caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[LV] Enable vectorization of multiple exit loops w/computable exit counts

Browse Source 
This change enables vectorization of multiple exit loops when the exit count is statically computable. That requirement - shared with the rest of LV - in turn requires each exit to be analyzeable and to dominate the latch.

The majority of work to support this was done in a set of previous patches. In particular,, 72314466 avoids having multiple edges from the middle block to the exits, and 4b33b2387 which added support for non-latch single exit and multiple exits with a single exiting block. As a result, this change is basically just removing a bailout and adjusting some tests now that the prerequisite work is done and has stuck in tree for a bit.

Differential Revision: https://reviews.llvm.org/D105817
This commit is contained in:
Philip Reames
2021-07-15 08:52:52 -07:00
parent f59209a86e
commit 95346ba877
4 changed files with 105 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
llvm/lib/Transforms/Vectorize/LoopVectorizationLegality.cpp
View File

@@ -1125,21 +1125,6 @@ bool LoopVectorizationLegality::canVectorizeLoopCFG(Loop *Lp,
return false;
}
// We currently must have a single "exit block" after the loop. Note that
// multiple "exiting blocks" inside the loop are allowed, provided they all
// reach the single exit block.
// TODO: This restriction can be relaxed in the near future, it's here solely
// to allow separation of changes for review. We need to generalize the phi
// update logic in a number of places.
if (!Lp->getUniqueExitBlock()) {
reportVectorizationFailure("The loop must have a unique exit block",
"loop control flow is not understood by vectorizer",
"CFGNotUnderstood", ORE, TheLoop);
if (DoExtraAnalysis)
Result = false;
else
return false;
}
return Result;
}

120
llvm/test/Transforms/LoopVectorize/loop-form.ll
View File

@@ -514,18 +514,47 @@ if.end:
define i32 @multiple_exit_blocks(i16* %p, i32 %n) {
; CHECK-LABEL: @multiple_exit_blocks(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[N:%.*]], i32 0)
; CHECK-NEXT: [[UMIN:%.*]] = call i32 @llvm.umin.i32(i32 [[SMAX]], i32 2096)
; CHECK-NEXT: [[TMP0:%.*]] = add nuw nsw i32 [[UMIN]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i32 [[TMP0]], 2
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 2
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i32 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[TMP1]], i32 2, i32 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[TMP2]]
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP3:%.*]] = add i32 [[INDEX]], 0
; CHECK-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 1
; CHECK-NEXT: [[TMP5:%.*]] = sext i32 [[TMP3]] to i64
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i16, i16* [[P:%.*]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i16, i16* [[TMP6]], i32 0
; CHECK-NEXT: [[TMP8:%.*]] = bitcast i16* [[TMP7]] to <2 x i16>*
; CHECK-NEXT: store <2 x i16> zeroinitializer, <2 x i16>* [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2>
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_COND:%.*]]
; CHECK: for.cond:
; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I]], [[N:%.*]]
; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I]], [[N]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[IF_END:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IPROM:%.*]] = sext i32 [[I]] to i64
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i16, i16* [[P:%.*]], i64 [[IPROM]]
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i16, i16* [[P]], i64 [[IPROM]]
; CHECK-NEXT: store i16 0, i16* [[B]], align 4
; CHECK-NEXT: [[INC]] = add nsw i32 [[I]], 1
; CHECK-NEXT: [[CMP2:%.*]] = icmp slt i32 [[I]], 2096
; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_COND]], label [[IF_END2:%.*]]
; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_COND]], label [[IF_END2:%.*]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK: if.end:
; CHECK-NEXT: ret i32 0
; CHECK: if.end2:
@@ -577,18 +606,47 @@ if.end2:
define i32 @multiple_exit_blocks2(i16* %p, i32 %n) {
; CHECK-LABEL: @multiple_exit_blocks2(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[N:%.*]], i32 0)
; CHECK-NEXT: [[UMIN:%.*]] = call i32 @llvm.umin.i32(i32 [[SMAX]], i32 2096)
; CHECK-NEXT: [[TMP0:%.*]] = add nuw nsw i32 [[UMIN]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i32 [[TMP0]], 2
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 2
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i32 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[TMP1]], i32 2, i32 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[TMP2]]
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP3:%.*]] = add i32 [[INDEX]], 0
; CHECK-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 1
; CHECK-NEXT: [[TMP5:%.*]] = sext i32 [[TMP3]] to i64
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i16, i16* [[P:%.*]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i16, i16* [[TMP6]], i32 0
; CHECK-NEXT: [[TMP8:%.*]] = bitcast i16* [[TMP7]] to <2 x i16>*
; CHECK-NEXT: store <2 x i16> zeroinitializer, <2 x i16>* [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2>
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_COND:%.*]]
; CHECK: for.cond:
; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I]], [[N:%.*]]
; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I]], [[N]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[IF_END:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IPROM:%.*]] = sext i32 [[I]] to i64
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i16, i16* [[P:%.*]], i64 [[IPROM]]
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i16, i16* [[P]], i64 [[IPROM]]
; CHECK-NEXT: store i16 0, i16* [[B]], align 4
; CHECK-NEXT: [[INC]] = add nsw i32 [[I]], 1
; CHECK-NEXT: [[CMP2:%.*]] = icmp slt i32 [[I]], 2096
; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_COND]], label [[IF_END2:%.*]]
; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_COND]], label [[IF_END2:%.*]], !llvm.loop [[LOOP15:![0-9]+]]
; CHECK: if.end:
; CHECK-NEXT: [[I_LCSSA:%.*]] = phi i32 [ [[I]], [[FOR_COND]] ]
; CHECK-NEXT: ret i32 [[I_LCSSA]]
@@ -644,18 +702,48 @@ if.end2:
define i32 @multiple_exit_blocks3(i16* %p, i32 %n) {
; CHECK-LABEL: @multiple_exit_blocks3(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[N:%.*]], i32 0)
; CHECK-NEXT: [[UMIN:%.*]] = call i32 @llvm.umin.i32(i32 [[SMAX]], i32 2096)
; CHECK-NEXT: [[TMP0:%.*]] = add nuw nsw i32 [[UMIN]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i32 [[TMP0]], 2
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 2
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i32 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[TMP1]], i32 2, i32 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[TMP2]]
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP3:%.*]] = add i32 [[INDEX]], 0
; CHECK-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 1
; CHECK-NEXT: [[TMP5:%.*]] = sext i32 [[TMP3]] to i64
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i16, i16* [[P:%.*]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i16, i16* [[TMP6]], i32 0
; CHECK-NEXT: [[TMP8:%.*]] = bitcast i16* [[TMP7]] to <2 x i16>*
; CHECK-NEXT: store <2 x i16> zeroinitializer, <2 x i16>* [[TMP8]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = add nsw <2 x i32> [[VEC_IND]], <i32 1, i32 1>
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2>
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[FOR_COND:%.*]]
; CHECK: for.cond:
; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I]], [[N:%.*]]
; CHECK-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I]], [[N]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[IF_END:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IPROM:%.*]] = sext i32 [[I]] to i64
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i16, i16* [[P:%.*]], i64 [[IPROM]]
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i16, i16* [[P]], i64 [[IPROM]]
; CHECK-NEXT: store i16 0, i16* [[B]], align 4
; CHECK-NEXT: [[INC]] = add nsw i32 [[I]], 1
; CHECK-NEXT: [[CMP2:%.*]] = icmp slt i32 [[I]], 2096
; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_COND]], label [[IF_END2:%.*]]
; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_COND]], label [[IF_END2:%.*]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK: if.end:
; CHECK-NEXT: [[I_LCSSA:%.*]] = phi i32 [ [[I]], [[FOR_COND]] ]
; CHECK-NEXT: ret i32 [[I_LCSSA]]
@@ -983,7 +1071,7 @@ define void @scalar_predication(float* %addr) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 200
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -1003,7 +1091,7 @@ define void @scalar_predication(float* %addr) {
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: loop.latch:
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: br label [[LOOP_HEADER]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK-NEXT: br label [[LOOP_HEADER]], !llvm.loop [[LOOP19:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
@@ -1074,7 +1162,7 @@ define i32 @me_reduction(i32* %addr) {
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 200
; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[TMP7:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[TMP5]])
; CHECK-NEXT: br label [[SCALAR_PH]]
@@ -1093,7 +1181,7 @@ define i32 @me_reduction(i32* %addr) {
; CHECK-NEXT: [[ACCUM_NEXT]] = add i32 [[ACCUM]], [[TMP8]]
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND2_NOT:%.*]] = icmp eq i64 [[IV]], 400
; CHECK-NEXT: br i1 [[EXITCOND2_NOT]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP15:![0-9]+]]
; CHECK-NEXT: br i1 [[EXITCOND2_NOT]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP21:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: [[LCSSA:%.*]] = phi i32 [ 0, [[LOOP_HEADER]] ], [ [[ACCUM_NEXT]], [[LOOP_LATCH]] ]
; CHECK-NEXT: ret i32 [[LCSSA]]

17
llvm/test/Transforms/LoopVectorize/loop-legality-checks.ll
View File

@@ -1,23 +1,6 @@
; RUN: opt < %s -loop-vectorize -debug-only=loop-vectorize -S -disable-output 2>&1 | FileCheck %s
; REQUIRES: asserts
; Make sure LV legal bails out when there is no exiting block
; CHECK-LABEL: "no_exiting_block"
; CHECK: LV: Not vectorizing: The loop must have a unique exit block.
define i32 @no_exiting_block() {
entry:
br label %for.body
for.body:
%i.02 = phi i32 [ 0, %entry ], [ %inc, %for.body ], [%inc, %for.second]
%inc = add nsw i32 %i.02, 1
%cmp = icmp slt i32 %inc, 16
br i1 %cmp, label %for.body, label %for.second
for.second:
br label %for.body
}
; Make sure LV legal bails out when there is a non-int, non-ptr phi
; CHECK-LABEL: "invalid_phi_types"
; CHECK: LV: Not vectorizing: Found a non-int non-pointer PHI.

2
llvm/test/Transforms/LoopVectorize/remarks-multi-exit-loops.ll
View File

@@ -3,7 +3,7 @@
; Make sure LV does not crash when generating remarks for loops with non-unique
; exit blocks.
define i32 @test_non_unique_exit_blocks(i32* nocapture readonly align 4 dereferenceable(1024) %data, i32 %x) {
; CHECK: loop not vectorized: loop control flow is not understood by vectorizer
; CHECK: loop not vectorized: could not determine number of loop iterations
;
entry:
br label %for.header