new vector shuffle lowering. You may note that it "regresses" many of the vperm2x128 test cases -- these were actually "improved" by the naive lowering that the new shuffle lowering previously did. This regression gave me fits. I had this patch ready-to-go about an hour after flipping the switch but wasn't sure how to have the best of both worlds here and thought the correct solution might be a completely different approach to lowering these vector shuffles. I'm now convinced this is the correct lowering and the missed optimizations shown in vperm2x128 are actually due to missing target-independent DAG combines. I've even written most of the needed DAG combine and will submit it shortly, but this part is ready and should help some real-world benchmarks out. llvm-svn: 219079
2014-10-05 19:41:36 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX1-NEXT: retq
;
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX2-LABEL: shuffle_v8i32_u5u7cdef:
[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
; AVX2: # %bb.0: # %entry
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX2-NEXT: vpcmpeqd %ymm2, %ymm2, %ymm2
; AVX2-NEXT: vpsubd %ymm2, %ymm0, %ymm0
[x86] Teach the new vector shuffle lowering how to lower 128-bit shuffles using AVX and AVX2 instructions. This fixes PR21138, one of the few remaining regressions impacting benchmarks from the new vector shuffle lowering. You may note that it "regresses" many of the vperm2x128 test cases -- these were actually "improved" by the naive lowering that the new shuffle lowering previously did. This regression gave me fits. I had this patch ready-to-go about an hour after flipping the switch but wasn't sure how to have the best of both worlds here and thought the correct solution might be a completely different approach to lowering these vector shuffles. I'm now convinced this is the correct lowering and the missed optimizations shown in vperm2x128 are actually due to missing target-independent DAG combines. I've even written most of the needed DAG combine and will submit it shortly, but this part is ready and should help some real-world benchmarks out. llvm-svn: 219079
2014-10-05 19:41:36 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX2-NEXT: retq
entry:
; add forces execution domain
%a2 = add <8 x i32> %a, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%shuffle = shufflevector <8 x i32> %a2, <8 x i32> %b, <8 x i32> <i32 undef, i32 5, i32 undef, i32 7, i32 12, i32 13, i32 14, i32 15>
ret <8 x i32> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <16 x i16> @shuffle_v16i16_4501(<16 x i16> %a, <16 x i16> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: shuffle_v16i16_4501:
[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
; AVX1: # %bb.0: # %entry
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX1-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpsubw %xmm2, %xmm0, %xmm0
[x86] Teach the new vector shuffle lowering how to lower 128-bit shuffles using AVX and AVX2 instructions. This fixes PR21138, one of the few remaining regressions impacting benchmarks from the new vector shuffle lowering. You may note that it "regresses" many of the vperm2x128 test cases -- these were actually "improved" by the naive lowering that the new shuffle lowering previously did. This regression gave me fits. I had this patch ready-to-go about an hour after flipping the switch but wasn't sure how to have the best of both worlds here and thought the correct solution might be a completely different approach to lowering these vector shuffles. I'm now convinced this is the correct lowering and the missed optimizations shown in vperm2x128 are actually due to missing target-independent DAG combines. I've even written most of the needed DAG combine and will submit it shortly, but this part is ready and should help some real-world benchmarks out. llvm-svn: 219079
2014-10-05 19:41:36 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX1-NEXT: retq
;
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX2-LABEL: shuffle_v16i16_4501:
[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
; AVX2: # %bb.0: # %entry
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX2-NEXT: vpcmpeqd %ymm2, %ymm2, %ymm2
; AVX2-NEXT: vpsubw %ymm2, %ymm0, %ymm0
[x86] Teach the new vector shuffle lowering how to lower 128-bit shuffles using AVX and AVX2 instructions. This fixes PR21138, one of the few remaining regressions impacting benchmarks from the new vector shuffle lowering. You may note that it "regresses" many of the vperm2x128 test cases -- these were actually "improved" by the naive lowering that the new shuffle lowering previously did. This regression gave me fits. I had this patch ready-to-go about an hour after flipping the switch but wasn't sure how to have the best of both worlds here and thought the correct solution might be a completely different approach to lowering these vector shuffles. I'm now convinced this is the correct lowering and the missed optimizations shown in vperm2x128 are actually due to missing target-independent DAG combines. I've even written most of the needed DAG combine and will submit it shortly, but this part is ready and should help some real-world benchmarks out. llvm-svn: 219079
2014-10-05 19:41:36 +08:00
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX2-NEXT: retq
entry:
; add forces execution domain
%a2 = add <16 x i16> %a, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%shuffle = shufflevector <16 x i16> %a2, <16 x i16> %b, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
ret <16 x i16> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <16 x i16> @shuffle_v16i16_4501_mem(<16 x i16>* %a, <16 x i16>* %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: shuffle_v16i16_4501_mem:
[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
; AVX1: # %bb.0: # %entry
[x86] Enable the new vector shuffle lowering by default. Update the entire regression test suite for the new shuffles. Remove most of the old testing which was devoted to the old shuffle lowering path and is no longer relevant really. Also remove a few other random tests that only really exercised shuffles and only incidently or without any interesting aspects to them. Benchmarking that I have done shows a few small regressions with this on LNT, zero measurable regressions on real, large applications, and for several benchmarks where the loop vectorizer fires in the hot path it shows 5% to 40% improvements for SSE2 and SSE3 code running on Sandy Bridge machines. Running on AMD machines shows even more dramatic improvements. When using newer ISA vector extensions the gains are much more modest, but the code is still better on the whole. There are a few regressions being tracked (PR21137, PR21138, PR21139) but by and large this is expected to be a win for x86 generated code performance. It is also more correct than the code it replaces. I have fuzz tested this extensively with ISA extensions up through AVX2 and found no crashes or miscompiles (yet...). The old lowering had a few miscompiles and crashers after a somewhat smaller amount of fuzz testing. There is one significant area where the new code path lags behind and that is in AVX-512 support. However, there was *extremely little* support for that already and so this isn't a significant step backwards and the new framework will probably make it easier to implement lowering that uses the full power of AVX-512's table-based shuffle+blend (IMO). Many thanks to Quentin, Andrea, Robert, and others for benchmarking assistance. Thanks to Adam and others for help with AVX-512. Thanks to Hal, Eric, and *many* others for answering my incessant questions about how the backend actually works. =] I will leave the old code path in the tree until the 3 PRs above are at least resolved to folks' satisfaction. Then I will rip it (and 1000s of lines of code) out. =] I don't expect this flag to stay around for very long. It may not survive next week. llvm-svn: 219046
2014-10-04 11:52:55 +08:00
; AVX1-NEXT: vmovdqa (%rdi), %ymm0
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX1-NEXT: vpcmpeqd %xmm1, %xmm1, %xmm1
; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm0
[x86] use vperm2f128 rather than vinsertf128 when there's a chance to fold a 32-byte load I was looking closer at the x86 test diffs in D33866, and the first change seems like it shouldn't happen in the first place. So this patch will resolve that. Using Agner's tables and AMD docs, vperm2f128 and vinsertf128 have identical timing for any given CPU model, so we should be able to interchange those without affecting perf. But as we can see in some of the diffs here, using vperm2f128 allows load folding, so we should take that opportunity to reduce code size and register pressure. A secondary advantage is making AVX1 and AVX2 codegen more similar. Given that vperm2f128 was introduced with AVX1, we should be selecting it in all of the same situations that we would with AVX2. If there's some reason that an AVX1 CPU would not want to use this instruction, that should be fixed up in a later pass. Differential Revision: https://reviews.llvm.org/D33938 llvm-svn: 305171
2017-06-12 05:18:58 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[0,1],ymm0[0,1]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX1-NEXT: retq
;
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX2-LABEL: shuffle_v16i16_4501_mem:
[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
; AVX2: # %bb.0: # %entry
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX2-NEXT: vmovdqa (%rdi), %ymm0
[x86] transform vector inc/dec to use -1 constant (PR33483) Convert vector increment or decrement to sub/add with an all-ones constant: add X, <1, 1...> --> sub X, <-1, -1...> sub X, <1, 1...> --> add X, <-1, -1...> The all-ones vector constant can be materialized using a pcmpeq instruction that is commonly recognized as an idiom (has no register dependency), so that's better than loading a splat 1 constant. AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better way to produce 512 one-bits. The general advantages of this lowering are: 1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables, so in theory, this could be better for perf, but... 2. That seems unlikely to affect any OOO implementation, and I can't measure any real perf difference from this transform on Haswell or Jaguar, but... 3. It doesn't look like it from the diffs, but this is an overall size win because we eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting a scalar load (which might itself be a bug), then we're replacing a scalar constant load + broadcast with a single cheap op, so that should always be smaller/better too. 4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1 and psub x, -1, so we should use that form for +1 too because we can. If there's some reason to favor a constant load on some CPU, let's make the reverse transform for all of these cases (either here in the DAG or in a later machine pass). This should fix: https://bugs.llvm.org/show_bug.cgi?id=33483 Differential Revision: https://reviews.llvm.org/D34336 llvm-svn: 306289
2017-06-26 22:19:26 +08:00
; AVX2-NEXT: vpcmpeqd %ymm1, %ymm1, %ymm1
; AVX2-NEXT: vpsubw %ymm1, %ymm0, %ymm0
[x86] use vperm2f128 rather than vinsertf128 when there's a chance to fold a 32-byte load I was looking closer at the x86 test diffs in D33866, and the first change seems like it shouldn't happen in the first place. So this patch will resolve that. Using Agner's tables and AMD docs, vperm2f128 and vinsertf128 have identical timing for any given CPU model, so we should be able to interchange those without affecting perf. But as we can see in some of the diffs here, using vperm2f128 allows load folding, so we should take that opportunity to reduce code size and register pressure. A secondary advantage is making AVX1 and AVX2 codegen more similar. Given that vperm2f128 was introduced with AVX1, we should be selecting it in all of the same situations that we would with AVX2. If there's some reason that an AVX1 CPU would not want to use this instruction, that should be fixed up in a later pass. Differential Revision: https://reviews.llvm.org/D33938 llvm-svn: 305171
2017-06-12 05:18:58 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = mem[0,1],ymm0[0,1]
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
; AVX2-NEXT: retq
entry:
[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
%c = load <16 x i16>, <16 x i16>* %a
%d = load <16 x i16>, <16 x i16>* %b
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
%c2 = add <16 x i16> %c, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%shuffle = shufflevector <16 x i16> %c2, <16 x i16> %d, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
ret <16 x i16> %shuffle
}
;;;; Cases with undef indicies mixed in the mask
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu67u9ub(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_uu67u9ub:
[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
; ALL: # %bb.0: # %entry
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
[x86] Teach the new vector shuffle lowering how to lower 128-bit shuffles using AVX and AVX2 instructions. This fixes PR21138, one of the few remaining regressions impacting benchmarks from the new vector shuffle lowering. You may note that it "regresses" many of the vperm2x128 test cases -- these were actually "improved" by the naive lowering that the new shuffle lowering previously did. This regression gave me fits. I had this patch ready-to-go about an hour after flipping the switch but wasn't sure how to have the best of both worlds here and thought the correct solution might be a completely different approach to lowering these vector shuffles. I'm now convinced this is the correct lowering and the missed optimizations shown in vperm2x128 are actually due to missing target-independent DAG combines. I've even written most of the needed DAG combine and will submit it shortly, but this part is ready and should help some real-world benchmarks out. llvm-svn: 219079
2014-10-05 19:41:36 +08:00
; ALL-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 9, i32 undef, i32 11>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu67uu67(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
[X86] Prefer VPERMQ over VPERM2F128 for any unary shuffle, not just the ones that can be done with a insertf128 The early out for AVX2 in lowerV2X128VectorShuffle is positioned in a weird spot below some shuffle mask equivalency checks. But I think we want to allow VPERMQ for any unary shuffle. Differential Revision: https://reviews.llvm.org/D37893 llvm-svn: 313373
2017-09-16 02:11:13 +08:00
; AVX1-LABEL: shuffle_v8f32_uu67uu67:
[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
; AVX1: # %bb.0: # %entry
[X86] Prefer VPERMQ over VPERM2F128 for any unary shuffle, not just the ones that can be done with a insertf128 The early out for AVX2 in lowerV2X128VectorShuffle is positioned in a weird spot below some shuffle mask equivalency checks. But I think we want to allow VPERMQ for any unary shuffle. Differential Revision: https://reviews.llvm.org/D37893 llvm-svn: 313373
2017-09-16 02:11:13 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_uu67uu67:
[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
; AVX2: # %bb.0: # %entry
[X86] Prefer VPERMQ over VPERM2F128 for any unary shuffle, not just the ones that can be done with a insertf128 The early out for AVX2 in lowerV2X128VectorShuffle is positioned in a weird spot below some shuffle mask equivalency checks. But I think we want to allow VPERMQ for any unary shuffle. Differential Revision: https://reviews.llvm.org/D37893 llvm-svn: 313373
2017-09-16 02:11:13 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
; AVX2-NEXT: retq
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu67uuab(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_uu67uuab:
[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
; ALL: # %bb.0: # %entry
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; ALL-NEXT: retq
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 10, i32 11>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu67uuef(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_uu67uuef:
[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
; ALL: # %bb.0: # %entry
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; ALL-NEXT: retq
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 14, i32 15>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu674567(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
[X86] Prefer VPERMQ over VPERM2F128 for any unary shuffle, not just the ones that can be done with a insertf128 The early out for AVX2 in lowerV2X128VectorShuffle is positioned in a weird spot below some shuffle mask equivalency checks. But I think we want to allow VPERMQ for any unary shuffle. Differential Revision: https://reviews.llvm.org/D37893 llvm-svn: 313373
2017-09-16 02:11:13 +08:00
; AVX1-LABEL: shuffle_v8f32_uu674567:
[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
; AVX1: # %bb.0: # %entry
[X86] Prefer VPERMQ over VPERM2F128 for any unary shuffle, not just the ones that can be done with a insertf128 The early out for AVX2 in lowerV2X128VectorShuffle is positioned in a weird spot below some shuffle mask equivalency checks. But I think we want to allow VPERMQ for any unary shuffle. Differential Revision: https://reviews.llvm.org/D37893 llvm-svn: 313373
2017-09-16 02:11:13 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_uu674567:
[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
; AVX2: # %bb.0: # %entry
[X86] Prefer VPERMQ over VPERM2F128 for any unary shuffle, not just the ones that can be done with a insertf128 The early out for AVX2 in lowerV2X128VectorShuffle is positioned in a weird spot below some shuffle mask equivalency checks. But I think we want to allow VPERMQ for any unary shuffle. Differential Revision: https://reviews.llvm.org/D37893 llvm-svn: 313373
2017-09-16 02:11:13 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
; AVX2-NEXT: retq
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu6789ab(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_uu6789ab:
[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
; ALL: # %bb.0: # %entry
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
; ALL-NEXT: retq
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_4567uu67(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-LABEL: shuffle_v8f32_4567uu67:
[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
; AVX1: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
; AVX2-LABEL: shuffle_v8f32_4567uu67:
[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
; AVX2: # %bb.0: # %entry
[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there. If we have VPERMQ/PD we should prefer those since they only have a single input. Differential Revision: https://reviews.llvm.org/D37947 llvm-svn: 313542
2017-09-19 00:39:49 +08:00
; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX2-NEXT: retq
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7>
ret <8 x float> %shuffle
}
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_4567uuef(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_4567uuef:
[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
; ALL: # %bb.0: # %entry
[X86][AVX] Fix wrong lowering of VPERM2X128 nodes There were cases where the backend computed a wrong permute mask for a VPERM2X128 node. Example: \code define <8 x float> @foo(<8 x float> %a, <8 x float> %b) { %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7> ret <8 x float> %shuffle } \code end Before this patch, llc (with -mattr=+avx) emitted the following vperm2f128: vperm2f128 $0, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[0,1,0,1] With this patch, llc emits a vperm2f128 with a correct permute mask: vperm2f128 $17, %ymm0, %ymm0, %ymm0 # ymm0 = ymm0[2,3,2,3] Differential Revision: http://reviews.llvm.org/D8119 llvm-svn: 231601
2015-03-09 00:28:47 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; ALL-NEXT: retq
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 14, i32 15>
ret <8 x float> %shuffle
}
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
;;;; Cases we must not select vperm2f128
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <8 x float> @shuffle_v8f32_uu67ucuf(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
; ALL-LABEL: shuffle_v8f32_uu67ucuf:
[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
; ALL: # %bb.0: # %entry
[x86] Teach the x86 vector shuffle lowering to detect mergable 128-bit lanes. By special casing these we can often either reduce the total number of shuffles significantly or reduce the number of (high latency on Haswell) AVX2 shuffles that potentially cross 128-bit lanes. Even when these don't actually cross lanes, they have much higher latency to support that. Doing two of them and a blend is worse than doing a single insert across the 128-bit lanes to blend and then doing a single interleaved shuffle. While this seems like a narrow case, it kept cropping up on me and the difference is *huge* as you can see in many of the test cases. I first hit this trying to perfectly fix the interleaving shuffle patterns used by Halide for AVX2. llvm-svn: 222533
2014-11-21 21:56:05 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; ALL-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,0,2,3,4,4,6,7]
; ALL-NEXT: retq
[x86] Move the vperm2f128 test to be vperm2x128 and test both the floating point and integer domains. Merge the AVX2 test into it and add an extra RUN line. Generate clean FileCheck statements with my script. Remove the now merged AVX2 tests. llvm-svn: 218903
2014-10-03 04:11:11 +08:00
entry:
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 undef, i32 12, i32 undef, i32 15>
ret <8 x float> %shuffle
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Regenerate tests. llvm-svn: 251263
2015-10-26 05:47:09 +08:00
;; Test zero mask generation.
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
;; PR22984: https://llvm.org/bugs/show_bug.cgi?id=22984
;; Prefer xor+vblendpd over vperm2f128 because that has better performance.
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
;; TODO: When building for optsize we should use vperm2f128.
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_zz01(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_zz01:
[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
; ALL: # %bb.0:
[X86][AVX] Add support for shuffle decoding of vperm2f128/vperm2i128 with zero'd lanes The vperm2f128/vperm2i128 shuffle mask decoding was not attempting to deal with shuffles that give zero lanes. This patch fixes this so that the assembly printer can provide shuffle comments. As this decoder is also used in X86ISelLowering for shuffle combining, I've added an early-out to match existing behaviour. The hope is that we can add zero support in the future, this would allow other ops' decodes (e.g. insertps) to be combined as well. Differential Revision: http://reviews.llvm.org/D10593 llvm-svn: 241516
2015-07-07 06:46:46 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = zero,zero,ymm0[0,1]
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_zz01_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_zz01_optsize:
[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
; ALL: # %bb.0:
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = zero,zero,ymm0[0,1]
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_zz23(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_zz23:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][AVX] Add support for shuffle decoding of vperm2f128/vperm2i128 with zero'd lanes The vperm2f128/vperm2i128 shuffle mask decoding was not attempting to deal with shuffles that give zero lanes. This patch fixes this so that the assembly printer can provide shuffle comments. As this decoder is also used in X86ISelLowering for shuffle combining, I've added an early-out to match existing behaviour. The hope is that we can add zero support in the future, this would allow other ops' decodes (e.g. insertps) to be combined as well. Differential Revision: http://reviews.llvm.org/D10593 llvm-svn: 241516
2015-07-07 06:46:46 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_zz23_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_zz23_optsize:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_zz45(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_zz45:
[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
; ALL: # %bb.0:
[X86][AVX] Add support for shuffle decoding of vperm2f128/vperm2i128 with zero'd lanes The vperm2f128/vperm2i128 shuffle mask decoding was not attempting to deal with shuffles that give zero lanes. This patch fixes this so that the assembly printer can provide shuffle comments. As this decoder is also used in X86ISelLowering for shuffle combining, I've added an early-out to match existing behaviour. The hope is that we can add zero support in the future, this would allow other ops' decodes (e.g. insertps) to be combined as well. Differential Revision: http://reviews.llvm.org/D10593 llvm-svn: 241516
2015-07-07 06:46:46 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = zero,zero,ymm0[0,1]
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_zz45_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_zz45_optsize:
[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
; ALL: # %bb.0:
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = zero,zero,ymm0[0,1]
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_zz67(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_zz67:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][AVX] Add support for shuffle decoding of vperm2f128/vperm2i128 with zero'd lanes The vperm2f128/vperm2i128 shuffle mask decoding was not attempting to deal with shuffles that give zero lanes. This patch fixes this so that the assembly printer can provide shuffle comments. As this decoder is also used in X86ISelLowering for shuffle combining, I've added an early-out to match existing behaviour. The hope is that we can add zero support in the future, this would allow other ops' decodes (e.g. insertps) to be combined as well. Differential Revision: http://reviews.llvm.org/D10593 llvm-svn: 241516
2015-07-07 06:46:46 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_zz67_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_zz67_optsize:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_01zz(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_01zz:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][SSE] Shuffle blends with zero This patch generalizes the zeroing of vector elements with the BLEND instructions. Currently a zero vector will only blend if the shuffled elements are correctly inline, this patch recognises when a vector input is zero (or zeroable) and modifies a local copy of the shuffle mask to support a blend. As a zeroable vector input may not be all zeroes, the zeroable vector is regenerated if necessary. Differential Revision: http://reviews.llvm.org/D14050 llvm-svn: 251659
2015-10-30 06:11:28 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_01zz_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_01zz_optsize:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_23zz(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_23zz:
[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
; ALL: # %bb.0:
[X86][AVX] Add support for shuffle decoding of vperm2f128/vperm2i128 with zero'd lanes The vperm2f128/vperm2i128 shuffle mask decoding was not attempting to deal with shuffles that give zero lanes. This patch fixes this so that the assembly printer can provide shuffle comments. As this decoder is also used in X86ISelLowering for shuffle combining, I've added an early-out to match existing behaviour. The hope is that we can add zero support in the future, this would allow other ops' decodes (e.g. insertps) to be combined as well. Differential Revision: http://reviews.llvm.org/D10593 llvm-svn: 241516
2015-07-07 06:46:46 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_23zz_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_23zz_optsize:
[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
; ALL: # %bb.0:
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero
; ALL-NEXT: retq
%s = shufflevector <4 x double> %a, <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_45zz(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_45zz:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][SSE] Shuffle blends with zero This patch generalizes the zeroing of vector elements with the BLEND instructions. Currently a zero vector will only blend if the shuffled elements are correctly inline, this patch recognises when a vector input is zero (or zeroable) and modifies a local copy of the shuffle mask to support a blend. As a zeroable vector input may not be all zeroes, the zeroable vector is regenerated if necessary. Differential Revision: http://reviews.llvm.org/D14050 llvm-svn: 251659
2015-10-30 06:11:28 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
ret <4 x double> %s
}
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
define <4 x double> @shuffle_v4f64_45zz_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_45zz_optsize:
[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
; ALL: # %bb.0:
[X86] SET0 to use XMM registers where possible PR26018 PR32862 Differential Revision: https://reviews.llvm.org/D35839 llvm-svn: 309298
2017-07-28 01:47:01 +08:00
; ALL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
ret <4 x double> %s
}
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x double> @shuffle_v4f64_67zz(<4 x double> %a) {
; ALL-LABEL: shuffle_v4f64_67zz:
[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
; ALL: # %bb.0:
[X86][AVX] Add support for shuffle decoding of vperm2f128/vperm2i128 with zero'd lanes The vperm2f128/vperm2i128 shuffle mask decoding was not attempting to deal with shuffles that give zero lanes. This patch fixes this so that the assembly printer can provide shuffle comments. As this decoder is also used in X86ISelLowering for shuffle combining, I've added an early-out to match existing behaviour. The hope is that we can add zero support in the future, this would allow other ops' decodes (e.g. insertps) to be combined as well. Differential Revision: http://reviews.llvm.org/D10593 llvm-svn: 241516
2015-07-07 06:46:46 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
ret <4 x double> %s
}
define <4 x double> @shuffle_v4f64_67zz_optsize(<4 x double> %a) optsize {
; ALL-LABEL: shuffle_v4f64_67zz_optsize:
[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
; ALL: # %bb.0:
[X86][AVX] Added zero+blend vs vperm2f128 optsize tests cases (PR22984) We should be trying to use vperm2f128 instead of zero+blend (if we're only the user of zero?) when optsize is enabled. llvm-svn: 266632
2016-04-19 01:14:04 +08:00
; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
; ALL-NEXT: retq
%s = shufflevector <4 x double> <double 0.0, double 0.0, double undef, double undef>, <4 x double> %a, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
ret <4 x double> %s
}
;; With AVX2 select the integer version of the instruction. Use an add to force the domain selection.
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
define <4 x i64> @shuffle_v4i64_67zz(<4 x i64> %a, <4 x i64> %b) {
; AVX1-LABEL: shuffle_v4i64_67zz:
[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
; AVX1: # %bb.0:
[X86][SSE] Shuffle blends with zero This patch generalizes the zeroing of vector elements with the BLEND instructions. Currently a zero vector will only blend if the shuffled elements are correctly inline, this patch recognises when a vector input is zero (or zeroable) and modifies a local copy of the shuffle mask to support a blend. As a zeroable vector input may not be all zeroes, the zeroable vector is regenerated if necessary. Differential Revision: http://reviews.llvm.org/D14050 llvm-svn: 251659
2015-10-30 06:11:28 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero
; AVX1-NEXT: vpaddq %xmm0, %xmm1, %xmm0
[X86] Genericize the handling of INSERT_SUBVECTOR from an EXTRACT_SUBVECTOR to support 512-bit vectors with 128-bit or 256-bit subvectors. We now detect that both the extract and insert indices are non-zero and convert to a shuffle. This will be lowered as a blend for 256-bit vectors or as a vshuf operations for 512-bit vectors. llvm-svn: 294931
2017-02-13 12:53:29 +08:00
; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
[X86][SSE] Shuffle blends with zero This patch generalizes the zeroing of vector elements with the BLEND instructions. Currently a zero vector will only blend if the shuffled elements are correctly inline, this patch recognises when a vector input is zero (or zeroable) and modifies a local copy of the shuffle mask to support a blend. As a zeroable vector input may not be all zeroes, the zeroable vector is regenerated if necessary. Differential Revision: http://reviews.llvm.org/D14050 llvm-svn: 251659
2015-10-30 06:11:28 +08:00
; AVX1-NEXT: retq
;
[X86][AVX] Renamed vperm2f128 tests to make it quicker to review llvm-svn: 266621
2016-04-18 23:37:45 +08:00
; AVX2-LABEL: shuffle_v4i64_67zz:
[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
; AVX2: # %bb.0:
[X86][SSE] Shuffle blends with zero This patch generalizes the zeroing of vector elements with the BLEND instructions. Currently a zero vector will only blend if the shuffled elements are correctly inline, this patch recognises when a vector input is zero (or zeroable) and modifies a local copy of the shuffle mask to support a blend. As a zeroable vector input may not be all zeroes, the zeroable vector is regenerated if necessary. Differential Revision: http://reviews.llvm.org/D14050 llvm-svn: 251659
2015-10-30 06:11:28 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero
; AVX2-NEXT: vpaddq %ymm0, %ymm1, %ymm0
; AVX2-NEXT: retq
[X86, AVX] recognize shufflevector with zero input as a vperm2 (PR22984) vperm2x128 instructions have the special ability (aka free hardware capability) to shuffle zero values into a vector. This patch recognizes that type of shuffle and generates the appropriate control byte. https://llvm.org/bugs/show_bug.cgi?id=22984 Differential Revision: http://reviews.llvm.org/D8563 llvm-svn: 233100
2015-03-25 03:19:07 +08:00
%s = shufflevector <4 x i64> <i64 0, i64 0, i64 undef, i64 undef>, <4 x i64> %a, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
%c = add <4 x i64> %b, %s
ret <4 x i64> %c
}
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
;;; Memory folding cases
define <4 x double> @ld0_hi0_lo1_4f64(<4 x double> * %pa, <4 x double> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld0_hi0_lo1_4f64:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX1-NEXT: vaddpd {{.*}}(%rip), %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld0_hi0_lo1_4f64:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
[X86][AVX] Regenerated and cleaned up AVX1 intrinsic tests. Cleaned up triple settings, added 32-bit/64-bit targets where useful, added broadcast comments llvm-svn: 309100
2017-07-26 18:54:51 +08:00
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1,1,1,1]
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%a = load <4 x double>, <4 x double> * %pa
%shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
%res = fadd <4 x double> %shuffle, <double 1.0, double 1.0, double 1.0, double 1.0>
ret <4 x double> %res
}
define <4 x double> @ld1_hi0_hi1_4f64(<4 x double> %a, <4 x double> * %pb) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld1_hi0_hi1_4f64:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX1-NEXT: vaddpd {{.*}}(%rip), %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld1_hi0_hi1_4f64:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
[X86][AVX] Regenerated and cleaned up AVX1 intrinsic tests. Cleaned up triple settings, added 32-bit/64-bit targets where useful, added broadcast comments llvm-svn: 309100
2017-07-26 18:54:51 +08:00
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1,1,1,1]
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%b = load <4 x double>, <4 x double> * %pb
%shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
%res = fadd <4 x double> %shuffle, <double 1.0, double 1.0, double 1.0, double 1.0>
ret <4 x double> %res
}
define <8 x float> @ld0_hi0_lo1_8f32(<8 x float> * %pa, <8 x float> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld0_hi0_lo1_8f32:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX1-NEXT: vaddps {{.*}}(%rip), %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld0_hi0_lo1_8f32:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
[X86][AVX] Regenerated and cleaned up AVX1 intrinsic tests. Cleaned up triple settings, added 32-bit/64-bit targets where useful, added broadcast comments llvm-svn: 309100
2017-07-26 18:54:51 +08:00
; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%a = load <8 x float>, <8 x float> * %pa
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
%res = fadd <8 x float> %shuffle, <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>
ret <8 x float> %res
}
define <8 x float> @ld1_hi0_hi1_8f32(<8 x float> %a, <8 x float> * %pb) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld1_hi0_hi1_8f32:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX1-NEXT: vaddps {{.*}}(%rip), %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld1_hi0_hi1_8f32:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
[X86][AVX] Regenerated and cleaned up AVX1 intrinsic tests. Cleaned up triple settings, added 32-bit/64-bit targets where useful, added broadcast comments llvm-svn: 309100
2017-07-26 18:54:51 +08:00
; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%b = load <8 x float>, <8 x float> * %pb
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
%res = fadd <8 x float> %shuffle, <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>
ret <8 x float> %res
}
define <4 x i64> @ld0_hi0_lo1_4i64(<4 x i64> * %pa, <4 x i64> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld0_hi0_lo1_4i64:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX1-NEXT: vpaddq {{.*}}(%rip), %xmm0, %xmm1
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT: vpaddq {{.*}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld0_hi0_lo1_4i64:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX2-NEXT: vpaddq {{.*}}(%rip), %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%a = load <4 x i64>, <4 x i64> * %pa
%shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
%res = add <4 x i64> %shuffle, <i64 1, i64 2, i64 3, i64 4>
ret <4 x i64> %res
}
define <4 x i64> @ld1_hi0_hi1_4i64(<4 x i64> %a, <4 x i64> * %pb) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld1_hi0_hi1_4i64:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX1-NEXT: vpaddq {{.*}}(%rip), %xmm0, %xmm1
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT: vpaddq {{.*}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld1_hi0_hi1_4i64:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX2-NEXT: vpaddq {{.*}}(%rip), %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%b = load <4 x i64>, <4 x i64> * %pb
%shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
%res = add <4 x i64> %shuffle, <i64 1, i64 2, i64 3, i64 4>
ret <4 x i64> %res
}
define <8 x i32> @ld0_hi0_lo1_8i32(<8 x i32> * %pa, <8 x i32> %b) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld0_hi0_lo1_8i32:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [1,2,3,4]
; AVX1-NEXT: vpaddd %xmm2, %xmm1, %xmm1
; AVX1-NEXT: vpaddd %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld0_hi0_lo1_8i32:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX2-NEXT: vpaddd {{.*}}(%rip), %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%a = load <8 x i32>, <8 x i32> * %pa
%shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
%res = add <8 x i32> %shuffle, <i32 1, i32 2, i32 3, i32 4, i32 1, i32 2, i32 3, i32 4>
ret <8 x i32> %res
}
define <8 x i32> @ld1_hi0_hi1_8i32(<8 x i32> %a, <8 x i32> * %pb) nounwind uwtable readnone ssp {
; AVX1-LABEL: ld1_hi0_hi1_8i32:
[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
; AVX1: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [1,2,3,4]
; AVX1-NEXT: vpaddd %xmm2, %xmm1, %xmm1
; AVX1-NEXT: vpaddd %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: ld1_hi0_hi1_8i32:
[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
; AVX2: # %bb.0: # %entry
[X86][AVX] Add commutation support for VPERM2X128 instructions Its main use is to allow memory folding of the 1st operand Differential Revision: http://reviews.llvm.org/D16521 llvm-svn: 258726
2016-01-26 05:51:34 +08:00
; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX2-NEXT: vpaddd {{.*}}(%rip), %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
%b = load <8 x i32>, <8 x i32> * %pb
%shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 12, i32 13, i32 14, i32 15>
%res = add <8 x i32> %shuffle, <i32 1, i32 2, i32 3, i32 4, i32 1, i32 2, i32 3, i32 4>
ret <8 x i32> %res
}