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/widened-broadcast.ll

563 lines
22 KiB
LLVM
Raw Normal View History

[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=SSE --check-prefix=SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=SSE --check-prefix=SSE42
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX --check-prefix=AVX512 --check-prefix=AVX512F
; Widened shuffle broadcast loads
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
define <4 x float> @load_splat_4f32_4f32_0101(<4 x float>* %ptr) nounwind uwtable readnone ssp {
; SSE2-LABEL: load_splat_4f32_4f32_0101:
[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
; SSE2: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; SSE2-NEXT: movaps (%rdi), %xmm0
; SSE2-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0,0]
; SSE2-NEXT: retq
;
; SSE42-LABEL: load_splat_4f32_4f32_0101:
[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
; SSE42: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; SSE42-NEXT: movddup {{.*#+}} xmm0 = mem[0,0]
; SSE42-NEXT: retq
;
; AVX-LABEL: load_splat_4f32_4f32_0101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; AVX-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
; AVX-NEXT: retq
entry:
%ld = load <4 x float>, <4 x float>* %ptr
%ret = shufflevector <4 x float> %ld, <4 x float> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
ret <4 x float> %ret
}
define <8 x float> @load_splat_8f32_4f32_01010101(<4 x float>* %ptr) nounwind uwtable readnone ssp {
; SSE2-LABEL: load_splat_8f32_4f32_01010101:
[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
; SSE2: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; SSE2-NEXT: movaps (%rdi), %xmm0
; SSE2-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0,0]
; SSE2-NEXT: movaps %xmm0, %xmm1
; SSE2-NEXT: retq
;
; SSE42-LABEL: load_splat_8f32_4f32_01010101:
[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
; SSE42: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; SSE42-NEXT: movddup {{.*#+}} xmm0 = mem[0,0]
; SSE42-NEXT: movapd %xmm0, %xmm1
; SSE42-NEXT: retq
;
; AVX1-LABEL: load_splat_8f32_4f32_01010101:
[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][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_8f32_4f32_01010101:
[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] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_8f32_4f32_01010101:
[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
; AVX512: # %bb.0: # %entry
[X86] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX512-NEXT: vbroadcastsd (%rdi), %ymm0
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; AVX512-NEXT: retq
entry:
%ld = load <4 x float>, <4 x float>* %ptr
%ret = shufflevector <4 x float> %ld, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <8 x float> %ret
}
define <8 x float> @load_splat_8f32_8f32_01010101(<8 x float>* %ptr) nounwind uwtable readnone ssp {
; SSE2-LABEL: load_splat_8f32_8f32_01010101:
[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
; SSE2: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; SSE2-NEXT: movaps (%rdi), %xmm0
; SSE2-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0,0]
; SSE2-NEXT: movaps %xmm0, %xmm1
; SSE2-NEXT: retq
;
; SSE42-LABEL: load_splat_8f32_8f32_01010101:
[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
; SSE42: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; SSE42-NEXT: movddup {{.*#+}} xmm0 = mem[0,0]
; SSE42-NEXT: movapd %xmm0, %xmm1
; SSE42-NEXT: retq
;
; AVX-LABEL: load_splat_8f32_8f32_01010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0: # %entry
[X86][SSE] Added float widened broadcast tests llvm-svn: 276535
2016-07-24 05:24:02 +08:00
; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX-NEXT: retq
entry:
%ld = load <8 x float>, <8 x float>* %ptr
%ret = shufflevector <8 x float> %ld, <8 x float> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <8 x float> %ret
}
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
define <4 x i32> @load_splat_4i32_4i32_0101(<4 x i32>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_4i32_4i32_0101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_4i32_4i32_0101:
[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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,1,0,1]
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_4i32_4i32_0101:
[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][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX2-NEXT: vpbroadcastq (%rdi), %xmm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_4i32_4i32_0101:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX512-NEXT: vpbroadcastq (%rdi), %xmm0
; AVX512-NEXT: retq
entry:
%ld = load <4 x i32>, <4 x i32>* %ptr
%ret = shufflevector <4 x i32> %ld, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
ret <4 x i32> %ret
}
define <8 x i32> @load_splat_8i32_4i32_01010101(<4 x i32>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_8i32_4i32_01010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
[X86][AVX] lowerVectorShuffleAsBroadcast - aggressively peek through BITCASTs Assuming we can safely adjust the broadcast index for the new type to keep it suitably aligned, then peek through BITCASTs when looking for the broadcast source. Fixes PR32007 llvm-svn: 320933
2017-12-17 07:32:18 +08:00
; AVX-LABEL: load_splat_8i32_4i32_01010101:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX-NEXT: retq
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
entry:
%ld = load <4 x i32>, <4 x i32>* %ptr
%ret = shufflevector <4 x i32> %ld, <4 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <8 x i32> %ret
}
define <8 x i32> @load_splat_8i32_8i32_01010101(<8 x i32>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_8i32_8i32_01010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_8i32_8i32_01010101:
[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
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_8i32_8i32_01010101:
[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][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_8i32_8i32_01010101:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX512-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX512-NEXT: retq
entry:
%ld = load <8 x i32>, <8 x i32>* %ptr
%ret = shufflevector <8 x i32> %ld, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <8 x i32> %ret
}
define <8 x i16> @load_splat_8i16_8i16_01010101(<8 x i16>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_8i16_8i16_01010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0]
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_8i16_8i16_01010101:
[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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,0,0,0]
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_8i16_8i16_01010101:
[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][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX2-NEXT: vbroadcastss (%rdi), %xmm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_8i16_8i16_01010101:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX512-NEXT: vbroadcastss (%rdi), %xmm0
; AVX512-NEXT: retq
entry:
%ld = load <8 x i16>, <8 x i16>* %ptr
%ret = shufflevector <8 x i16> %ld, <8 x i16> undef, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <8 x i16> %ret
}
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
define <8 x i16> @load_splat_8i16_8i16_01230123(<8 x i16>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_8i16_8i16_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
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_8i16_8i16_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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,1,0,1]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_8i16_8i16_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][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: vpbroadcastq (%rdi), %xmm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_8i16_8i16_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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: vpbroadcastq (%rdi), %xmm0
; AVX512-NEXT: retq
entry:
%ld = load <8 x i16>, <8 x i16>* %ptr
%ret = shufflevector <8 x i16> %ld, <8 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <8 x i16> %ret
}
define <16 x i16> @load_splat_16i16_8i16_0101010101010101(<8 x i16>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i16_8i16_0101010101010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-LABEL: load_splat_16i16_8i16_0101010101010101:
[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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_16i16_8i16_0101010101010101:
[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] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX2-NEXT: vbroadcastss (%rdi), %ymm0
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_16i16_8i16_0101010101010101:
[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
; AVX512: # %bb.0: # %entry
[X86] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX512-NEXT: vbroadcastss (%rdi), %ymm0
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: retq
entry:
%ld = load <8 x i16>, <8 x i16>* %ptr
%ret = shufflevector <8 x i16> %ld, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <16 x i16> %ret
}
define <16 x i16> @load_splat_16i16_8i16_0123012301230123(<8 x i16>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i16_8i16_0123012301230123:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
[X86][AVX] lowerVectorShuffleAsBroadcast - aggressively peek through BITCASTs Assuming we can safely adjust the broadcast index for the new type to keep it suitably aligned, then peek through BITCASTs when looking for the broadcast source. Fixes PR32007 llvm-svn: 320933
2017-12-17 07:32:18 +08:00
; AVX-LABEL: load_splat_16i16_8i16_0123012301230123:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX-NEXT: retq
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
entry:
%ld = load <8 x i16>, <8 x i16>* %ptr
%ret = shufflevector <8 x i16> %ld, <8 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3,i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <16 x i16> %ret
}
define <16 x i16> @load_splat_16i16_16i16_0101010101010101(<16 x i16>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i16_16i16_0101010101010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_16i16_16i16_0101010101010101:
[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
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,0,0,0]
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-LABEL: load_splat_16i16_16i16_0101010101010101:
[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][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX2-NEXT: vbroadcastss (%rdi), %ymm0
; AVX2-NEXT: retq
;
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-LABEL: load_splat_16i16_16i16_0101010101010101:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added tests where we should be trying to widen a load+splat into a broadcast llvm-svn: 276527
2016-07-24 00:19:17 +08:00
; AVX512-NEXT: vbroadcastss (%rdi), %ymm0
; AVX512-NEXT: retq
entry:
%ld = load <16 x i16>, <16 x i16>* %ptr
%ret = shufflevector <16 x i16> %ld, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <16 x i16> %ret
}
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
define <16 x i16> @load_splat_16i16_16i16_0123012301230123(<16 x i16>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i16_16i16_0123012301230123:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX-LABEL: load_splat_16i16_16i16_0123012301230123:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0: # %entry
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX-NEXT: retq
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
entry:
%ld = load <16 x i16>, <16 x i16>* %ptr
%ret = shufflevector <16 x i16> %ld, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3,i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <16 x i16> %ret
}
define <16 x i8> @load_splat_16i8_16i8_0101010101010101(<16 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i8_16i8_0101010101010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Match PSHUFLW/PSHUFHW + PSHUFD vXi16 shuffle patterns (PR34686) As noted in PR34686, we are relying on a PSHUFD+PSHUFLW+PSHUFHW shuffle chain for most general vXi16 unary shuffles. This patch checks for simpler PSHUFLW+PSHUFD and PSHUFHW+PSHUFD cases beforehand, building on some existing code that just handled splat shuffles. By doing so we also prevent premature use of PSHUFB shuffles which can be slower and require the creation/loading of constant shuffle masks. We now have the 'fast-variable-shuffle' option for hardware that prefers combining 2 or more shuffles to VPSHUFB etc. Differential Revision: https://reviews.llvm.org/D38318 llvm-svn: 321553
2017-12-29 22:41:50 +08:00
; SSE-NEXT: pshuflw {{.*#+}} xmm0 = mem[0,0,2,3,4,5,6,7]
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_16i8_16i8_0101010101010101:
[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][SSE] Match PSHUFLW/PSHUFHW + PSHUFD vXi16 shuffle patterns (PR34686) As noted in PR34686, we are relying on a PSHUFD+PSHUFLW+PSHUFHW shuffle chain for most general vXi16 unary shuffles. This patch checks for simpler PSHUFLW+PSHUFD and PSHUFHW+PSHUFD cases beforehand, building on some existing code that just handled splat shuffles. By doing so we also prevent premature use of PSHUFB shuffles which can be slower and require the creation/loading of constant shuffle masks. We now have the 'fast-variable-shuffle' option for hardware that prefers combining 2 or more shuffles to VPSHUFB etc. Differential Revision: https://reviews.llvm.org/D38318 llvm-svn: 321553
2017-12-29 22:41:50 +08:00
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = mem[0,0,2,3,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_16i8_16i8_0101010101010101:
[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][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: vpbroadcastw (%rdi), %xmm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_16i8_16i8_0101010101010101:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: vpbroadcastw (%rdi), %xmm0
; AVX512-NEXT: retq
entry:
%ld = load <16 x i8>, <16 x i8>* %ptr
%ret = shufflevector <16 x i8> %ld, <16 x i8> undef, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <16 x i8> %ret
}
define <16 x i8> @load_splat_16i8_16i8_0123012301230123(<16 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i8_16i8_0123012301230123:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0]
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_16i8_16i8_0123012301230123:
[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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_16i8_16i8_0123012301230123:
[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][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: vbroadcastss (%rdi), %xmm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_16i8_16i8_0123012301230123:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: vbroadcastss (%rdi), %xmm0
; AVX512-NEXT: retq
entry:
%ld = load <16 x i8>, <16 x i8>* %ptr
%ret = shufflevector <16 x i8> %ld, <16 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <16 x i8> %ret
}
define <16 x i8> @load_splat_16i8_16i8_0123456701234567(<16 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_16i8_16i8_0123456701234567:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_16i8_16i8_0123456701234567:
[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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,1,0,1]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_16i8_16i8_0123456701234567:
[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][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: vpbroadcastq (%rdi), %xmm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_16i8_16i8_0123456701234567:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: vpbroadcastq (%rdi), %xmm0
; AVX512-NEXT: retq
entry:
%ld = load <16 x i8>, <16 x i8>* %ptr
%ret = shufflevector <16 x i8> %ld, <16 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
ret <16 x i8> %ret
}
define <32 x i8> @load_splat_32i8_16i8_01010101010101010101010101010101(<16 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_32i8_16i8_01010101010101010101010101010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Match PSHUFLW/PSHUFHW + PSHUFD vXi16 shuffle patterns (PR34686) As noted in PR34686, we are relying on a PSHUFD+PSHUFLW+PSHUFHW shuffle chain for most general vXi16 unary shuffles. This patch checks for simpler PSHUFLW+PSHUFD and PSHUFHW+PSHUFD cases beforehand, building on some existing code that just handled splat shuffles. By doing so we also prevent premature use of PSHUFB shuffles which can be slower and require the creation/loading of constant shuffle masks. We now have the 'fast-variable-shuffle' option for hardware that prefers combining 2 or more shuffles to VPSHUFB etc. Differential Revision: https://reviews.llvm.org/D38318 llvm-svn: 321553
2017-12-29 22:41:50 +08:00
; SSE-NEXT: pshuflw {{.*#+}} xmm0 = mem[0,0,2,3,4,5,6,7]
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_32i8_16i8_01010101010101010101010101010101:
[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][SSE] Match PSHUFLW/PSHUFHW + PSHUFD vXi16 shuffle patterns (PR34686) As noted in PR34686, we are relying on a PSHUFD+PSHUFLW+PSHUFHW shuffle chain for most general vXi16 unary shuffles. This patch checks for simpler PSHUFLW+PSHUFD and PSHUFHW+PSHUFD cases beforehand, building on some existing code that just handled splat shuffles. By doing so we also prevent premature use of PSHUFB shuffles which can be slower and require the creation/loading of constant shuffle masks. We now have the 'fast-variable-shuffle' option for hardware that prefers combining 2 or more shuffles to VPSHUFB etc. Differential Revision: https://reviews.llvm.org/D38318 llvm-svn: 321553
2017-12-29 22:41:50 +08:00
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = mem[0,0,2,3,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_32i8_16i8_01010101010101010101010101010101:
[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] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX2-NEXT: vpbroadcastw (%rdi), %ymm0
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_32i8_16i8_01010101010101010101010101010101:
[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
; AVX512: # %bb.0: # %entry
[X86] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX512-NEXT: vpbroadcastw (%rdi), %ymm0
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: retq
entry:
%ld = load <16 x i8>, <16 x i8>* %ptr
%ret = shufflevector <16 x i8> %ld, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <32 x i8> %ret
}
define <32 x i8> @load_splat_32i8_16i8_01230123012301230123012301230123(<16 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_32i8_16i8_01230123012301230123012301230123:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_32i8_16i8_01230123012301230123012301230123:
[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] Teach execution domain fixing to convert between VPERMILPS and VPSHUFD. llvm-svn: 313507
2017-09-18 11:29:47 +08:00
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_32i8_16i8_01230123012301230123012301230123:
[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] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX2-NEXT: vbroadcastss (%rdi), %ymm0
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_32i8_16i8_01230123012301230123012301230123:
[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
; AVX512: # %bb.0: # %entry
[X86] Don't create VBROADCAST nodes with 256-bit or 512-bit input types Summary: We don't seem to have great rules on what a valid VBROADCAST node looks like. And as a consequence we end up with a lot of patterns to try to catch everything. We have patterns with scalar inputs, 128-bit vector inputs, 256-bit vector inputs, and 512-bit vector inputs. As you can see from the things improved here we are currently missing patterns for 128-bit loads being extended to 256-bit before the vbroadcast. I'd like to propose that VBROADCAST should always take a 128-bit vector type as input. As a first step towards that this patch adds an EXTRACT_SUBVECTOR in front of VBROADCAST when the input is 256 or 512-bits. In the future I would like to add scalar_to_vector around all the scalar operations. And maybe we should consider adding a VBROADCAST+load node to avoid separating loads from the broadcasting operation when the load itself isn't foldable. This requires an additional change in target shuffle combining to look for the extract subvector and look through it to find the original operand. I'm sure this change isn't perfect but was enough to fix a few test failures that were being caused. Another interesting thing I noticed is that the changes in masked_gather_scatter.ll show cases were we don't remove a useless insert into element 1 before broadcasting element 0. Reviewers: delena, RKSimon, zvi Reviewed By: zvi Subscribers: igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D28747 llvm-svn: 295155
2017-02-15 14:58:47 +08:00
; AVX512-NEXT: vbroadcastss (%rdi), %ymm0
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: retq
entry:
%ld = load <16 x i8>, <16 x i8>* %ptr
%ret = shufflevector <16 x i8> %ld, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <32 x i8> %ret
}
define <32 x i8> @load_splat_32i8_16i8_01234567012345670123456701234567(<16 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_32i8_16i8_01234567012345670123456701234567:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
[X86][AVX] lowerVectorShuffleAsBroadcast - aggressively peek through BITCASTs Assuming we can safely adjust the broadcast index for the new type to keep it suitably aligned, then peek through BITCASTs when looking for the broadcast source. Fixes PR32007 llvm-svn: 320933
2017-12-17 07:32:18 +08:00
; AVX-LABEL: load_splat_32i8_16i8_01234567012345670123456701234567:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX-NEXT: retq
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
entry:
%ld = load <16 x i8>, <16 x i8>* %ptr
%ret = shufflevector <16 x i8> %ld, <16 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
ret <32 x i8> %ret
}
define <32 x i8> @load_splat_32i8_32i8_01010101010101010101010101010101(<32 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_32i8_32i8_01010101010101010101010101010101:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Match PSHUFLW/PSHUFHW + PSHUFD vXi16 shuffle patterns (PR34686) As noted in PR34686, we are relying on a PSHUFD+PSHUFLW+PSHUFHW shuffle chain for most general vXi16 unary shuffles. This patch checks for simpler PSHUFLW+PSHUFD and PSHUFHW+PSHUFD cases beforehand, building on some existing code that just handled splat shuffles. By doing so we also prevent premature use of PSHUFB shuffles which can be slower and require the creation/loading of constant shuffle masks. We now have the 'fast-variable-shuffle' option for hardware that prefers combining 2 or more shuffles to VPSHUFB etc. Differential Revision: https://reviews.llvm.org/D38318 llvm-svn: 321553
2017-12-29 22:41:50 +08:00
; SSE-NEXT: pshuflw {{.*#+}} xmm0 = mem[0,0,2,3,4,5,6,7]
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
; AVX1-LABEL: load_splat_32i8_32i8_01010101010101010101010101010101:
[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][SSE] Match PSHUFLW/PSHUFHW + PSHUFD vXi16 shuffle patterns (PR34686) As noted in PR34686, we are relying on a PSHUFD+PSHUFLW+PSHUFHW shuffle chain for most general vXi16 unary shuffles. This patch checks for simpler PSHUFLW+PSHUFD and PSHUFHW+PSHUFD cases beforehand, building on some existing code that just handled splat shuffles. By doing so we also prevent premature use of PSHUFB shuffles which can be slower and require the creation/loading of constant shuffle masks. We now have the 'fast-variable-shuffle' option for hardware that prefers combining 2 or more shuffles to VPSHUFB etc. Differential Revision: https://reviews.llvm.org/D38318 llvm-svn: 321553
2017-12-29 22:41:50 +08:00
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = mem[0,0,2,3,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: load_splat_32i8_32i8_01010101010101010101010101010101:
[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][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX2-NEXT: vpbroadcastw (%rdi), %ymm0
; AVX2-NEXT: retq
;
; AVX512-LABEL: load_splat_32i8_32i8_01010101010101010101010101010101:
[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
; AVX512: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX512-NEXT: vpbroadcastw (%rdi), %ymm0
; AVX512-NEXT: retq
entry:
%ld = load <32 x i8>, <32 x i8>* %ptr
%ret = shufflevector <32 x i8> %ld, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <32 x i8> %ret
}
define <32 x i8> @load_splat_32i8_32i8_01230123012301230123012301230123(<32 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_32i8_32i8_01230123012301230123012301230123:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
[X86][AVX] Fix failed broadcast fold Strip excess BITCASTs from EXTRACT_SUBVECTOR input llvm-svn: 320930
2017-12-17 06:57:17 +08:00
; AVX-LABEL: load_splat_32i8_32i8_01230123012301230123012301230123:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vbroadcastss (%rdi), %ymm0
; AVX-NEXT: retq
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
entry:
%ld = load <32 x i8>, <32 x i8>* %ptr
%ret = shufflevector <32 x i8> %ld, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <32 x i8> %ret
}
define <32 x i8> @load_splat_32i8_32i8_01234567012345670123456701234567(<32 x i8>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_32i8_32i8_01234567012345670123456701234567:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm0 = mem[0,1,0,1]
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: retq
;
; AVX-LABEL: load_splat_32i8_32i8_01234567012345670123456701234567:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0: # %entry
[X86][SSE] Added more widened broadcast tests Added more vXi16 and vXi8 tests llvm-svn: 276534
2016-07-24 05:15:31 +08:00
; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
; AVX-NEXT: retq
entry:
%ld = load <32 x i8>, <32 x i8>* %ptr
%ret = shufflevector <32 x i8> %ld, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
ret <32 x i8> %ret
}
[X86][SSE] Add tests for broadcast from larger vector loads llvm-svn: 301583
2017-04-28 04:19:00 +08:00
define <4 x float> @load_splat_4f32_8f32_0000(<8 x float>* %ptr) nounwind uwtable readnone ssp {
; SSE-LABEL: load_splat_4f32_8f32_0000:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0: # %entry
[X86][SSE] Add tests for broadcast from larger vector loads llvm-svn: 301583
2017-04-28 04:19:00 +08:00
; SSE-NEXT: movaps (%rdi), %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0,0,0]
; SSE-NEXT: retq
;
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX-LABEL: load_splat_4f32_8f32_0000:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0: # %entry
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX-NEXT: vbroadcastss (%rdi), %xmm0
; AVX-NEXT: retq
[X86][SSE] Add tests for broadcast from larger vector loads llvm-svn: 301583
2017-04-28 04:19:00 +08:00
entry:
%ld = load <8 x float>, <8 x float>* %ptr
%ret = shufflevector <8 x float> %ld, <8 x float> undef, <4 x i32> zeroinitializer
ret <4 x float> %ret
}
define <8 x float> @load_splat_8f32_16f32_89898989(<16 x float>* %ptr) nounwind uwtable readnone ssp {
; SSE2-LABEL: load_splat_8f32_16f32_89898989:
[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
; SSE2: # %bb.0: # %entry
[X86][SSE] Add tests for broadcast from larger vector loads llvm-svn: 301583
2017-04-28 04:19:00 +08:00
; SSE2-NEXT: movaps 32(%rdi), %xmm0
; SSE2-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0,0]
; SSE2-NEXT: movaps %xmm0, %xmm1
; SSE2-NEXT: retq
;
; SSE42-LABEL: load_splat_8f32_16f32_89898989:
[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
; SSE42: # %bb.0: # %entry
[X86][SSE] Add tests for broadcast from larger vector loads llvm-svn: 301583
2017-04-28 04:19:00 +08:00
; SSE42-NEXT: movddup {{.*#+}} xmm0 = mem[0,0]
; SSE42-NEXT: movapd %xmm0, %xmm1
; SSE42-NEXT: retq
;
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX-LABEL: load_splat_8f32_16f32_89898989:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0: # %entry
[DAGCombiner] use narrow load to avoid vector extract If we have (extract_subvector(load wide vector)) with no other users, that can just be (load narrow vector). This is intentionally conservative. Follow-ups may loosen the one-use constraint to account for the extract cost or just remove the one-use check. The memop chain updating is based on code that already exists multiple times in x86 lowering, so that should be pulled into a helper function as a follow-up. Background: this is a potential improvement noticed via regressions caused by making x86's peekThroughBitcasts() not loop on consecutive bitcasts (see comments in D33137). Differential Revision: https://reviews.llvm.org/D33578 llvm-svn: 304072
2017-05-27 22:07:03 +08:00
; AVX-NEXT: vbroadcastsd 32(%rdi), %ymm0
; AVX-NEXT: retq
[X86][SSE] Add tests for broadcast from larger vector loads llvm-svn: 301583
2017-04-28 04:19:00 +08:00
entry:
%ld = load <16 x float>, <16 x float>* %ptr
%ret = shufflevector <16 x float> %ld, <16 x float> undef, <8 x i32> <i32 8, i32 9, i32 8, i32 9, i32 8, i32 9, i32 8, i32 9>
ret <8 x float> %ret
}