shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/inloop-reduction.ll
Sander de Smalen 5a115452c4 Reland D143267: [LoopVectorize] Use DataLayout::getIndexType instead of i32 for non-constant GEP indices. 
Fixed issue where 'ConstantInt::get(IndextTy, -Part)' was executed with the wrong type for Part,
e.g. IndexTy was i64, but Part was 'unsigned', which led to things like 'mul i64 .., 4294967292',
which was obviously wrong.

Also changed sve-vector-reverse.ll to be vectorized with UF>1 to test this.

This reverts commit 1f01cdda68614dba12af3cc3aff38541d0abcc6b.
2023-02-09 09:42:29 +00:00

166 lines
10 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -mtriple riscv64-linux-gnu -mattr=+v,+d -passes=loop-vectorize < %s -S -o - | FileCheck %s -check-prefix=OUTLOOP
; RUN: opt -mtriple riscv64-linux-gnu -mattr=+v,+d -passes=loop-vectorize -prefer-inloop-reductions < %s -S -o - | FileCheck %s -check-prefix=INLOOP
target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128"
target triple = "riscv64"
define i32 @add_i16_i32(ptr nocapture readonly %x, i32 %n) {
; OUTLOOP-LABEL: @add_i16_i32(
; OUTLOOP-NEXT: entry:
; OUTLOOP-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[N:%.*]], 0
; OUTLOOP-NEXT: br i1 [[CMP6]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_COND_CLEANUP:%.*]]
; OUTLOOP: for.body.preheader:
; OUTLOOP-NEXT: [[TMP0:%.*]] = call i32 @llvm.vscale.i32()
; OUTLOOP-NEXT: [[TMP1:%.*]] = mul i32 [[TMP0]], 4
; OUTLOOP-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], [[TMP1]]
; OUTLOOP-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; OUTLOOP: vector.ph:
; OUTLOOP-NEXT: [[TMP2:%.*]] = call i32 @llvm.vscale.i32()
; OUTLOOP-NEXT: [[TMP3:%.*]] = mul i32 [[TMP2]], 4
; OUTLOOP-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[N]], [[TMP3]]
; OUTLOOP-NEXT: [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
; OUTLOOP-NEXT: br label [[VECTOR_BODY:%.*]]
; OUTLOOP: vector.body:
; OUTLOOP-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; OUTLOOP-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
; OUTLOOP-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP19:%.*]], [[VECTOR_BODY]] ]
; OUTLOOP-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 0
; OUTLOOP-NEXT: [[TMP5:%.*]] = call i32 @llvm.vscale.i32()
; OUTLOOP-NEXT: [[TMP6:%.*]] = mul i32 [[TMP5]], 2
; OUTLOOP-NEXT: [[TMP7:%.*]] = add i32 [[TMP6]], 0
; OUTLOOP-NEXT: [[TMP8:%.*]] = mul i32 [[TMP7]], 1
; OUTLOOP-NEXT: [[TMP9:%.*]] = add i32 [[INDEX]], [[TMP8]]
; OUTLOOP-NEXT: [[TMP10:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[TMP4]]
; OUTLOOP-NEXT: [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[TMP9]]
; OUTLOOP-NEXT: [[TMP12:%.*]] = getelementptr inbounds i16, ptr [[TMP10]], i32 0
; OUTLOOP-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i16>, ptr [[TMP12]], align 2
; OUTLOOP-NEXT: [[TMP13:%.*]] = call i64 @llvm.vscale.i64()
; OUTLOOP-NEXT: [[TMP14:%.*]] = mul i64 [[TMP13]], 2
; OUTLOOP-NEXT: [[TMP15:%.*]] = getelementptr inbounds i16, ptr [[TMP10]], i64 [[TMP14]]
; OUTLOOP-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 2 x i16>, ptr [[TMP15]], align 2
; OUTLOOP-NEXT: [[TMP16:%.*]] = sext <vscale x 2 x i16> [[WIDE_LOAD]] to <vscale x 2 x i32>
; OUTLOOP-NEXT: [[TMP17:%.*]] = sext <vscale x 2 x i16> [[WIDE_LOAD2]] to <vscale x 2 x i32>
; OUTLOOP-NEXT: [[TMP18]] = add <vscale x 2 x i32> [[VEC_PHI]], [[TMP16]]
; OUTLOOP-NEXT: [[TMP19]] = add <vscale x 2 x i32> [[VEC_PHI1]], [[TMP17]]
; OUTLOOP-NEXT: [[TMP20:%.*]] = call i32 @llvm.vscale.i32()
; OUTLOOP-NEXT: [[TMP21:%.*]] = mul i32 [[TMP20]], 4
; OUTLOOP-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], [[TMP21]]
; OUTLOOP-NEXT: [[TMP22:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; OUTLOOP-NEXT: br i1 [[TMP22]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; OUTLOOP: middle.block:
; OUTLOOP-NEXT: [[BIN_RDX:%.*]] = add <vscale x 2 x i32> [[TMP19]], [[TMP18]]
; OUTLOOP-NEXT: [[TMP23:%.*]] = call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> [[BIN_RDX]])
; OUTLOOP-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
; OUTLOOP-NEXT: br i1 [[CMP_N]], label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; OUTLOOP: scalar.ph:
; OUTLOOP-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; OUTLOOP-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[FOR_BODY_PREHEADER]] ], [ [[TMP23]], [[MIDDLE_BLOCK]] ]
; OUTLOOP-NEXT: br label [[FOR_BODY:%.*]]
; OUTLOOP: for.body:
; OUTLOOP-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; OUTLOOP-NEXT: [[R_07:%.*]] = phi i32 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; OUTLOOP-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_08]]
; OUTLOOP-NEXT: [[TMP24:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; OUTLOOP-NEXT: [[CONV:%.*]] = sext i16 [[TMP24]] to i32
; OUTLOOP-NEXT: [[ADD]] = add nsw i32 [[R_07]], [[CONV]]
; OUTLOOP-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; OUTLOOP-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], [[N]]
; OUTLOOP-NEXT: br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; OUTLOOP: for.cond.cleanup.loopexit:
; OUTLOOP-NEXT: [[ADD_LCSSA:%.*]] = phi i32 [ [[ADD]], [[FOR_BODY]] ], [ [[TMP23]], [[MIDDLE_BLOCK]] ]
; OUTLOOP-NEXT: br label [[FOR_COND_CLEANUP]]
; OUTLOOP: for.cond.cleanup:
; OUTLOOP-NEXT: [[R_0_LCSSA:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[ADD_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT]] ]
; OUTLOOP-NEXT: ret i32 [[R_0_LCSSA]]
;
; INLOOP-LABEL: @add_i16_i32(
; INLOOP-NEXT: entry:
; INLOOP-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[N:%.*]], 0
; INLOOP-NEXT: br i1 [[CMP6]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_COND_CLEANUP:%.*]]
; INLOOP: for.body.preheader:
; INLOOP-NEXT: [[TMP0:%.*]] = call i32 @llvm.vscale.i32()
; INLOOP-NEXT: [[TMP1:%.*]] = mul i32 [[TMP0]], 8
; INLOOP-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], [[TMP1]]
; INLOOP-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; INLOOP: vector.ph:
; INLOOP-NEXT: [[TMP2:%.*]] = call i32 @llvm.vscale.i32()
; INLOOP-NEXT: [[TMP3:%.*]] = mul i32 [[TMP2]], 8
; INLOOP-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[N]], [[TMP3]]
; INLOOP-NEXT: [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
; INLOOP-NEXT: br label [[VECTOR_BODY:%.*]]
; INLOOP: vector.body:
; INLOOP-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; INLOOP-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP19:%.*]], [[VECTOR_BODY]] ]
; INLOOP-NEXT: [[VEC_PHI1:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
; INLOOP-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 0
; INLOOP-NEXT: [[TMP5:%.*]] = call i32 @llvm.vscale.i32()
; INLOOP-NEXT: [[TMP6:%.*]] = mul i32 [[TMP5]], 4
; INLOOP-NEXT: [[TMP7:%.*]] = add i32 [[TMP6]], 0
; INLOOP-NEXT: [[TMP8:%.*]] = mul i32 [[TMP7]], 1
; INLOOP-NEXT: [[TMP9:%.*]] = add i32 [[INDEX]], [[TMP8]]
; INLOOP-NEXT: [[TMP10:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[TMP4]]
; INLOOP-NEXT: [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[TMP9]]
; INLOOP-NEXT: [[TMP12:%.*]] = getelementptr inbounds i16, ptr [[TMP10]], i32 0
; INLOOP-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i16>, ptr [[TMP12]], align 2
; INLOOP-NEXT: [[TMP13:%.*]] = call i64 @llvm.vscale.i64()
; INLOOP-NEXT: [[TMP14:%.*]] = mul i64 [[TMP13]], 4
; INLOOP-NEXT: [[TMP15:%.*]] = getelementptr inbounds i16, ptr [[TMP10]], i64 [[TMP14]]
; INLOOP-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 4 x i16>, ptr [[TMP15]], align 2
; INLOOP-NEXT: [[TMP16:%.*]] = sext <vscale x 4 x i16> [[WIDE_LOAD]] to <vscale x 4 x i32>
; INLOOP-NEXT: [[TMP17:%.*]] = sext <vscale x 4 x i16> [[WIDE_LOAD2]] to <vscale x 4 x i32>
; INLOOP-NEXT: [[TMP18:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP16]])
; INLOOP-NEXT: [[TMP19]] = add i32 [[TMP18]], [[VEC_PHI]]
; INLOOP-NEXT: [[TMP20:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP17]])
; INLOOP-NEXT: [[TMP21]] = add i32 [[TMP20]], [[VEC_PHI1]]
; INLOOP-NEXT: [[TMP22:%.*]] = call i32 @llvm.vscale.i32()
; INLOOP-NEXT: [[TMP23:%.*]] = mul i32 [[TMP22]], 8
; INLOOP-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], [[TMP23]]
; INLOOP-NEXT: [[TMP24:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; INLOOP-NEXT: br i1 [[TMP24]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; INLOOP: middle.block:
; INLOOP-NEXT: [[BIN_RDX:%.*]] = add i32 [[TMP21]], [[TMP19]]
; INLOOP-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
; INLOOP-NEXT: br i1 [[CMP_N]], label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; INLOOP: scalar.ph:
; INLOOP-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; INLOOP-NEXT: [[BC_MERGE_RDX:%.*]] = phi i32 [ 0, [[FOR_BODY_PREHEADER]] ], [ [[BIN_RDX]], [[MIDDLE_BLOCK]] ]
; INLOOP-NEXT: br label [[FOR_BODY:%.*]]
; INLOOP: for.body:
; INLOOP-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INLOOP-NEXT: [[R_07:%.*]] = phi i32 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; INLOOP-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_08]]
; INLOOP-NEXT: [[TMP25:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; INLOOP-NEXT: [[CONV:%.*]] = sext i16 [[TMP25]] to i32
; INLOOP-NEXT: [[ADD]] = add nsw i32 [[R_07]], [[CONV]]
; INLOOP-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; INLOOP-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], [[N]]
; INLOOP-NEXT: br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; INLOOP: for.cond.cleanup.loopexit:
; INLOOP-NEXT: [[ADD_LCSSA:%.*]] = phi i32 [ [[ADD]], [[FOR_BODY]] ], [ [[BIN_RDX]], [[MIDDLE_BLOCK]] ]
; INLOOP-NEXT: br label [[FOR_COND_CLEANUP]]
; INLOOP: for.cond.cleanup:
; INLOOP-NEXT: [[R_0_LCSSA:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[ADD_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT]] ]
; INLOOP-NEXT: ret i32 [[R_0_LCSSA]]
;
entry:
%cmp6 = icmp sgt i32 %n, 0
br i1 %cmp6, label %for.body, label %for.cond.cleanup
for.body: ; preds = %entry, %for.body
%i.08 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
%r.07 = phi i32 [ %add, %for.body ], [ 0, %entry ]
%arrayidx = getelementptr inbounds i16, ptr %x, i32 %i.08
%0 = load i16, ptr %arrayidx, align 2
%conv = sext i16 %0 to i32
%add = add nsw i32 %r.07, %conv
%inc = add nuw nsw i32 %i.08, 1
%exitcond = icmp eq i32 %inc, %n
br i1 %exitcond, label %for.cond.cleanup, label %for.body
for.cond.cleanup: ; preds = %for.body, %entry
%r.0.lcssa = phi i32 [ 0, %entry ], [ %add, %for.body ]
ret i32 %r.0.lcssa
}
Reference in New Issue View Git Blame Copy Permalink