Skip to content
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

Jump to bottom

[LV][EVL] Peek through combination tail-folded + predicated masks #133430

Merged
merged 3 commits into from
Mar 31, 2025
Merged

[LV][EVL] Peek through combination tail-folded + predicated masks #133430

merged 3 commits into from
Mar 31, 2025

Conversation

lukel97
Copy link
Contributor

@lukel97 lukel97 commented Mar 28, 2025 •
edited
Loading

If a recipe was predicated and tail folded at the same time, it will have a mask like

EMIT vp<%header-mask> = icmp ule canonical-iv, backedge-tc
EMIT vp<%mask> = logical-and vp<%header-mask>, vp<%pred-mask>

When converting to an EVL recipe, if the mask isn't exactly just the header-mask we copy the whole logical-and.
We can remove this redundant logical-and (because it's now covered by EVL) and just use vp<%pred-mask> instead.

This lets us remove the widened canonical IV in more places.

@lukel97
[LV][EVL] Peek through combination tail-folded + predicated masks
77f93ae 
If an instruction was predicated and tail folded at the same time, it will have a mask like

    EMIT vp<%header-mask> = icmp ule canonical-iv, backedge-tc
    EMIT vp<%mask> = logical-and vp<%header-mask>, vp<%pred-mask>

When converting to an EVL recipe, if the mask isn't exactly just the header-mask we copy the whole logical-and.
We can remove this redundant logical-and (because we it's now covered by EVL) and just use vp<%pred-mask> instead.

This lets us remove the widened canonical IV in more places.
@llvmbot
Copy link
Member

llvmbot commented Mar 28, 2025

@llvm/pr-subscribers-vectorizers

Author: Luke Lau (lukel97)

Changes

If an instruction was predicated and tail folded at the same time, it will have a mask like

EMIT vp&lt;%header-mask&gt; = icmp ule canonical-iv, backedge-tc
EMIT vp&lt;%mask&gt; = logical-and vp&lt;%header-mask&gt;, vp&lt;%pred-mask&gt;

When converting to an EVL recipe, if the mask isn't exactly just the header-mask we copy the whole logical-and.
We can remove this redundant logical-and (because we it's now covered by EVL) and just use vp<%pred-mask> instead.

This lets us remove the widened canonical IV in more places.

Full diff: https://github.com/llvm/llvm-project/pull/133430.diff

4 Files Affected:

  • (modified) llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp (+4)
  • (modified) llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll (+1-11)
  • (modified) llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll (+2-12)
  • (modified) llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll (+2-9)
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index a2e6182cf16e2..3d5eb2aa82389 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1764,6 +1764,10 @@ static VPRecipeBase *createEVLRecipe(VPValue *HeaderMask,
   using namespace llvm::VPlanPatternMatch;
   auto GetNewMask = [&](VPValue *OrigMask) -> VPValue * {
     assert(OrigMask && "Unmasked recipe when folding tail");
+    using namespace llvm::VPlanPatternMatch;
+    VPValue *Mask;
+    if (match(OrigMask, m_LogicalAnd(m_Specific(HeaderMask), m_VPValue(Mask))))
+      return Mask;
     return HeaderMask == OrigMask ? nullptr : OrigMask;
   };
 
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll
index a40255c031619..354cc91c6f01f 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll
@@ -348,11 +348,8 @@ define i32 @cond_add_pred(ptr %a, i64 %n, i32 %start) {
 ; IF-EVL-INLOOP-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP8]]
 ; IF-EVL-INLOOP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP7]]
 ; IF-EVL-INLOOP-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
-; IF-EVL-INLOOP-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
 ; IF-EVL-INLOOP-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
 ; IF-EVL-INLOOP-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 4
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
 ; IF-EVL-INLOOP-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; IF-EVL-INLOOP:       vector.body:
 ; IF-EVL-INLOOP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -360,18 +357,11 @@ define i32 @cond_add_pred(ptr %a, i64 %n, i32 %start) {
 ; IF-EVL-INLOOP-NEXT:    [[VEC_PHI:%.*]] = phi i32 [ [[START]], [[VECTOR_PH]] ], [ [[TMP22:%.*]], [[VECTOR_BODY]] ]
 ; IF-EVL-INLOOP-NEXT:    [[TMP11:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP12:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP11]], i32 4, i1 true)
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
-; IF-EVL-INLOOP-NEXT:    [[TMP14:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
-; IF-EVL-INLOOP-NEXT:    [[TMP15:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP14]]
-; IF-EVL-INLOOP-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP15]]
-; IF-EVL-INLOOP-NEXT:    [[TMP16:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], [[BROADCAST_SPLAT2]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[EVL_BASED_IV]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i32 0
 ; IF-EVL-INLOOP-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP18]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP12]])
 ; IF-EVL-INLOOP-NEXT:    [[TMP19:%.*]] = icmp sgt <vscale x 4 x i32> [[VP_OP_LOAD]], splat (i32 3)
-; IF-EVL-INLOOP-NEXT:    [[TMP20:%.*]] = select <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP19]], <vscale x 4 x i1> zeroinitializer
-; IF-EVL-INLOOP-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vp.reduce.add.nxv4i32(i32 0, <vscale x 4 x i32> [[VP_OP_LOAD]], <vscale x 4 x i1> [[TMP20]], i32 [[TMP12]])
+; IF-EVL-INLOOP-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vp.reduce.add.nxv4i32(i32 0, <vscale x 4 x i32> [[VP_OP_LOAD]], <vscale x 4 x i1> [[TMP19]], i32 [[TMP12]])
 ; IF-EVL-INLOOP-NEXT:    [[TMP22]] = add i32 [[TMP21]], [[VEC_PHI]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP23:%.*]] = zext i32 [[TMP12]] to i64
 ; IF-EVL-INLOOP-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP23]], [[EVL_BASED_IV]]
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
index bf3f01343eb24..73c14f86e2782 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
@@ -24,33 +24,23 @@ define void @masked_loadstore(ptr noalias %a, ptr noalias %b, i64 %n) {
 ; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP6]]
 ; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
 ; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
-; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
 ; IF-EVL-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
 ; IF-EVL-NEXT:    [[TMP8:%.*]] = mul i64 [[TMP7]], 4
-; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
-; IF-EVL-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
 ; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; IF-EVL:       vector.body:
 ; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; IF-EVL-NEXT:    [[TMP9:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[TMP10:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP9]], i32 4, i1 true)
-; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
-; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
-; IF-EVL-NEXT:    [[TMP12:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
-; IF-EVL-NEXT:    [[TMP13:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP12]]
-; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP13]]
-; IF-EVL-NEXT:    [[TMP14:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], [[BROADCAST_SPLAT2]]
 ; IF-EVL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP15]], i32 0
 ; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP16]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP10]])
 ; IF-EVL-NEXT:    [[TMP17:%.*]] = icmp ne <vscale x 4 x i32> [[VP_OP_LOAD]], zeroinitializer
-; IF-EVL-NEXT:    [[TMP18:%.*]] = select <vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> [[TMP17]], <vscale x 4 x i1> zeroinitializer
 ; IF-EVL-NEXT:    [[TMP19:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[TMP19]], i32 0
-; IF-EVL-NEXT:    [[VP_OP_LOAD3:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP18]], i32 [[TMP10]])
+; IF-EVL-NEXT:    [[VP_OP_LOAD3:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP17]], i32 [[TMP10]])
 ; IF-EVL-NEXT:    [[VP_OP:%.*]] = add <vscale x 4 x i32> [[VP_OP_LOAD]], [[VP_OP_LOAD3]]
-; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP]], ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP18]], i32 [[TMP10]])
+; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP]], ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP17]], i32 [[TMP10]])
 ; IF-EVL-NEXT:    [[TMP21:%.*]] = zext i32 [[TMP10]] to i64
 ; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP21]], [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP8]]
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
index b6d92caa46ab0..695af0d241159 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
@@ -132,25 +132,18 @@ define void @reverse_load_store_masked(i64 %startval, ptr noalias %ptr, ptr noal
 ; IF-EVL-NEXT:    [[TMP5:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
 ; IF-EVL-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL]], [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[OFFSET_IDX3:%.*]] = trunc i64 [[EVL_BASED_IV]] to i32
-; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
-; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
-; IF-EVL-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
-; IF-EVL-NEXT:    [[TMP9:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP8]]
-; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP9]]
-; IF-EVL-NEXT:    [[TMP10:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], splat (i64 1023)
 ; IF-EVL-NEXT:    [[TMP11:%.*]] = add i64 [[OFFSET_IDX]], -1
 ; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i32 [[OFFSET_IDX3]]
 ; IF-EVL-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 0
 ; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP13]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[TMP14:%.*]] = icmp slt <vscale x 4 x i32> [[VP_OP_LOAD]], splat (i32 100)
-; IF-EVL-NEXT:    [[TMP15:%.*]] = select <vscale x 4 x i1> [[TMP10]], <vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> zeroinitializer
 ; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr i32, ptr [[PTR1:%.*]], i64 [[TMP11]]
 ; IF-EVL-NEXT:    [[TMP26:%.*]] = zext i32 [[TMP5]] to i64
 ; IF-EVL-NEXT:    [[TMP17:%.*]] = mul i64 0, [[TMP26]]
 ; IF-EVL-NEXT:    [[TMP18:%.*]] = sub i64 1, [[TMP26]]
 ; IF-EVL-NEXT:    [[TMP19:%.*]] = getelementptr i32, ptr [[TMP16]], i64 [[TMP17]]
 ; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[TMP19]], i64 [[TMP18]]
-; IF-EVL-NEXT:    [[VP_REVERSE_MASK:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP15]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
+; IF-EVL-NEXT:    [[VP_REVERSE_MASK:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[VP_OP_LOAD4:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[VP_REVERSE_MASK]], i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[VP_REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> [[VP_OP_LOAD4]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[PTR2:%.*]], i64 [[TMP11]]
@@ -160,7 +153,7 @@ define void @reverse_load_store_masked(i64 %startval, ptr noalias %ptr, ptr noal
 ; IF-EVL-NEXT:    [[TMP24:%.*]] = getelementptr i32, ptr [[TMP21]], i64 [[TMP22]]
 ; IF-EVL-NEXT:    [[TMP25:%.*]] = getelementptr i32, ptr [[TMP24]], i64 [[TMP23]]
 ; IF-EVL-NEXT:    [[VP_REVERSE5:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> [[VP_REVERSE]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
-; IF-EVL-NEXT:    [[VP_REVERSE_MASK6:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP15]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
+; IF-EVL-NEXT:    [[VP_REVERSE_MASK6:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_REVERSE5]], ptr align 4 [[TMP25]], <vscale x 4 x i1> [[VP_REVERSE_MASK6]], i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[TMP28:%.*]] = zext i32 [[TMP5]] to i64
 ; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add nuw i64 [[TMP28]], [[EVL_BASED_IV]]

@llvmbot
Copy link
Member

llvmbot commented Mar 28, 2025

@llvm/pr-subscribers-llvm-transforms

Author: Luke Lau (lukel97)

Changes

If an instruction was predicated and tail folded at the same time, it will have a mask like

EMIT vp&lt;%header-mask&gt; = icmp ule canonical-iv, backedge-tc
EMIT vp&lt;%mask&gt; = logical-and vp&lt;%header-mask&gt;, vp&lt;%pred-mask&gt;

When converting to an EVL recipe, if the mask isn't exactly just the header-mask we copy the whole logical-and.
We can remove this redundant logical-and (because we it's now covered by EVL) and just use vp<%pred-mask> instead.

This lets us remove the widened canonical IV in more places.

Full diff: https://github.com/llvm/llvm-project/pull/133430.diff

4 Files Affected:

  • (modified) llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp (+4)
  • (modified) llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll (+1-11)
  • (modified) llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll (+2-12)
  • (modified) llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll (+2-9)
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index a2e6182cf16e2..3d5eb2aa82389 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1764,6 +1764,10 @@ static VPRecipeBase *createEVLRecipe(VPValue *HeaderMask,
   using namespace llvm::VPlanPatternMatch;
   auto GetNewMask = [&](VPValue *OrigMask) -> VPValue * {
     assert(OrigMask && "Unmasked recipe when folding tail");
+    using namespace llvm::VPlanPatternMatch;
+    VPValue *Mask;
+    if (match(OrigMask, m_LogicalAnd(m_Specific(HeaderMask), m_VPValue(Mask))))
+      return Mask;
     return HeaderMask == OrigMask ? nullptr : OrigMask;
   };
 
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll
index a40255c031619..354cc91c6f01f 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-cond-reduction.ll
@@ -348,11 +348,8 @@ define i32 @cond_add_pred(ptr %a, i64 %n, i32 %start) {
 ; IF-EVL-INLOOP-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP8]]
 ; IF-EVL-INLOOP-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP7]]
 ; IF-EVL-INLOOP-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
-; IF-EVL-INLOOP-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
 ; IF-EVL-INLOOP-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
 ; IF-EVL-INLOOP-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 4
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
 ; IF-EVL-INLOOP-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; IF-EVL-INLOOP:       vector.body:
 ; IF-EVL-INLOOP-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -360,18 +357,11 @@ define i32 @cond_add_pred(ptr %a, i64 %n, i32 %start) {
 ; IF-EVL-INLOOP-NEXT:    [[VEC_PHI:%.*]] = phi i32 [ [[START]], [[VECTOR_PH]] ], [ [[TMP22:%.*]], [[VECTOR_BODY]] ]
 ; IF-EVL-INLOOP-NEXT:    [[TMP11:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP12:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP11]], i32 4, i1 true)
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
-; IF-EVL-INLOOP-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
-; IF-EVL-INLOOP-NEXT:    [[TMP14:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
-; IF-EVL-INLOOP-NEXT:    [[TMP15:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP14]]
-; IF-EVL-INLOOP-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP15]]
-; IF-EVL-INLOOP-NEXT:    [[TMP16:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], [[BROADCAST_SPLAT2]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[EVL_BASED_IV]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i32 0
 ; IF-EVL-INLOOP-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP18]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP12]])
 ; IF-EVL-INLOOP-NEXT:    [[TMP19:%.*]] = icmp sgt <vscale x 4 x i32> [[VP_OP_LOAD]], splat (i32 3)
-; IF-EVL-INLOOP-NEXT:    [[TMP20:%.*]] = select <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP19]], <vscale x 4 x i1> zeroinitializer
-; IF-EVL-INLOOP-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vp.reduce.add.nxv4i32(i32 0, <vscale x 4 x i32> [[VP_OP_LOAD]], <vscale x 4 x i1> [[TMP20]], i32 [[TMP12]])
+; IF-EVL-INLOOP-NEXT:    [[TMP21:%.*]] = call i32 @llvm.vp.reduce.add.nxv4i32(i32 0, <vscale x 4 x i32> [[VP_OP_LOAD]], <vscale x 4 x i1> [[TMP19]], i32 [[TMP12]])
 ; IF-EVL-INLOOP-NEXT:    [[TMP22]] = add i32 [[TMP21]], [[VEC_PHI]]
 ; IF-EVL-INLOOP-NEXT:    [[TMP23:%.*]] = zext i32 [[TMP12]] to i64
 ; IF-EVL-INLOOP-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP23]], [[EVL_BASED_IV]]
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
index bf3f01343eb24..73c14f86e2782 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-masked-loadstore.ll
@@ -24,33 +24,23 @@ define void @masked_loadstore(ptr noalias %a, ptr noalias %b, i64 %n) {
 ; IF-EVL-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP6]]
 ; IF-EVL-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
 ; IF-EVL-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
-; IF-EVL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[N]], 1
 ; IF-EVL-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
 ; IF-EVL-NEXT:    [[TMP8:%.*]] = mul i64 [[TMP7]], 4
-; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
-; IF-EVL-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
 ; IF-EVL-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; IF-EVL:       vector.body:
 ; IF-EVL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; IF-EVL-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; IF-EVL-NEXT:    [[TMP9:%.*]] = sub i64 [[N]], [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[TMP10:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[TMP9]], i32 4, i1 true)
-; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
-; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
-; IF-EVL-NEXT:    [[TMP12:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
-; IF-EVL-NEXT:    [[TMP13:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP12]]
-; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP13]]
-; IF-EVL-NEXT:    [[TMP14:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], [[BROADCAST_SPLAT2]]
 ; IF-EVL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP15]], i32 0
 ; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP16]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP10]])
 ; IF-EVL-NEXT:    [[TMP17:%.*]] = icmp ne <vscale x 4 x i32> [[VP_OP_LOAD]], zeroinitializer
-; IF-EVL-NEXT:    [[TMP18:%.*]] = select <vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> [[TMP17]], <vscale x 4 x i1> zeroinitializer
 ; IF-EVL-NEXT:    [[TMP19:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[TMP19]], i32 0
-; IF-EVL-NEXT:    [[VP_OP_LOAD3:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP18]], i32 [[TMP10]])
+; IF-EVL-NEXT:    [[VP_OP_LOAD3:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP17]], i32 [[TMP10]])
 ; IF-EVL-NEXT:    [[VP_OP:%.*]] = add <vscale x 4 x i32> [[VP_OP_LOAD]], [[VP_OP_LOAD3]]
-; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP]], ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP18]], i32 [[TMP10]])
+; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_OP]], ptr align 4 [[TMP20]], <vscale x 4 x i1> [[TMP17]], i32 [[TMP10]])
 ; IF-EVL-NEXT:    [[TMP21:%.*]] = zext i32 [[TMP10]] to i64
 ; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP21]], [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP8]]
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
index b6d92caa46ab0..695af0d241159 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/vectorize-force-tail-with-evl-reverse-load-store.ll
@@ -132,25 +132,18 @@ define void @reverse_load_store_masked(i64 %startval, ptr noalias %ptr, ptr noal
 ; IF-EVL-NEXT:    [[TMP5:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 4, i1 true)
 ; IF-EVL-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[STARTVAL]], [[EVL_BASED_IV]]
 ; IF-EVL-NEXT:    [[OFFSET_IDX3:%.*]] = trunc i64 [[EVL_BASED_IV]] to i32
-; IF-EVL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[EVL_BASED_IV]], i64 0
-; IF-EVL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
-; IF-EVL-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
-; IF-EVL-NEXT:    [[TMP9:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP8]]
-; IF-EVL-NEXT:    [[VEC_IV:%.*]] = add <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[TMP9]]
-; IF-EVL-NEXT:    [[TMP10:%.*]] = icmp ule <vscale x 4 x i64> [[VEC_IV]], splat (i64 1023)
 ; IF-EVL-NEXT:    [[TMP11:%.*]] = add i64 [[OFFSET_IDX]], -1
 ; IF-EVL-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i32 [[OFFSET_IDX3]]
 ; IF-EVL-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i32 0
 ; IF-EVL-NEXT:    [[VP_OP_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP13]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[TMP14:%.*]] = icmp slt <vscale x 4 x i32> [[VP_OP_LOAD]], splat (i32 100)
-; IF-EVL-NEXT:    [[TMP15:%.*]] = select <vscale x 4 x i1> [[TMP10]], <vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> zeroinitializer
 ; IF-EVL-NEXT:    [[TMP16:%.*]] = getelementptr i32, ptr [[PTR1:%.*]], i64 [[TMP11]]
 ; IF-EVL-NEXT:    [[TMP26:%.*]] = zext i32 [[TMP5]] to i64
 ; IF-EVL-NEXT:    [[TMP17:%.*]] = mul i64 0, [[TMP26]]
 ; IF-EVL-NEXT:    [[TMP18:%.*]] = sub i64 1, [[TMP26]]
 ; IF-EVL-NEXT:    [[TMP19:%.*]] = getelementptr i32, ptr [[TMP16]], i64 [[TMP17]]
 ; IF-EVL-NEXT:    [[TMP20:%.*]] = getelementptr i32, ptr [[TMP19]], i64 [[TMP18]]
-; IF-EVL-NEXT:    [[VP_REVERSE_MASK:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP15]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
+; IF-EVL-NEXT:    [[VP_REVERSE_MASK:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[VP_OP_LOAD4:%.*]] = call <vscale x 4 x i32> @llvm.vp.load.nxv4i32.p0(ptr align 4 [[TMP20]], <vscale x 4 x i1> [[VP_REVERSE_MASK]], i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[VP_REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> [[VP_OP_LOAD4]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[PTR2:%.*]], i64 [[TMP11]]
@@ -160,7 +153,7 @@ define void @reverse_load_store_masked(i64 %startval, ptr noalias %ptr, ptr noal
 ; IF-EVL-NEXT:    [[TMP24:%.*]] = getelementptr i32, ptr [[TMP21]], i64 [[TMP22]]
 ; IF-EVL-NEXT:    [[TMP25:%.*]] = getelementptr i32, ptr [[TMP24]], i64 [[TMP23]]
 ; IF-EVL-NEXT:    [[VP_REVERSE5:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> [[VP_REVERSE]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
-; IF-EVL-NEXT:    [[VP_REVERSE_MASK6:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP15]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
+; IF-EVL-NEXT:    [[VP_REVERSE_MASK6:%.*]] = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> [[TMP14]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP5]])
 ; IF-EVL-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[VP_REVERSE5]], ptr align 4 [[TMP25]], <vscale x 4 x i1> [[VP_REVERSE_MASK6]], i32 [[TMP5]])
 ; IF-EVL-NEXT:    [[TMP28:%.*]] = zext i32 [[TMP5]] to i64
 ; IF-EVL-NEXT:    [[INDEX_EVL_NEXT]] = add nuw i64 [[TMP28]], [[EVL_BASED_IV]]

alexey-bataev
alexey-bataev approved these changes Mar 28, 2025
View reviewed changes
Copy link
Member

@alexey-bataev alexey-bataev left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LG

fhahn
fhahn reviewed Mar 28, 2025
View reviewed changes
llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp Outdated
@@ -1764,6 +1764,10 @@ static VPRecipeBase *createEVLRecipe(VPValue *HeaderMask,
using namespace llvm::VPlanPatternMatch;
auto GetNewMask = [&](VPValue *OrigMask) -> VPValue * {
assert(OrigMask && "Unmasked recipe when folding tail");
using namespace llvm::VPlanPatternMatch;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this needed? Does the using at line 1764 not apply?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oops yes, removed

fhahn
fhahn approved these changes Mar 30, 2025
View reviewed changes
Copy link
Contributor

@fhahn fhahn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, thanks!

arcbbb
arcbbb approved these changes Mar 31, 2025
View reviewed changes
Copy link
Contributor

@arcbbb arcbbb left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@lukel97 lukel97 merged commit 6afe5e5 into llvm:main Mar 31, 2025
11 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@fhahn fhahn fhahn approved these changes

@arcbbb arcbbb arcbbb approved these changes

@alexey-bataev alexey-bataev alexey-bataev approved these changes

@ElvisWang123 ElvisWang123 Awaiting requested review from ElvisWang123

@Mel-Chen Mel-Chen Awaiting requested review from Mel-Chen

@mshockwave mshockwave Awaiting requested review from mshockwave

Assignees
No one assigned
Labels
llvm:transforms vectorizers
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
5 participants
@lukel97 @llvmbot @fhahn @arcbbb @alexey-bataev