[VPlan] Remove VPVectorPointer for part 0 after unrolling. #149735
+3,795
−6,032
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@llvm/pr-subscribers-backend-systemz @llvm/pr-subscribers-backend-risc-v Author: Florian Hahn (fhahn) ChangesVPVectorPointer for part 0 is just the pointer operand. Simplify it after unrolling. This removes a large number of redundant GEPs with index 0. Patch is 2.18 MiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/149735.diff 289 Files Affected:
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 204268e586b43..d80de9b960d14 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -1835,6 +1835,8 @@ class VPVectorPointerRecipe : public VPRecipeWithIRFlags,
getGEPNoWrapFlags(), getDebugLoc());
}
+ bool isPart0() { return getUnrollPart(*this) == 0; }
+
/// Return the cost of this VPHeaderPHIRecipe.
InstructionCost computeCost(ElementCount VF,
VPCostContext &Ctx) const override {
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 2a920832f272f..666ff2354c53a 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1015,6 +1015,14 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
if (Op->isLiveIn())
PredPHI->replaceAllUsesWith(Op);
}
+ if (auto *VecPtr= dyn_cast<VPVectorPointerRecipe>(&R)) {
+ if (VecPtr->getParent()->getPlan()->isUnrolled() && VecPtr->isPart0()) {
+ VecPtr->replaceAllUsesWith(VecPtr->getOperand(0));
+ return;
+ }
+ }
+
+
VPValue *A;
if (match(Def, m_Trunc(m_ZExtOrSExt(m_VPValue(A))))) {
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/blend-costs.ll b/llvm/test/Transforms/LoopVectorize/AArch64/blend-costs.ll
index 43b942458a39e..62329b63bc5e5 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/blend-costs.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/blend-costs.ll
@@ -22,8 +22,7 @@ define void @test_blend_feeding_replicated_store_1(i64 %N, ptr noalias %src, ptr
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE30:.*]] ]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP5]], align 4
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp slt <16 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP7:%.*]] = select <16 x i1> [[TMP6]], <16 x i1> zeroinitializer, <16 x i1> zeroinitializer
; CHECK-NEXT: [[TMP8:%.*]] = xor <16 x i1> [[TMP6]], splat (i1 true)
@@ -213,8 +212,7 @@ define void @test_blend_feeding_replicated_store_2(ptr noalias %src, ptr %dst, i
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE30:.*]] ]
; CHECK-NEXT: [[GEP_SRC:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i32 [[IV]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[GEP_SRC]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP2]], align 1
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[GEP_SRC]], align 1
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq <16 x i8> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP4:%.*]] = xor <16 x i1> [[TMP3]], splat (i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = select <16 x i1> [[TMP4]], <16 x i1> [[TMP5]], <16 x i1> zeroinitializer
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/call-costs.ll b/llvm/test/Transforms/LoopVectorize/AArch64/call-costs.ll
index 8c2a48aa38695..7f36e07f924e3 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/call-costs.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/call-costs.ll
@@ -14,18 +14,16 @@ define void @fshl_operand_first_order_recurrence(ptr %dst, ptr noalias %src) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <2 x i64> [ <i64 poison, i64 0>, %[[VECTOR_PH]] ], [ [[WIDE_LOAD1:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[SRC]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 2
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP2]], align 8
; CHECK-NEXT: [[WIDE_LOAD1]] = load <2 x i64>, ptr [[TMP5]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <2 x i64> [[VECTOR_RECUR]], <2 x i64> [[WIDE_LOAD]], <2 x i32> <i32 1, i32 2>
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <2 x i64> [[WIDE_LOAD]], <2 x i64> [[WIDE_LOAD1]], <2 x i32> <i32 1, i32 2>
; CHECK-NEXT: [[TMP8:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> splat (i64 1), <2 x i64> [[TMP6]], <2 x i64> splat (i64 1))
; CHECK-NEXT: [[TMP9:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> splat (i64 1), <2 x i64> [[TMP7]], <2 x i64> splat (i64 1))
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP10]], i32 0
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[TMP10]], i32 2
-; CHECK-NEXT: store <2 x i64> [[TMP8]], ptr [[TMP12]], align 8
+; CHECK-NEXT: store <2 x i64> [[TMP8]], ptr [[TMP10]], align 8
; CHECK-NEXT: store <2 x i64> [[TMP9]], ptr [[TMP13]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
@@ -79,11 +77,9 @@ define void @powi_call(ptr %P) {
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[P]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[P]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = call <2 x double> @llvm.powi.v2f64.i32(<2 x double> [[WIDE_LOAD]], i32 3)
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[P]], i32 0
-; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[TMP4]], align 8
+; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[P]], align 8
; CHECK-NEXT: br label %[[MIDDLE_BLOCK:.*]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/clamped-trip-count.ll b/llvm/test/Transforms/LoopVectorize/AArch64/clamped-trip-count.ll
index 95f3eb7b21f4e..795de3d978e74 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/clamped-trip-count.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/clamped-trip-count.ll
@@ -33,8 +33,7 @@ define void @clamped_tc_8(ptr nocapture %dst, i32 %n, i64 %val) vscale_range(1,1
; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw <vscale x 8 x i64> [[VEC_IND]], splat (i64 3)
; CHECK-NEXT: [[TMP11:%.*]] = lshr <vscale x 8 x i64> [[BROADCAST_SPLAT]], [[TMP10]]
; CHECK-NEXT: [[TMP14:%.*]] = trunc <vscale x 8 x i64> [[TMP11]] to <vscale x 8 x i8>
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 0
-; CHECK-NEXT: call void @llvm.masked.store.nxv8i8.p0(<vscale x 8 x i8> [[TMP14]], ptr [[TMP17]], i32 1, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: call void @llvm.masked.store.nxv8i8.p0(<vscale x 8 x i8> [[TMP14]], ptr [[NEXT_GEP]], i32 1, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_NEXT]], i64 8)
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 8 x i64> [[VEC_IND]], [[DOTSPLAT]]
@@ -117,8 +116,7 @@ define void @clamped_tc_max_8(ptr nocapture %dst, i32 %n, i64 %val) vscale_range
; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw <vscale x 8 x i64> [[VEC_IND]], splat (i64 3)
; CHECK-NEXT: [[TMP11:%.*]] = lshr <vscale x 8 x i64> [[BROADCAST_SPLAT]], [[TMP10]]
; CHECK-NEXT: [[TMP14:%.*]] = trunc <vscale x 8 x i64> [[TMP11]] to <vscale x 8 x i8>
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 0
-; CHECK-NEXT: call void @llvm.masked.store.nxv8i8.p0(<vscale x 8 x i8> [[TMP14]], ptr [[TMP17]], i32 1, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: call void @llvm.masked.store.nxv8i8.p0(<vscale x 8 x i8> [[TMP14]], ptr [[NEXT_GEP]], i32 1, <vscale x 8 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64 [[INDEX_NEXT]], i64 [[WIDE_TRIP_COUNT]])
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 8 x i64> [[VEC_IND]], [[DOTSPLAT]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll b/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
index 46a194dacad9e..9a75d797bf0b7 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
@@ -82,9 +82,8 @@ define void @loop_dependent_cond(ptr %src, ptr noalias %dst, i64 %N) {
; DEFAULT: [[VECTOR_BODY]]:
; DEFAULT-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE7:.*]] ]
; DEFAULT-NEXT: [[TMP3:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; DEFAULT-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[TMP3]], i32 0
; DEFAULT-NEXT: [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i32 2
-; DEFAULT-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP5]], align 8
+; DEFAULT-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
; DEFAULT-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x double>, ptr [[TMP6]], align 8
; DEFAULT-NEXT: [[TMP7:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[WIDE_LOAD]])
; DEFAULT-NEXT: [[TMP8:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[WIDE_LOAD1]])
@@ -341,9 +340,8 @@ define void @latch_branch_cost(ptr %dst) {
; DEFAULT: [[VECTOR_BODY]]:
; DEFAULT-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; DEFAULT-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX]]
-; DEFAULT-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP2]], i32 0
; DEFAULT-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[TMP2]], i32 16
-; DEFAULT-NEXT: store <16 x i8> zeroinitializer, ptr [[TMP6]], align 1
+; DEFAULT-NEXT: store <16 x i8> zeroinitializer, ptr [[TMP2]], align 1
; DEFAULT-NEXT: store <16 x i8> zeroinitializer, ptr [[TMP5]], align 1
; DEFAULT-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 32
; DEFAULT-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 96
@@ -358,8 +356,7 @@ define void @latch_branch_cost(ptr %dst) {
; DEFAULT: [[VEC_EPILOG_VECTOR_BODY]]:
; DEFAULT-NEXT: [[INDEX1:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT2:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
; DEFAULT-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[DST]], i64 [[INDEX1]]
-; DEFAULT-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i32 0
-; DEFAULT-NEXT: store <4 x i8> zeroinitializer, ptr [[TMP9]], align 1
+; DEFAULT-NEXT: store <4 x i8> zeroinitializer, ptr [[TMP8]], align 1
; DEFAULT-NEXT: [[INDEX_NEXT2]] = add nuw i64 [[INDEX1]], 4
; DEFAULT-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT2]], 100
; DEFAULT-NEXT: br i1 [[TMP10]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
@@ -575,8 +572,7 @@ define i32 @header_mask_and_invariant_compare(ptr %A, ptr %B, ptr %C, ptr %D, pt
; DEFAULT-NEXT: store i32 [[TMP22]], ptr [[E]], align 4, !alias.scope [[META14]], !noalias [[META16]]
; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE37]]
; DEFAULT: [[PRED_STORE_CONTINUE37]]:
-; DEFAULT-NEXT: [[TMP17:%.*]] = getelementptr i32, ptr [[TMP16]], i32 0
-; DEFAULT-NEXT: call void @llvm.masked.store.v4i32.p0(<4 x i32> zeroinitializer, ptr [[TMP17]], i32 4, <4 x i1> [[TMP8]]), !alias.scope [[META18:![0-9]+]], !noalias [[META19:![0-9]+]]
+; DEFAULT-NEXT: call void @llvm.masked.store.v4i32.p0(<4 x i32> zeroinitializer, ptr [[TMP16]], i32 4, <4 x i1> [[TMP8]]), !alias.scope [[META18:![0-9]+]], !noalias [[META19:![0-9]+]]
; DEFAULT-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; DEFAULT-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; DEFAULT-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
@@ -674,8 +670,7 @@ define void @multiple_exit_conditions(ptr %src, ptr noalias %dst) #1 {
; DEFAULT-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x i16> [[BROADCAST_SPLATINSERT]], <8 x i16> poison, <8 x i32> zeroinitializer
; DEFAULT-NEXT: [[TMP2:%.*]] = or <8 x i16> [[BROADCAST_SPLAT]], splat (i16 1)
; DEFAULT-NEXT: [[TMP3:%.*]] = uitofp <8 x i16> [[TMP2]] to <8 x double>
-; DEFAULT-NEXT: [[TMP4:%.*]] = getelementptr double, ptr [[NEXT_GEP]], i32 0
-; DEFAULT-NEXT: store <8 x double> [[TMP3]], ptr [[TMP4]], align 8
+; DEFAULT-NEXT: store <8 x double> [[TMP3]], ptr [[NEXT_GEP]], align 8
; DEFAULT-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; DEFAULT-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
; DEFAULT-NEXT: br i1 [[TMP5]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
@@ -730,8 +725,7 @@ define void @multiple_exit_conditions(ptr %src, ptr noalias %dst) #1 {
; PRED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i16> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i16> poison, <vscale x 2 x i32> zeroinitializer
; PRED-NEXT: [[TMP13:%.*]] = or <vscale x 2 x i16> [[BROADCAST_SPLAT]], splat (i16 1)
; PRED-NEXT: [[TMP14:%.*]] = uitofp <vscale x 2 x i16> [[TMP13]] to <vscale x 2 x double>
-; PRED-NEXT: [[TMP15:%.*]] = getelementptr double, ptr [[NEXT_GEP]], i32 0
-; PRED-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP14]], ptr [[TMP15]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; PRED-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP14]], ptr [[NEXT_GEP]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
; PRED-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP5]]
; PRED-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP10]])
; PRED-NEXT: [[TMP16:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], splat (i1 true)
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll b/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll
index d42be20ea1e73..1ad1e42678c5a 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/divs-with-scalable-vfs.ll
@@ -38,11 +38,10 @@ define void @sdiv_feeding_gep(ptr %dst, i32 %x, i64 %M, i64 %conv6, i64 %N) {
; CHECK-NEXT: [[TMP30:%.*]] = add i32 [[TMP28]], [[TMP26]]
; CHECK-NEXT: [[TMP32:%.*]] = sext i32 [[TMP30]] to i64
; CHECK-NEXT: [[TMP34:%.*]] = getelementptr double, ptr [[DST]], i64 [[TMP32]]
-; CHECK-NEXT: [[TMP36:%.*]] = getelementptr double, ptr [[TMP34]], i32 0
; CHECK-NEXT: [[TMP37:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], 2
; CHECK-NEXT: [[TMP39:%.*]] = getelementptr double, ptr [[TMP34]], i64 [[TMP38]]
-; CHECK-NEXT: store <vscale x 2 x double> zeroinitializer, ptr [[TMP36]], align 8
+; CHECK-NEXT: store <vscale x 2 x double> zeroinitializer, ptr [[TMP34]], align 8
; CHECK-NEXT: store <vscale x 2 x double> zeroinitializer, ptr [[TMP39]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
; CHECK-NEXT: [[TMP40:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -149,8 +148,7 @@ define void @sdiv_feeding_gep_predicated(ptr %dst, i32 %x, i64 %M, i64 %conv6, i
; CHECK-NEXT: [[TMP32:%.*]] = add i32 [[TMP31]], [[TMP30]]
; CHECK-NEXT: [[TMP33:%.*]] = sext i32 [[TMP32]] to i64
; CHECK-NEXT: [[TMP34:%.*]] = getelementptr double, ptr [[DST]], i64 [[TMP33]]
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr double, ptr [[TMP34]], i32 0
-; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> zeroinitializer, ptr [[TMP35]], i32 8, <vscale x 2 x i1> [[TMP23]])
+; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> zeroinitializer, ptr [[TMP34]], i32 8, <vscale x 2 x i1> [[TMP23]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP9]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP14]])
; CHECK-NEXT: [[TMP36:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], splat (i1 true)
@@ -275,8 +273,7 @@ define void @udiv_urem_feeding_gep(i64 %x, ptr %dst, i64 %N) {
; CHECK-NEXT: [[TMP36:%.*]] = shl i64 [[TMP35]], 32
; CHECK-NEXT: [[TMP37:%.*]] = ashr i64 [[TMP36]], 32
; CHECK-NEXT: [[TMP38:%.*]] = getelementptr i64, ptr [[DST]], i64 [[TMP37]]
-; CHECK-NEXT: [[TMP39:%.*]] = getelementptr i64, ptr [[TMP38]], i32 0
-; CHECK-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[TMP23]], ptr [[TMP39]], i32 4, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[TMP23]], ptr [[TMP38]], i32 4, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP9]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX]], i64 [[TMP14]])
; CHECK-NEXT: [[TMP47:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], splat (i1 true)
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/drop-poison-generating-flags.ll b/llvm/test/Transforms/LoopVectorize/AArch64/drop-poison-generating-flags.ll
index e28c79eac1e5c..db9d05b7bbf29 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/drop-poison-generating-flags.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/drop-poison-generating-flags.ll
@@ -15,15 +15,13 @@ define void @check_widen_intrinsic_with_nnan(ptr noalias %dst.0, ptr noalias %ds
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_LOAD_CONTINUE6:.*]] ]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[SRC_1]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[TMP1]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP2]], align 8
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[WIDE_LOAD]])
; CHECK-NEXT: [[TMP4:%.*]] = fcmp olt <4 x double> [[TMP3]], splat (double 1.000000e+00)
; CHECK-NEXT: [[TMP5:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
; CHECK-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], -1
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr double, ptr [[DST_0]], i64 [[TMP6]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[TMP7]], i32 0
-; CHECK-NEXT: call void @llvm.masked.store.v4f64.p0(<4 x double> zeroinitializer, ptr [[TMP8]], i32 8, <4 x i1> [[TMP5]])
+; CHECK-NEXT: call void @llvm.masked.store.v4f64.p0(<4 x double> zeroinitializer, ptr [[TMP7]], i32 8, <4 x i1> [[TMP5]])
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP4]], i32 0
; CHECK-NEXT: br i1 [[TMP9]], label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
; CHECK: [[PRED_LOAD_IF]]:
@@ -58,16 +56,14 @@ define void @check_widen_intrinsic_with_nnan(ptr noalias %dst.0, ptr noalias %ds
; CHECK-NEXT: [[TMP24:%.*]] = phi <4 x double> [ [[TMP20]], %[[PRED_LOAD_CONTINUE4]] ], [ [[TMP23]], %[[PRED_LOAD_IF5]] ]
; CHECK-NEXT: [[TMP25:%.*]] = add i64 [[INDEX]], -1
; CHECK-NEXT: [[TMP26:%.*]] = getelementptr double, ptr [[DST_0]], i64 [[TMP25]]
-; CHECK-NEXT: [[TMP27:%.*]] = getelementptr double, ptr [[TMP26]], i32 0
-; CHECK-NEXT: call void @llvm.masked.store.v4f64.p0(<4 x double> zeroinitializer, ptr [[TMP27]], i32 8, <4 x i1> [[TMP4]])
+; CHECK-NEXT: call void @llvm.masked.store...
[truncated]
|
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
VPVectorPointer for part 0 is just the pointer operand. Simplify it after unrolling. This removes a large number of redundant GEPs with index 0.
fhahn
force-pushed
the
vplan-more-unroll-simps
branch
from
b0f2110 to
81d97e2
Compare
|
Looks like there's a test failing in |
lukel97
reviewed
Jul 21, 2025
| @@ -1015,6 +1015,12 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) { | |||
| if (Op->isLiveIn()) | |||
| PredPHI->replaceAllUsesWith(Op); | |||
| } | |||
| if (auto *VecPtr = dyn_cast<VPVectorPointerRecipe>(&R)) { | |||
| if (VecPtr->getParent()->getPlan()->isUnrolled() && VecPtr->isPart0()) { | |||
There was a problem hiding this comment.
Is it necessary to check if the plan is unrolled? I.e. can we also simplify it for UF=1?
Suggested change
| if (VecPtr->getParent()->getPlan()->isUnrolled() && VecPtr->isPart0()) { | |
| if (VecPtr->isPart0()) { |
Mel-Chen
reviewed
Jul 21, 2025
| @@ -1015,6 +1015,12 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) { | |||
| if (Op->isLiveIn()) | |||
| PredPHI->replaceAllUsesWith(Op); | |||
| } | |||
| if (auto *VecPtr = dyn_cast<VPVectorPointerRecipe>(&R)) { | |||
| if (VecPtr->getParent()->getPlan()->isUnrolled() && VecPtr->isPart0()) { | |||
| @@ -1835,6 +1835,8 @@ class VPVectorPointerRecipe : public VPRecipeWithIRFlags, | |||
| getGEPNoWrapFlags(), getDebugLoc()); | |||
| } | |||
|
|
|||
| bool isPart0() { return getUnrollPart(*this) == 0; } | |||
There was a problem hiding this comment.
nit: isFirstPart() or isZeroPart()
| @@ -1835,6 +1835,8 @@ class VPVectorPointerRecipe : public VPRecipeWithIRFlags, | |||
| getGEPNoWrapFlags(), getDebugLoc()); | |||
| } | |||
|
|
|||
| bool isPart0() { return getUnrollPart(*this) == 0; } | |||
There was a problem hiding this comment.
Suggested change
| bool isPart0() { return getUnrollPart(*this) == 0; } | |
| bool isPart0() const { return getUnrollPart(*this) == 0; } |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
VPVectorPointer for part 0 is just the pointer operand. Simplify it after unrolling. This removes a large number of redundant GEPs with index 0.