Skip to content

[DAG][X86] Fold mgather/mscatter/etc with splat index #65980

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Sep 13, 2023
Merged

[DAG][X86] Fold mgather/mscatter/etc with splat index #65980

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 15 additions & 2 deletions llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11637,8 +11637,6 @@ static SDValue ConvertSelectToConcatVector(SDNode *N, SelectionDAG &DAG) {

bool refineUniformBase(SDValue &BasePtr, SDValue &Index, bool IndexIsScaled,
SelectionDAG &DAG, const SDLoc &DL) {
if (Index.getOpcode() != ISD::ADD)
return false;

// Only perform the transformation when existing operands can be reused.
if (IndexIsScaled)
Expand All @@ -11648,6 +11646,21 @@ bool refineUniformBase(SDValue &BasePtr, SDValue &Index, bool IndexIsScaled,
return false;

EVT VT = BasePtr.getValueType();

if (SDValue SplatVal = DAG.getSplatValue(Index);
SplatVal && !isNullConstant(SplatVal) &&
SplatVal.getValueType() == VT) {
if (isNullConstant(BasePtr))
BasePtr = SplatVal;
else
BasePtr = DAG.getNode(ISD::ADD, DL, VT, BasePtr, SplatVal);
Index = DAG.getSplat(Index.getValueType(), DL, DAG.getConstant(0, DL, VT));
return true;
}

if (Index.getOpcode() != ISD::ADD)
return false;

if (SDValue SplatVal = DAG.getSplatValue(Index.getOperand(0));
SplatVal && SplatVal.getValueType() == VT) {
if (isNullConstant(BasePtr))
Expand Down
24 changes: 11 additions & 13 deletions llvm/test/CodeGen/X86/masked_gather.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1747,29 +1747,27 @@ define <8 x i32> @gather_v8i32_v8i32(<8 x i32> %trigger) {
; AVX512F-NEXT: vptestnmd %zmm0, %zmm0, %k0
; AVX512F-NEXT: kshiftlw $8, %k0, %k0
; AVX512F-NEXT: kshiftrw $8, %k0, %k1
; AVX512F-NEXT: vpbroadcastd {{.*#+}} zmm0 = [12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12]
; AVX512F-NEXT: vpxor %xmm0, %xmm0, %xmm0
; AVX512F-NEXT: vpxor %xmm1, %xmm1, %xmm1
; AVX512F-NEXT: vpxor %xmm2, %xmm2, %xmm2
; AVX512F-NEXT: kmovw %k1, %k2
; AVX512F-NEXT: vpgatherdd c(,%zmm0), %zmm2 {%k2}
; AVX512F-NEXT: vpbroadcastd {{.*#+}} zmm0 = [28,28,28,28,28,28,28,28,28,28,28,28,28,28,28,28]
; AVX512F-NEXT: vpgatherdd c(,%zmm0), %zmm1 {%k1}
; AVX512F-NEXT: vpaddd %ymm1, %ymm2, %ymm0
; AVX512F-NEXT: vpaddd %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: vpgatherdd c+12(,%zmm0), %zmm1 {%k2}
; AVX512F-NEXT: vpxor %xmm2, %xmm2, %xmm2
; AVX512F-NEXT: vpgatherdd c+28(,%zmm0), %zmm2 {%k1}
; AVX512F-NEXT: vpaddd %ymm2, %ymm1, %ymm0
; AVX512F-NEXT: vpaddd %ymm2, %ymm0, %ymm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: gather_v8i32_v8i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vptestnmd %ymm0, %ymm0, %k1
; AVX512VL-NEXT: vpxor %xmm0, %xmm0, %xmm0
; AVX512VL-NEXT: vpbroadcastd {{.*#+}} ymm1 = [12,12,12,12,12,12,12,12]
; AVX512VL-NEXT: kmovw %k1, %k2
; AVX512VL-NEXT: vpxor %xmm1, %xmm1, %xmm1
; AVX512VL-NEXT: vpgatherdd c+12(,%ymm0), %ymm1 {%k2}
; AVX512VL-NEXT: vpxor %xmm2, %xmm2, %xmm2
; AVX512VL-NEXT: vpgatherdd c(,%ymm1), %ymm2 {%k2}
; AVX512VL-NEXT: vpbroadcastd {{.*#+}} ymm1 = [28,28,28,28,28,28,28,28]
; AVX512VL-NEXT: vpgatherdd c(,%ymm1), %ymm0 {%k1}
; AVX512VL-NEXT: vpaddd %ymm0, %ymm2, %ymm1
; AVX512VL-NEXT: vpaddd %ymm0, %ymm1, %ymm0
; AVX512VL-NEXT: vpgatherdd c+28(,%ymm0), %ymm2 {%k1}
; AVX512VL-NEXT: vpaddd %ymm2, %ymm1, %ymm0
; AVX512VL-NEXT: vpaddd %ymm2, %ymm0, %ymm0
; AVX512VL-NEXT: retq
%1 = icmp eq <8 x i32> %trigger, zeroinitializer
%2 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> getelementptr (%struct.a, <8 x ptr> <ptr @c, ptr @c, ptr @c, ptr @c, ptr @c, ptr @c, ptr @c, ptr @c>, <8 x i64> zeroinitializer, i32 0, <8 x i64> <i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3>), i32 4, <8 x i1> %1, <8 x i32> undef)
Expand Down