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/combine-srl.ll

452 lines
16 KiB
LLVM
Raw Normal View History

[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
[X86] Add common CHECK prefix to shift combine tests llvm-svn: 324638
2018-02-09 03:23:47 +08:00
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=CHECK,SSE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2-SLOW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fast-variable-shuffle | FileCheck %s --check-prefixes=CHECK,AVX,AVX2-FAST
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; fold (srl 0, x) -> 0
define <4 x i32> @combine_vec_lshr_zero(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_zero:
[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:
[DAGCombiner] Add vector support to fold (shl/srl 0, x) -> 0 llvm-svn: 302641
2017-05-10 20:34:27 +08:00
; SSE-NEXT: xorps %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_zero:
[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:
[DAGCombiner] Add vector support to fold (shl/srl 0, x) -> 0 llvm-svn: 302641
2017-05-10 20:34:27 +08:00
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: retq
%1 = lshr <4 x i32> zeroinitializer, %x
ret <4 x i32> %1
}
; fold (srl x, c >= size(x)) -> undef
define <4 x i32> @combine_vec_lshr_outofrange0(<4 x i32> %x) {
[X86] Add common CHECK prefix to shift combine tests llvm-svn: 324638
2018-02-09 03:23:47 +08:00
; CHECK-LABEL: combine_vec_lshr_outofrange0:
; CHECK: # %bb.0:
; CHECK-NEXT: retq
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
%1 = lshr <4 x i32> %x, <i32 33, i32 33, i32 33, i32 33>
ret <4 x i32> %1
}
define <4 x i32> @combine_vec_lshr_outofrange1(<4 x i32> %x) {
[X86] Add common CHECK prefix to shift combine tests llvm-svn: 324638
2018-02-09 03:23:47 +08:00
; CHECK-LABEL: combine_vec_lshr_outofrange1:
; CHECK: # %bb.0:
; CHECK-NEXT: retq
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
%1 = lshr <4 x i32> %x, <i32 33, i32 34, i32 35, i32 36>
ret <4 x i32> %1
}
; fold (srl x, 0) -> x
define <4 x i32> @combine_vec_lshr_by_zero(<4 x i32> %x) {
[X86] Add common CHECK prefix to shift combine tests llvm-svn: 324638
2018-02-09 03:23:47 +08:00
; CHECK-LABEL: combine_vec_lshr_by_zero:
; CHECK: # %bb.0:
; CHECK-NEXT: retq
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
%1 = lshr <4 x i32> %x, zeroinitializer
ret <4 x i32> %1
}
; if (srl x, c) is known to be zero, return 0
define <4 x i32> @combine_vec_lshr_known_zero0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_known_zero0:
[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:
[DAG] enhance computeKnownBits to handle SRL/SRA with vector splat constant llvm-svn: 284953
2016-10-24 07:13:31 +08:00
; SSE-NEXT: xorps %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_known_zero0:
[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:
[DAG] enhance computeKnownBits to handle SRL/SRA with vector splat constant llvm-svn: 284953
2016-10-24 07:13:31 +08:00
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: retq
%1 = and <4 x i32> %x, <i32 15, i32 15, i32 15, i32 15>
%2 = lshr <4 x i32> %1, <i32 4, i32 4, i32 4, i32 4>
ret <4 x i32> %2
}
define <4 x i32> @combine_vec_lshr_known_zero1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_known_zero1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86] Add vector shift by immediate to SimplifyDemandedBitsForTargetNode. Summary: This also enables some constant folding from KnownBits propagation. This helps on some cases vXi64 case in 32-bit mode where constant vectors appear as vXi32 and a bitcast. This can prevent getNode from constant folding sra/shl/srl. Reviewers: RKSimon, spatel Reviewed By: spatel Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D54069 llvm-svn: 346102
2018-11-05 01:31:27 +08:00
; SSE-NEXT: xorps %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_known_zero1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][AVX] Regenerate combine tests with constant broadcast comments llvm-svn: 308128
2017-07-16 19:36:11 +08:00
; AVX-NEXT: vpbroadcastd {{.*#+}} xmm1 = [15,15,15,15]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX-NEXT: vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%1 = and <4 x i32> %x, <i32 15, i32 15, i32 15, i32 15>
%2 = lshr <4 x i32> %1, <i32 8, i32 9, i32 10, i32 11>
ret <4 x i32> %2
}
; fold (srl (srl x, c1), c2) -> (srl x, (add c1, c2))
define <4 x i32> @combine_vec_lshr_lshr0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_lshr0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrld $6, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_lshr0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrld $6, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 2, i32 2, i32 2, i32 2>
%2 = lshr <4 x i32> %1, <i32 4, i32 4, i32 4, i32 4>
ret <4 x i32> %2
}
define <4 x i32> @combine_vec_lshr_lshr1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_lshr1:
[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:
[DAGCombiner] Match ISD::SRL non-uniform constant vectors patterns using predicates. Use predicate matchers introduced in D35492 to match more ISD::SRL constant folds llvm-svn: 308602
2017-07-20 19:03:30 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: psrld $10, %xmm1
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm2
[DAGCombiner] Match ISD::SRL non-uniform constant vectors patterns using predicates. Use predicate matchers introduced in D35492 to match more ISD::SRL constant folds llvm-svn: 308602
2017-07-20 19:03:30 +08:00
; SSE-NEXT: psrld $6, %xmm2
; SSE-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3],xmm1[4,5,6,7]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm1
[DAGCombiner] Match ISD::SRL non-uniform constant vectors patterns using predicates. Use predicate matchers introduced in D35492 to match more ISD::SRL constant folds llvm-svn: 308602
2017-07-20 19:03:30 +08:00
; SSE-NEXT: psrld $8, %xmm1
; SSE-NEXT: psrld $4, %xmm0
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_lshr1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 0, i32 1, i32 2, i32 3>
%2 = lshr <4 x i32> %1, <i32 4, i32 5, i32 6, i32 7>
ret <4 x i32> %2
}
; fold (srl (srl x, c1), c2) -> 0
define <4 x i32> @combine_vec_lshr_lshr_zero0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_lshr_zero0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: xorps %xmm0, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_lshr_zero0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 16, i32 16, i32 16, i32 16>
%2 = lshr <4 x i32> %1, <i32 20, i32 20, i32 20, i32 20>
ret <4 x i32> %2
}
define <4 x i32> @combine_vec_lshr_lshr_zero1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_lshr_zero1:
[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:
[DAGCombiner] Match ISD::SRL non-uniform constant vectors patterns using predicates. Use predicate matchers introduced in D35492 to match more ISD::SRL constant folds llvm-svn: 308602
2017-07-20 19:03:30 +08:00
; SSE-NEXT: xorps %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_lshr_zero1:
[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:
[DAGCombiner] Match ISD::SRL non-uniform constant vectors patterns using predicates. Use predicate matchers introduced in D35492 to match more ISD::SRL constant folds llvm-svn: 308602
2017-07-20 19:03:30 +08:00
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 17, i32 18, i32 19, i32 20>
%2 = lshr <4 x i32> %1, <i32 25, i32 26, i32 27, i32 28>
ret <4 x i32> %2
}
; fold (srl (trunc (srl x, c1)), c2) -> (trunc (srl x, (add c1, c2)))
define <4 x i32> @combine_vec_lshr_trunc_lshr0(<4 x i64> %x) {
; SSE-LABEL: combine_vec_lshr_trunc_lshr0:
[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:
[DAGCombiner] Add vector support for (srl (trunc (srl x, c1)), c2) combine. llvm-svn: 301305
2017-04-25 20:40:45 +08:00
; SSE-NEXT: psrlq $48, %xmm1
; SSE-NEXT: psrlq $48, %xmm0
[X86][SSE] Add PACKUS support to combineVectorTruncation Similar to the existing code to lower to PACKSS, we can use PACKUS if the input vector's leading zero bits extend all the way to the packed/truncated value. We have to account for pre-SSE41 targets not supporting PACKUSDW llvm-svn: 317315
2017-11-03 19:33:48 +08:00
; SSE-NEXT: packusdw %xmm1, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_trunc_lshr0:
[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:
[DAGCombiner] Add vector support for (srl (trunc (srl x, c1)), c2) combine. llvm-svn: 301305
2017-04-25 20:40:45 +08:00
; AVX-NEXT: vpsrlq $48, %ymm0, %ymm0
[X86][SSE] Add PACKUS support to LowerTruncate Similar to the existing code to lower to PACKSS, we can use PACKUS if the input vector's leading zero bits extend all the way to the packed/truncated value. We have to account for pre-SSE41 targets not supporting PACKUSDW llvm-svn: 317128
2017-11-02 05:52:29 +08:00
; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
; AVX-NEXT: vpackusdw %xmm1, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
%1 = lshr <4 x i64> %x, <i64 32, i64 32, i64 32, i64 32>
%2 = trunc <4 x i64> %1 to <4 x i32>
%3 = lshr <4 x i32> %2, <i32 16, i32 16, i32 16, i32 16>
ret <4 x i32> %3
}
define <4 x i32> @combine_vec_lshr_trunc_lshr1(<4 x i64> %x) {
; SSE-LABEL: combine_vec_lshr_trunc_lshr1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: movdqa %xmm1, %xmm2
; SSE-NEXT: psrlq $35, %xmm2
; SSE-NEXT: psrlq $34, %xmm1
; SSE-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm2[4,5,6,7]
; SSE-NEXT: movdqa %xmm0, %xmm2
[x86] use a single shufps when it can save instructions This is a tiny patch with a big pile of test changes. This partially fixes PR27885: https://llvm.org/bugs/show_bug.cgi?id=27885 My motivating case looks like this: - vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,2] - vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] - vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7] + vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] And this happens several times in the diffs. For chips with domain-crossing penalties, the instruction count and size reduction should usually overcome any potential domain-crossing penalty due to using an FP op in a sequence of int ops. For chips such as recent Intel big cores and Atom, there is no domain-crossing penalty for shufps, so using shufps is a pure win. So the test case diffs all appear to be improvements except one test in vector-shuffle-combining.ll where we miss an opportunity to use a shift to generate zero elements and one test in combine-sra.ll where multiple uses prevent the expected shuffle combining. Differential Revision: https://reviews.llvm.org/D27692 llvm-svn: 289837
2016-12-16 02:03:38 +08:00
; SSE-NEXT: psrlq $33, %xmm2
; SSE-NEXT: psrlq $32, %xmm0
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
; SSE-NEXT: movaps %xmm0, %xmm1
; SSE-NEXT: psrld $19, %xmm1
; SSE-NEXT: movaps %xmm0, %xmm2
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrld $17, %xmm2
; SSE-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3],xmm1[4,5,6,7]
[x86] use a single shufps when it can save instructions This is a tiny patch with a big pile of test changes. This partially fixes PR27885: https://llvm.org/bugs/show_bug.cgi?id=27885 My motivating case looks like this: - vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,2] - vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] - vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7] + vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] And this happens several times in the diffs. For chips with domain-crossing penalties, the instruction count and size reduction should usually overcome any potential domain-crossing penalty due to using an FP op in a sequence of int ops. For chips such as recent Intel big cores and Atom, there is no domain-crossing penalty for shufps, so using shufps is a pure win. So the test case diffs all appear to be improvements except one test in vector-shuffle-combining.ll where we miss an opportunity to use a shift to generate zero elements and one test in combine-sra.ll where multiple uses prevent the expected shuffle combining. Differential Revision: https://reviews.llvm.org/D27692 llvm-svn: 289837
2016-12-16 02:03:38 +08:00
; SSE-NEXT: movaps %xmm0, %xmm1
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrld $18, %xmm1
; SSE-NEXT: psrld $16, %xmm0
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
; SSE-NEXT: retq
;
X86 Tests: Update more isel tests with FastVariableShuffle feature Summary: Added the FastVariableShuffle feature to cases that resembled processors for which this fearure is on. For AVX2 there are processors with and w/o this fearue enable. For AVX512 only KNL does enable this feature so cases which only have +avx512f were left without the FastVariableShuffle enabled. Reviewers: RKSimon, craig.topper Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D41851 llvm-svn: 322090
2018-01-10 00:26:06 +08:00
; AVX2-SLOW-LABEL: combine_vec_lshr_trunc_lshr1:
; AVX2-SLOW: # %bb.0:
; AVX2-SLOW-NEXT: vpsrlvq {{.*}}(%rip), %ymm0, %ymm0
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-SLOW-NEXT: vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
; AVX2-SLOW-NEXT: vzeroupper
; AVX2-SLOW-NEXT: retq
;
; AVX2-FAST-LABEL: combine_vec_lshr_trunc_lshr1:
; AVX2-FAST: # %bb.0:
; AVX2-FAST-NEXT: vpsrlvq {{.*}}(%rip), %ymm0, %ymm0
; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm1 = [0,2,4,6,4,6,6,7]
; AVX2-FAST-NEXT: vpermd %ymm0, %ymm1, %ymm0
; AVX2-FAST-NEXT: vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
; AVX2-FAST-NEXT: vzeroupper
; AVX2-FAST-NEXT: retq
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
%1 = lshr <4 x i64> %x, <i64 32, i64 33, i64 34, i64 35>
%2 = trunc <4 x i64> %1 to <4 x i32>
%3 = lshr <4 x i32> %2, <i32 16, i32 17, i32 18, i32 19>
ret <4 x i32> %3
}
; fold (srl (trunc (srl x, c1)), c2) -> 0
define <4 x i32> @combine_vec_lshr_trunc_lshr_zero0(<4 x i64> %x) {
; SSE-LABEL: combine_vec_lshr_trunc_lshr_zero0:
[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:
[DAG] enhance computeKnownBits to handle SRL/SRA with vector splat constant llvm-svn: 284953
2016-10-24 07:13:31 +08:00
; SSE-NEXT: xorps %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_trunc_lshr_zero0:
[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:
[DAG] enhance computeKnownBits to handle SRL/SRA with vector splat constant llvm-svn: 284953
2016-10-24 07:13:31 +08:00
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: retq
%1 = lshr <4 x i64> %x, <i64 48, i64 48, i64 48, i64 48>
%2 = trunc <4 x i64> %1 to <4 x i32>
%3 = lshr <4 x i32> %2, <i32 24, i32 24, i32 24, i32 24>
ret <4 x i32> %3
}
define <4 x i32> @combine_vec_lshr_trunc_lshr_zero1(<4 x i64> %x) {
; SSE-LABEL: combine_vec_lshr_trunc_lshr_zero1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: movdqa %xmm1, %xmm2
; SSE-NEXT: psrlq $51, %xmm2
; SSE-NEXT: psrlq $50, %xmm1
; SSE-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm2[4,5,6,7]
; SSE-NEXT: movdqa %xmm0, %xmm2
[x86] use a single shufps when it can save instructions This is a tiny patch with a big pile of test changes. This partially fixes PR27885: https://llvm.org/bugs/show_bug.cgi?id=27885 My motivating case looks like this: - vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,2] - vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] - vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7] + vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] And this happens several times in the diffs. For chips with domain-crossing penalties, the instruction count and size reduction should usually overcome any potential domain-crossing penalty due to using an FP op in a sequence of int ops. For chips such as recent Intel big cores and Atom, there is no domain-crossing penalty for shufps, so using shufps is a pure win. So the test case diffs all appear to be improvements except one test in vector-shuffle-combining.ll where we miss an opportunity to use a shift to generate zero elements and one test in combine-sra.ll where multiple uses prevent the expected shuffle combining. Differential Revision: https://reviews.llvm.org/D27692 llvm-svn: 289837
2016-12-16 02:03:38 +08:00
; SSE-NEXT: psrlq $49, %xmm2
; SSE-NEXT: psrlq $48, %xmm0
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
[SelectionDAG] Teach computeKnownBits some improvements to ISD::SRL with a non-splat constant shift amount. If we have a non-splat constant shift amount, the minimum shift amount can be used to infer the number of zero upper bits of the result. There's probably a lot more that we can do here, but this fixes a case where I wanted to infer the sign bit as zero when all the shift amounts are non-zero. llvm-svn: 319639
2017-12-04 13:38:42 +08:00
; SSE-NEXT: packusdw %xmm1, %xmm0
; SSE-NEXT: movdqa %xmm0, %xmm1
[x86] use a single shufps when it can save instructions This is a tiny patch with a big pile of test changes. This partially fixes PR27885: https://llvm.org/bugs/show_bug.cgi?id=27885 My motivating case looks like this: - vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,2] - vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] - vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7] + vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] And this happens several times in the diffs. For chips with domain-crossing penalties, the instruction count and size reduction should usually overcome any potential domain-crossing penalty due to using an FP op in a sequence of int ops. For chips such as recent Intel big cores and Atom, there is no domain-crossing penalty for shufps, so using shufps is a pure win. So the test case diffs all appear to be improvements except one test in vector-shuffle-combining.ll where we miss an opportunity to use a shift to generate zero elements and one test in combine-sra.ll where multiple uses prevent the expected shuffle combining. Differential Revision: https://reviews.llvm.org/D27692 llvm-svn: 289837
2016-12-16 02:03:38 +08:00
; SSE-NEXT: psrld $27, %xmm1
[SelectionDAG] Teach computeKnownBits some improvements to ISD::SRL with a non-splat constant shift amount. If we have a non-splat constant shift amount, the minimum shift amount can be used to infer the number of zero upper bits of the result. There's probably a lot more that we can do here, but this fixes a case where I wanted to infer the sign bit as zero when all the shift amounts are non-zero. llvm-svn: 319639
2017-12-04 13:38:42 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm2
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrld $25, %xmm2
; SSE-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3],xmm1[4,5,6,7]
[SelectionDAG] Teach computeKnownBits some improvements to ISD::SRL with a non-splat constant shift amount. If we have a non-splat constant shift amount, the minimum shift amount can be used to infer the number of zero upper bits of the result. There's probably a lot more that we can do here, but this fixes a case where I wanted to infer the sign bit as zero when all the shift amounts are non-zero. llvm-svn: 319639
2017-12-04 13:38:42 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm1
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrld $26, %xmm1
; SSE-NEXT: psrld $24, %xmm0
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_trunc_lshr_zero1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrlvq {{.*}}(%rip), %ymm0, %ymm0
[SelectionDAG] Teach computeKnownBits some improvements to ISD::SRL with a non-splat constant shift amount. If we have a non-splat constant shift amount, the minimum shift amount can be used to infer the number of zero upper bits of the result. There's probably a lot more that we can do here, but this fixes a case where I wanted to infer the sign bit as zero when all the shift amounts are non-zero. llvm-svn: 319639
2017-12-04 13:38:42 +08:00
; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
; AVX-NEXT: vpackusdw %xmm1, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrlvd {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
%1 = lshr <4 x i64> %x, <i64 48, i64 49, i64 50, i64 51>
%2 = trunc <4 x i64> %1 to <4 x i32>
%3 = lshr <4 x i32> %2, <i32 24, i32 25, i32 26, i32 27>
ret <4 x i32> %3
}
; fold (srl (shl x, c), c) -> (and x, cst2)
define <4 x i32> @combine_vec_lshr_shl_mask0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_shl_mask0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: andps {{.*}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_shl_mask0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][AVX] Regenerate combine tests with constant broadcast comments llvm-svn: 308128
2017-07-16 19:36:11 +08:00
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [1073741823,1073741823,1073741823,1073741823]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vandps %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = shl <4 x i32> %x, <i32 2, i32 2, i32 2, i32 2>
%2 = lshr <4 x i32> %1, <i32 2, i32 2, i32 2, i32 2>
ret <4 x i32> %2
}
define <4 x i32> @combine_vec_lshr_shl_mask1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_shl_mask1:
[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:
[DAGCombiner] Add general constant vector support to (srl (shl x, c), c) -> (and x, cst2) We already supported scalar constant / splatted constant vector - now accepts any (non opaque) constant scalar / vector llvm-svn: 284717
2016-10-20 19:10:21 +08:00
; SSE-NEXT: andps {{.*}}(%rip), %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_shl_mask1:
[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:
[DAGCombiner] Add general constant vector support to (srl (shl x, c), c) -> (and x, cst2) We already supported scalar constant / splatted constant vector - now accepts any (non opaque) constant scalar / vector llvm-svn: 284717
2016-10-20 19:10:21 +08:00
; AVX-NEXT: vandps {{.*}}(%rip), %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: retq
%1 = shl <4 x i32> %x, <i32 2, i32 3, i32 4, i32 5>
%2 = lshr <4 x i32> %1, <i32 2, i32 3, i32 4, i32 5>
ret <4 x i32> %2
}
; fold (srl (sra X, Y), 31) -> (srl X, 31)
define <4 x i32> @combine_vec_lshr_ashr_sign(<4 x i32> %x, <4 x i32> %y) {
; SSE-LABEL: combine_vec_lshr_ashr_sign:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrld $31, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_ashr_sign:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrld $31, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = ashr <4 x i32> %x, %y
%2 = lshr <4 x i32> %1, <i32 31, i32 31, i32 31, i32 31>
ret <4 x i32> %2
}
; fold (srl (ctlz x), "5") -> x iff x has one bit set (the low bit).
define <4 x i32> @combine_vec_lshr_lzcnt_bit0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_lzcnt_bit0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: pand {{.*}}(%rip), %xmm0
; SSE-NEXT: psrld $4, %xmm0
; SSE-NEXT: pxor {{.*}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_lzcnt_bit0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][AVX] Regenerate combine tests with constant broadcast comments llvm-svn: 308128
2017-07-16 19:36:11 +08:00
; AVX-NEXT: vpbroadcastd {{.*#+}} xmm1 = [16,16,16,16]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX-NEXT: vpsrld $4, %xmm0, %xmm0
[X86][AVX] Regenerate combine tests with constant broadcast comments llvm-svn: 308128
2017-07-16 19:36:11 +08:00
; AVX-NEXT: vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpxor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = and <4 x i32> %x, <i32 16, i32 16, i32 16, i32 16>
%2 = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %1, i1 0)
%3 = lshr <4 x i32> %2, <i32 5, i32 5, i32 5, i32 5>
ret <4 x i32> %3
}
define <4 x i32> @combine_vec_lshr_lzcnt_bit1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_lshr_lzcnt_bit1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: pand {{.*}}(%rip), %xmm0
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: psrlw $4, %xmm1
; SSE-NEXT: pxor %xmm2, %xmm2
[X86] Move promotion of vector and/or/xor from legalization to DAG combine Summary: I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that. This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better. In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53107 llvm-svn: 344487
2018-10-15 09:51:58 +08:00
; SSE-NEXT: movdqa {{.*#+}} xmm3 = [4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0]
; SSE-NEXT: movdqa %xmm3, %xmm4
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: pshufb %xmm1, %xmm3
; SSE-NEXT: pcmpeqb %xmm2, %xmm1
[X86] Move promotion of vector and/or/xor from legalization to DAG combine Summary: I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that. This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better. In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53107 llvm-svn: 344487
2018-10-15 09:51:58 +08:00
; SSE-NEXT: movdqa {{.*#+}} xmm5 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
; SSE-NEXT: pand %xmm0, %xmm5
; SSE-NEXT: pshufb %xmm5, %xmm4
; SSE-NEXT: pand %xmm1, %xmm4
; SSE-NEXT: paddb %xmm4, %xmm3
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: pcmpeqb %xmm2, %xmm1
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: psrlw $8, %xmm1
[X86] Move promotion of vector and/or/xor from legalization to DAG combine Summary: I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that. This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better. In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53107 llvm-svn: 344487
2018-10-15 09:51:58 +08:00
; SSE-NEXT: pand %xmm3, %xmm1
; SSE-NEXT: psrlw $8, %xmm3
; SSE-NEXT: paddw %xmm1, %xmm3
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: pcmpeqw %xmm2, %xmm0
; SSE-NEXT: psrld $16, %xmm0
[X86] Move promotion of vector and/or/xor from legalization to DAG combine Summary: I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that. This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better. In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53107 llvm-svn: 344487
2018-10-15 09:51:58 +08:00
; SSE-NEXT: pand %xmm3, %xmm0
; SSE-NEXT: psrld $16, %xmm3
; SSE-NEXT: paddd %xmm3, %xmm0
; SSE-NEXT: psrld $5, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_lshr_lzcnt_bit1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
[X86] Move promotion of vector and/or/xor from legalization to DAG combine Summary: I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that. This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better. In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53107 llvm-svn: 344487
2018-10-15 09:51:58 +08:00
; AVX-NEXT: vpsrlw $4, %xmm0, %xmm1
; AVX-NEXT: vpxor %xmm2, %xmm2, %xmm2
; AVX-NEXT: vpcmpeqb %xmm2, %xmm1, %xmm3
; AVX-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm4
; AVX-NEXT: vmovdqa {{.*#+}} xmm5 = [4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0]
; AVX-NEXT: vpshufb %xmm4, %xmm5, %xmm4
; AVX-NEXT: vpand %xmm3, %xmm4, %xmm3
; AVX-NEXT: vpshufb %xmm1, %xmm5, %xmm1
; AVX-NEXT: vpaddb %xmm1, %xmm3, %xmm1
; AVX-NEXT: vpcmpeqb %xmm2, %xmm0, %xmm3
; AVX-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX-NEXT: vpand %xmm3, %xmm1, %xmm3
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrlw $8, %xmm1, %xmm1
[X86] Move promotion of vector and/or/xor from legalization to DAG combine Summary: I've noticed that the bitcasts we introduce for these make computeKnownBits and computeNumSignBits not work well in LegalizeVectorOps. LegalizeVectorOps legalizes bottom up while LegalizeDAG legalizes top down. The bottom up strategy for LegalizeVectorOps means operands are legalized before their uses. So we promote and/or/xor before we legalize the operands that use them making computeKnownBits/computeNumSignBits in places like LowerTruncate suboptimal. I looked at changing LegalizeVectorOps to be top down as well, but that was more disruptive and caused some regressions. I also looked at just moving promotion of binops to LegalizeDAG, but that had a few issues one around matching AND,ANDN,OR into VSELECT because I had to create ANDN as vXi64, but the other nodes hadn't legalized yet, I didn't look too hard at fixing that. This patch seems to produce better results overall than my other attempts. We now form broadcasts of constants better in some cases. For at least some of them the AND was being introduced in LegalizeDAG, promoted to vXi64, and the BUILD_VECTOR was also legalized there. I think we got bad ordering of that. Now the promotion is out of the legalizer so we handle this better. In the longer term I think we really should evaluate whether we should be doing this promotion at all. It's really there to reduce isel pattern count, but I'm wondering if we'd be better served just eating the pattern cost or doing C++ based isel for vector and/or/xor in X86ISelDAGToDAG. The masked and/or/xor will definitely be difficult in patterns if a bitcast gets between the vselect and the and/or/xor node. That becomes a lot of permutations to cover. Reviewers: RKSimon, spatel Reviewed By: RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D53107 llvm-svn: 344487
2018-10-15 09:51:58 +08:00
; AVX-NEXT: vpaddw %xmm3, %xmm1, %xmm1
; AVX-NEXT: vpcmpeqw %xmm2, %xmm0, %xmm0
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; AVX-NEXT: vpsrld $16, %xmm0, %xmm0
; AVX-NEXT: vpand %xmm0, %xmm1, %xmm0
; AVX-NEXT: vpsrld $16, %xmm1, %xmm1
; AVX-NEXT: vpaddd %xmm0, %xmm1, %xmm0
; AVX-NEXT: vpsrld $5, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = and <4 x i32> %x, <i32 4, i32 32, i32 64, i32 128>
%2 = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %1, i1 0)
%3 = lshr <4 x i32> %2, <i32 5, i32 5, i32 5, i32 5>
ret <4 x i32> %3
}
declare <4 x i32> @llvm.ctlz.v4i32(<4 x i32>, i1)
; fold (srl x, (trunc (and y, c))) -> (srl x, (and (trunc y), (trunc c))).
define <4 x i32> @combine_vec_lshr_trunc_and(<4 x i32> %x, <4 x i64> %y) {
; SSE-LABEL: combine_vec_lshr_trunc_and:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[x86] use a single shufps when it can save instructions This is a tiny patch with a big pile of test changes. This partially fixes PR27885: https://llvm.org/bugs/show_bug.cgi?id=27885 My motivating case looks like this: - vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,2] - vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] - vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7] + vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] And this happens several times in the diffs. For chips with domain-crossing penalties, the instruction count and size reduction should usually overcome any potential domain-crossing penalty due to using an FP op in a sequence of int ops. For chips such as recent Intel big cores and Atom, there is no domain-crossing penalty for shufps, so using shufps is a pure win. So the test case diffs all appear to be improvements except one test in vector-shuffle-combining.ll where we miss an opportunity to use a shift to generate zero elements and one test in combine-sra.ll where multiple uses prevent the expected shuffle combining. Differential Revision: https://reviews.llvm.org/D27692 llvm-svn: 289837
2016-12-16 02:03:38 +08:00
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
; SSE-NEXT: andps {{.*}}(%rip), %xmm1
[X86][SSE] Reduce instruction/register usages for v4i32 vector shifts (PR37441) As suggested by Fabian on PR37441, use PSHUFLW to extend shift amount types for use with PSRAD/PSRLD to reduce register pressure. Some of this ideally would be done by combineTargetShuffle but its tricky to do as most of the shuffles are sharing inputs. Differential Revision: https://reviews.llvm.org/D46959 llvm-svn: 332524
2018-05-17 04:52:52 +08:00
; SSE-NEXT: pshuflw {{.*#+}} xmm2 = xmm1[2,3,3,3,4,5,6,7]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: movdqa %xmm0, %xmm3
; SSE-NEXT: psrld %xmm2, %xmm3
[X86][SSE] Reduce instruction/register usages for v4i32 vector shifts (PR37441) As suggested by Fabian on PR37441, use PSHUFLW to extend shift amount types for use with PSRAD/PSRLD to reduce register pressure. Some of this ideally would be done by combineTargetShuffle but its tricky to do as most of the shuffles are sharing inputs. Differential Revision: https://reviews.llvm.org/D46959 llvm-svn: 332524
2018-05-17 04:52:52 +08:00
; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
; SSE-NEXT: pshuflw {{.*#+}} xmm4 = xmm2[2,3,3,3,4,5,6,7]
; SSE-NEXT: movdqa %xmm0, %xmm5
; SSE-NEXT: psrld %xmm4, %xmm5
; SSE-NEXT: pblendw {{.*#+}} xmm5 = xmm3[0,1,2,3],xmm5[4,5,6,7]
; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,1,1,1,4,5,6,7]
; SSE-NEXT: movdqa %xmm0, %xmm3
; SSE-NEXT: psrld %xmm1, %xmm3
; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm2[0,1,1,1,4,5,6,7]
; SSE-NEXT: psrld %xmm1, %xmm0
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm3[0,1,2,3],xmm0[4,5,6,7]
; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm5[2,3],xmm0[4,5],xmm5[6,7]
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
; SSE-NEXT: retq
;
X86 Tests: Update more isel tests with FastVariableShuffle feature Summary: Added the FastVariableShuffle feature to cases that resembled processors for which this fearure is on. For AVX2 there are processors with and w/o this fearue enable. For AVX512 only KNL does enable this feature so cases which only have +avx512f were left without the FastVariableShuffle enabled. Reviewers: RKSimon, craig.topper Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D41851 llvm-svn: 322090
2018-01-10 00:26:06 +08:00
; AVX2-SLOW-LABEL: combine_vec_lshr_trunc_and:
; AVX2-SLOW: # %bb.0:
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-SLOW-NEXT: vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX2-SLOW-NEXT: vpsrlvd %xmm1, %xmm0, %xmm0
; AVX2-SLOW-NEXT: vzeroupper
; AVX2-SLOW-NEXT: retq
;
; AVX2-FAST-LABEL: combine_vec_lshr_trunc_and:
; AVX2-FAST: # %bb.0:
; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm2 = [0,2,4,6,4,6,6,7]
; AVX2-FAST-NEXT: vpermd %ymm1, %ymm2, %ymm1
; AVX2-FAST-NEXT: vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX2-FAST-NEXT: vpsrlvd %xmm1, %xmm0, %xmm0
; AVX2-FAST-NEXT: vzeroupper
; AVX2-FAST-NEXT: retq
[X86][SSE] Added vector lshr/shl combine tests This doesn't cover all combines in DAGCombiner::visitSRL/visitSHL yet, but identifies several cases where we fail to combine vectors (or non-splatted) vectors llvm-svn: 284518
2016-10-19 03:28:12 +08:00
%1 = and <4 x i64> %y, <i64 15, i64 255, i64 4095, i64 65535>
%2 = trunc <4 x i64> %1 to <4 x i32>
%3 = lshr <4 x i32> %x, %2
ret <4 x i32> %3
}