shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/scalable-basics.ll
Luke Lau 8d16c6809a [RISCV] Increase default vectorizer LMUL to 2 
After some discussion and experimentation, we have seen that changing the default number of vector register bits to LMUL=2 strikes a sweet spot.
Whilst we could be clever here and make the vectorizer smarter about dynamically selecting an LMUL that
a) Doesn't affect register pressure
b) Suitable for the microarchitecture
we would need to teach its heuristics about RISC-V register grouping specifics.
Instead this just does the easy, pragmatic thing by changing the default to a safe value that doesn't affect register pressure signifcantly[1], but should increase throughput and unlock more interleaving.

[1] Register spilling when compiling sqlite at various levels of `-riscv-v-register-bit-width-lmul`:

LMUL=1    2573 spills
LMUL=2    2583 spills
LMUL=4    2819 spills
LMUL=8    3256 spills

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D143723
2023-03-23 10:33:50 +00:00

716 lines
42 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=loop-vectorize -scalable-vectorization=on -mtriple riscv64-linux-gnu -mattr=+v,+f -S 2>%t | FileCheck %s -check-prefix=VLENUNK
; RUN: opt < %s -passes=loop-vectorize -scalable-vectorization=on -riscv-v-vector-bits-min=-1 -mtriple riscv64-linux-gnu -mattr=+v,+f -S 2>%t | FileCheck %s -check-prefix=VLEN128
; Note: +v implies a Zvl128b (i.e. minimal VLEN of 128), but as can be seen, we're currently
; not using that information unless an explicit vector width is set. (FIXME)
; A collection of fairly basic functional tests when both fixed and scalable vectorization is
; allowed. The primary goal of this is check for crashes during cost modeling, but it also
; exercises the default heuristics in a useful way.
target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n64-S128"
target triple = "riscv64"
define void @vector_add(ptr noalias nocapture %a, i64 %v, i64 %n) {
; VLENUNK-LABEL: @vector_add(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLENUNK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLENUNK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLENUNK: vector.ph:
; VLENUNK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLENUNK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLENUNK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLENUNK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[V:%.*]], i64 0
; VLENUNK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; VLENUNK-NEXT: br label [[VECTOR_BODY:%.*]]
; VLENUNK: vector.body:
; VLENUNK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLENUNK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP4]]
; VLENUNK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLENUNK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP6]], align 8
; VLENUNK-NEXT: [[TMP7:%.*]] = add <vscale x 2 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; VLENUNK-NEXT: store <vscale x 2 x i64> [[TMP7]], ptr [[TMP6]], align 8
; VLENUNK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 2
; VLENUNK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; VLENUNK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLENUNK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; VLENUNK: middle.block:
; VLENUNK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLENUNK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLENUNK: scalar.ph:
; VLENUNK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; VLENUNK-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; VLENUNK-NEXT: [[ADD:%.*]] = add i64 [[ELEM]], [[V]]
; VLENUNK-NEXT: store i64 [[ADD]], ptr [[ARRAYIDX]], align 8
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; VLENUNK: for.end:
; VLENUNK-NEXT: ret void
;
; VLEN128-LABEL: @vector_add(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLEN128-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLEN128-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLEN128: vector.ph:
; VLEN128-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLEN128-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLEN128-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLEN128-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[V:%.*]], i64 0
; VLEN128-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; VLEN128-NEXT: br label [[VECTOR_BODY:%.*]]
; VLEN128: vector.body:
; VLEN128-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLEN128-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP4]]
; VLEN128-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLEN128-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP6]], align 8
; VLEN128-NEXT: [[TMP7:%.*]] = add <vscale x 2 x i64> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; VLEN128-NEXT: store <vscale x 2 x i64> [[TMP7]], ptr [[TMP6]], align 8
; VLEN128-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 2
; VLEN128-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; VLEN128-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLEN128-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; VLEN128: middle.block:
; VLEN128-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLEN128-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLEN128: scalar.ph:
; VLEN128-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; VLEN128-NEXT: [[ELEM:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; VLEN128-NEXT: [[ADD:%.*]] = add i64 [[ELEM]], [[V]]
; VLEN128-NEXT: store i64 [[ADD]], ptr [[ARRAYIDX]], align 8
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; VLEN128: for.end:
; VLEN128-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
%elem = load i64, ptr %arrayidx
%add = add i64 %elem, %v
store i64 %add, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
; Same as above, but with op type of i32. We currently have a bug around
; etype=ELEN profitability in the vectorizer, and having a smaller element
; width test allows us to highlight different aspects of codegen.
define void @vector_add_i32(ptr noalias nocapture %a, i32 %v, i64 %n) {
; VLENUNK-LABEL: @vector_add_i32(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; VLENUNK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLENUNK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLENUNK: vector.ph:
; VLENUNK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; VLENUNK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLENUNK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLENUNK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[V:%.*]], i64 0
; VLENUNK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; VLENUNK-NEXT: br label [[VECTOR_BODY:%.*]]
; VLENUNK: vector.body:
; VLENUNK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLENUNK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP4]]
; VLENUNK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
; VLENUNK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 4
; VLENUNK-NEXT: [[TMP7:%.*]] = add <vscale x 4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; VLENUNK-NEXT: store <vscale x 4 x i32> [[TMP7]], ptr [[TMP6]], align 4
; VLENUNK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 4
; VLENUNK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; VLENUNK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLENUNK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; VLENUNK: middle.block:
; VLENUNK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLENUNK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLENUNK: scalar.ph:
; VLENUNK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
; VLENUNK-NEXT: [[ELEM:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; VLENUNK-NEXT: [[ADD:%.*]] = add i32 [[ELEM]], [[V]]
; VLENUNK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; VLENUNK: for.end:
; VLENUNK-NEXT: ret void
;
; VLEN128-LABEL: @vector_add_i32(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; VLEN128-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLEN128-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLEN128: vector.ph:
; VLEN128-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; VLEN128-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLEN128-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLEN128-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[V:%.*]], i64 0
; VLEN128-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; VLEN128-NEXT: br label [[VECTOR_BODY:%.*]]
; VLEN128: vector.body:
; VLEN128-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLEN128-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP4]]
; VLEN128-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
; VLEN128-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 4
; VLEN128-NEXT: [[TMP7:%.*]] = add <vscale x 4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; VLEN128-NEXT: store <vscale x 4 x i32> [[TMP7]], ptr [[TMP6]], align 4
; VLEN128-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 4
; VLEN128-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; VLEN128-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLEN128-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; VLEN128: middle.block:
; VLEN128-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLEN128-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLEN128: scalar.ph:
; VLEN128-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
; VLEN128-NEXT: [[ELEM:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; VLEN128-NEXT: [[ADD:%.*]] = add i32 [[ELEM]], [[V]]
; VLEN128-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; VLEN128: for.end:
; VLEN128-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i32, ptr %a, i64 %iv
%elem = load i32, ptr %arrayidx
%add = add i32 %elem, %v
store i32 %add, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
; a[b[i]] += v, mostly to exercise scatter/gather costing
; TODO: Currently fails to vectorize due to a memory conflict
define void @indexed_add(ptr noalias nocapture %a, ptr noalias nocapture %b, i64 %v, i64 %n) {
; VLENUNK-LABEL: @indexed_add(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[BADDR:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[IV]]
; VLENUNK-NEXT: [[AIDX:%.*]] = load i64, ptr [[BADDR]], align 8
; VLENUNK-NEXT: [[AADDR:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[AIDX]]
; VLENUNK-NEXT: [[ELEM:%.*]] = load i64, ptr [[AADDR]], align 8
; VLENUNK-NEXT: [[ADD:%.*]] = add i64 [[ELEM]], [[V:%.*]]
; VLENUNK-NEXT: store i64 [[ADD]], ptr [[AADDR]], align 8
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]]
; VLENUNK: for.end:
; VLENUNK-NEXT: ret void
;
; VLEN128-LABEL: @indexed_add(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[BADDR:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[IV]]
; VLEN128-NEXT: [[AIDX:%.*]] = load i64, ptr [[BADDR]], align 8
; VLEN128-NEXT: [[AADDR:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[AIDX]]
; VLEN128-NEXT: [[ELEM:%.*]] = load i64, ptr [[AADDR]], align 8
; VLEN128-NEXT: [[ADD:%.*]] = add i64 [[ELEM]], [[V:%.*]]
; VLEN128-NEXT: store i64 [[ADD]], ptr [[AADDR]], align 8
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]]
; VLEN128: for.end:
; VLEN128-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%baddr = getelementptr inbounds i64, ptr %b, i64 %iv
%aidx = load i64, ptr %baddr
%aaddr = getelementptr inbounds i64, ptr %a, i64 %aidx
%elem = load i64, ptr %aaddr
%add = add i64 %elem, %v
store i64 %add, ptr %aaddr
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
; a[b[i]] = v, exercise scatter support
define void @indexed_store(ptr noalias nocapture %a, ptr noalias nocapture %b, i64 %v, i64 %n) {
; VLENUNK-LABEL: @indexed_store(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLENUNK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLENUNK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLENUNK: vector.ph:
; VLENUNK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLENUNK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLENUNK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLENUNK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[V:%.*]], i64 0
; VLENUNK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; VLENUNK-NEXT: br label [[VECTOR_BODY:%.*]]
; VLENUNK: vector.body:
; VLENUNK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLENUNK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[TMP4]]
; VLENUNK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLENUNK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP6]], align 8
; VLENUNK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], <vscale x 2 x i64> [[WIDE_LOAD]]
; VLENUNK-NEXT: call void @llvm.masked.scatter.nxv2i64.nxv2p0(<vscale x 2 x i64> [[BROADCAST_SPLAT]], <vscale x 2 x ptr> [[TMP7]], i32 8, <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
; VLENUNK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 2
; VLENUNK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; VLENUNK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLENUNK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; VLENUNK: middle.block:
; VLENUNK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLENUNK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLENUNK: scalar.ph:
; VLENUNK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[BADDR:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
; VLENUNK-NEXT: [[AIDX:%.*]] = load i64, ptr [[BADDR]], align 8
; VLENUNK-NEXT: [[AADDR:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[AIDX]]
; VLENUNK-NEXT: store i64 [[V]], ptr [[AADDR]], align 8
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; VLENUNK: for.end:
; VLENUNK-NEXT: ret void
;
; VLEN128-LABEL: @indexed_store(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLEN128-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLEN128-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLEN128: vector.ph:
; VLEN128-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLEN128-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLEN128-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLEN128-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[V:%.*]], i64 0
; VLEN128-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; VLEN128-NEXT: br label [[VECTOR_BODY:%.*]]
; VLEN128: vector.body:
; VLEN128-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLEN128-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[TMP4]]
; VLEN128-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLEN128-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP6]], align 8
; VLEN128-NEXT: [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], <vscale x 2 x i64> [[WIDE_LOAD]]
; VLEN128-NEXT: call void @llvm.masked.scatter.nxv2i64.nxv2p0(<vscale x 2 x i64> [[BROADCAST_SPLAT]], <vscale x 2 x ptr> [[TMP7]], i32 8, <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
; VLEN128-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP9:%.*]] = mul i64 [[TMP8]], 2
; VLEN128-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
; VLEN128-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLEN128-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; VLEN128: middle.block:
; VLEN128-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLEN128-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLEN128: scalar.ph:
; VLEN128-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[BADDR:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
; VLEN128-NEXT: [[AIDX:%.*]] = load i64, ptr [[BADDR]], align 8
; VLEN128-NEXT: [[AADDR:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[AIDX]]
; VLEN128-NEXT: store i64 [[V]], ptr [[AADDR]], align 8
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; VLEN128: for.end:
; VLEN128-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%baddr = getelementptr inbounds i64, ptr %b, i64 %iv
%aidx = load i64, ptr %baddr
%aaddr = getelementptr inbounds i64, ptr %a, i64 %aidx
store i64 %v, ptr %aaddr
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define i64 @indexed_load(ptr noalias nocapture %a, ptr noalias nocapture %b, i64 %v, i64 %n) {
; VLENUNK-LABEL: @indexed_load(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLENUNK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLENUNK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLENUNK: vector.ph:
; VLENUNK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLENUNK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLENUNK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLENUNK-NEXT: br label [[VECTOR_BODY:%.*]]
; VLENUNK: vector.body:
; VLENUNK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLENUNK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[TMP4]]
; VLENUNK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLENUNK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP6]], align 8
; VLENUNK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], <vscale x 2 x i64> [[WIDE_LOAD]]
; VLENUNK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 2 x i64> @llvm.masked.gather.nxv2i64.nxv2p0(<vscale x 2 x ptr> [[TMP7]], i32 8, <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer), <vscale x 2 x i64> poison)
; VLENUNK-NEXT: [[TMP8]] = add <vscale x 2 x i64> [[VEC_PHI]], [[WIDE_MASKED_GATHER]]
; VLENUNK-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP10:%.*]] = mul i64 [[TMP9]], 2
; VLENUNK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP10]]
; VLENUNK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLENUNK-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; VLENUNK: middle.block:
; VLENUNK-NEXT: [[TMP12:%.*]] = call i64 @llvm.vector.reduce.add.nxv2i64(<vscale x 2 x i64> [[TMP8]])
; VLENUNK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLENUNK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLENUNK: scalar.ph:
; VLENUNK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLENUNK-NEXT: [[BC_MERGE_RDX:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[TMP12]], [[MIDDLE_BLOCK]] ]
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[SUM:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[SUM_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[BADDR:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
; VLENUNK-NEXT: [[AIDX:%.*]] = load i64, ptr [[BADDR]], align 8
; VLENUNK-NEXT: [[AADDR:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[AIDX]]
; VLENUNK-NEXT: [[ELEM:%.*]] = load i64, ptr [[AADDR]], align 8
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[SUM_NEXT]] = add i64 [[SUM]], [[ELEM]]
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; VLENUNK: for.end:
; VLENUNK-NEXT: [[SUM_NEXT_LCSSA:%.*]] = phi i64 [ [[SUM_NEXT]], [[FOR_BODY]] ], [ [[TMP12]], [[MIDDLE_BLOCK]] ]
; VLENUNK-NEXT: ret i64 [[SUM_NEXT_LCSSA]]
;
; VLEN128-LABEL: @indexed_load(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLEN128-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLEN128-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLEN128: vector.ph:
; VLEN128-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLEN128-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLEN128-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLEN128-NEXT: br label [[VECTOR_BODY:%.*]]
; VLEN128: vector.body:
; VLEN128-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLEN128-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[TMP4]]
; VLEN128-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLEN128-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP6]], align 8
; VLEN128-NEXT: [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], <vscale x 2 x i64> [[WIDE_LOAD]]
; VLEN128-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 2 x i64> @llvm.masked.gather.nxv2i64.nxv2p0(<vscale x 2 x ptr> [[TMP7]], i32 8, <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer), <vscale x 2 x i64> poison)
; VLEN128-NEXT: [[TMP8]] = add <vscale x 2 x i64> [[VEC_PHI]], [[WIDE_MASKED_GATHER]]
; VLEN128-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP10:%.*]] = mul i64 [[TMP9]], 2
; VLEN128-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP10]]
; VLEN128-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLEN128-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; VLEN128: middle.block:
; VLEN128-NEXT: [[TMP12:%.*]] = call i64 @llvm.vector.reduce.add.nxv2i64(<vscale x 2 x i64> [[TMP8]])
; VLEN128-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLEN128-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLEN128: scalar.ph:
; VLEN128-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLEN128-NEXT: [[BC_MERGE_RDX:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[TMP12]], [[MIDDLE_BLOCK]] ]
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[SUM:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[SUM_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[BADDR:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
; VLEN128-NEXT: [[AIDX:%.*]] = load i64, ptr [[BADDR]], align 8
; VLEN128-NEXT: [[AADDR:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[AIDX]]
; VLEN128-NEXT: [[ELEM:%.*]] = load i64, ptr [[AADDR]], align 8
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[SUM_NEXT]] = add i64 [[SUM]], [[ELEM]]
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; VLEN128: for.end:
; VLEN128-NEXT: [[SUM_NEXT_LCSSA:%.*]] = phi i64 [ [[SUM_NEXT]], [[FOR_BODY]] ], [ [[TMP12]], [[MIDDLE_BLOCK]] ]
; VLEN128-NEXT: ret i64 [[SUM_NEXT_LCSSA]]
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%sum = phi i64 [0, %entry], [%sum.next, %for.body]
%baddr = getelementptr inbounds i64, ptr %b, i64 %iv
%aidx = load i64, ptr %baddr
%aaddr = getelementptr inbounds i64, ptr %a, i64 %aidx
%elem = load i64, ptr %aaddr
%iv.next = add nuw nsw i64 %iv, 1
%sum.next = add i64 %sum, %elem
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret i64 %sum.next
}
define void @splat_int(ptr noalias nocapture %a, i64 %v, i64 %n) {
; VLENUNK-LABEL: @splat_int(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLENUNK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLENUNK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLENUNK: vector.ph:
; VLENUNK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLENUNK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLENUNK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLENUNK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[V:%.*]], i64 0
; VLENUNK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; VLENUNK-NEXT: br label [[VECTOR_BODY:%.*]]
; VLENUNK: vector.body:
; VLENUNK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLENUNK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP4]]
; VLENUNK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLENUNK-NEXT: store <vscale x 2 x i64> [[BROADCAST_SPLAT]], ptr [[TMP6]], align 8
; VLENUNK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 2
; VLENUNK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; VLENUNK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLENUNK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; VLENUNK: middle.block:
; VLENUNK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLENUNK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLENUNK: scalar.ph:
; VLENUNK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; VLENUNK-NEXT: store i64 [[V]], ptr [[ARRAYIDX]], align 8
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; VLENUNK: for.end:
; VLENUNK-NEXT: ret void
;
; VLEN128-LABEL: @splat_int(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLEN128-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLEN128-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLEN128: vector.ph:
; VLEN128-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLEN128-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLEN128-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLEN128-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[V:%.*]], i64 0
; VLEN128-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; VLEN128-NEXT: br label [[VECTOR_BODY:%.*]]
; VLEN128: vector.body:
; VLEN128-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLEN128-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP4]]
; VLEN128-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
; VLEN128-NEXT: store <vscale x 2 x i64> [[BROADCAST_SPLAT]], ptr [[TMP6]], align 8
; VLEN128-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 2
; VLEN128-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; VLEN128-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLEN128-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; VLEN128: middle.block:
; VLEN128-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLEN128-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLEN128: scalar.ph:
; VLEN128-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; VLEN128-NEXT: store i64 [[V]], ptr [[ARRAYIDX]], align 8
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; VLEN128: for.end:
; VLEN128-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
store i64 %v, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
define void @splat_ptr(ptr noalias nocapture %a, ptr %v, i64 %n) {
; VLENUNK-LABEL: @splat_ptr(
; VLENUNK-NEXT: entry:
; VLENUNK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLENUNK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLENUNK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLENUNK: vector.ph:
; VLENUNK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLENUNK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLENUNK-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLENUNK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x ptr> poison, ptr [[V:%.*]], i64 0
; VLENUNK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x ptr> [[BROADCAST_SPLATINSERT]], <vscale x 2 x ptr> poison, <vscale x 2 x i32> zeroinitializer
; VLENUNK-NEXT: br label [[VECTOR_BODY:%.*]]
; VLENUNK: vector.body:
; VLENUNK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLENUNK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLENUNK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP4]]
; VLENUNK-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[TMP5]], i32 0
; VLENUNK-NEXT: store <vscale x 2 x ptr> [[BROADCAST_SPLAT]], ptr [[TMP6]], align 8
; VLENUNK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
; VLENUNK-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 2
; VLENUNK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; VLENUNK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLENUNK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; VLENUNK: middle.block:
; VLENUNK-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLENUNK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLENUNK: scalar.ph:
; VLENUNK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLENUNK-NEXT: br label [[FOR_BODY:%.*]]
; VLENUNK: for.body:
; VLENUNK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLENUNK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; VLENUNK-NEXT: store ptr [[V]], ptr [[ARRAYIDX]], align 8
; VLENUNK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLENUNK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLENUNK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; VLENUNK: for.end:
; VLENUNK-NEXT: ret void
;
; VLEN128-LABEL: @splat_ptr(
; VLEN128-NEXT: entry:
; VLEN128-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP1:%.*]] = mul i64 [[TMP0]], 2
; VLEN128-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1024, [[TMP1]]
; VLEN128-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; VLEN128: vector.ph:
; VLEN128-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 2
; VLEN128-NEXT: [[N_MOD_VF:%.*]] = urem i64 1024, [[TMP3]]
; VLEN128-NEXT: [[N_VEC:%.*]] = sub i64 1024, [[N_MOD_VF]]
; VLEN128-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x ptr> poison, ptr [[V:%.*]], i64 0
; VLEN128-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x ptr> [[BROADCAST_SPLATINSERT]], <vscale x 2 x ptr> poison, <vscale x 2 x i32> zeroinitializer
; VLEN128-NEXT: br label [[VECTOR_BODY:%.*]]
; VLEN128: vector.body:
; VLEN128-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VLEN128-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; VLEN128-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[TMP4]]
; VLEN128-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[TMP5]], i32 0
; VLEN128-NEXT: store <vscale x 2 x ptr> [[BROADCAST_SPLAT]], ptr [[TMP6]], align 8
; VLEN128-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
; VLEN128-NEXT: [[TMP8:%.*]] = mul i64 [[TMP7]], 2
; VLEN128-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; VLEN128-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; VLEN128-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; VLEN128: middle.block:
; VLEN128-NEXT: [[CMP_N:%.*]] = icmp eq i64 1024, [[N_VEC]]
; VLEN128-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VLEN128: scalar.ph:
; VLEN128-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VLEN128-NEXT: br label [[FOR_BODY:%.*]]
; VLEN128: for.body:
; VLEN128-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; VLEN128-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
; VLEN128-NEXT: store ptr [[V]], ptr [[ARRAYIDX]], align 8
; VLEN128-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; VLEN128-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 1024
; VLEN128-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; VLEN128: for.end:
; VLEN128-NEXT: ret void
;
entry:
br label %for.body
for.body:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
%arrayidx = getelementptr inbounds i64, ptr %a, i64 %iv
store ptr %v, ptr %arrayidx
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 1024
br i1 %exitcond.not, label %for.end, label %for.body
for.end:
ret void
}
Reference in New Issue View Git Blame Copy Permalink