e6fb9ba3f2
This patch adds a case in isTriviallyVectorizable in llvm/lib/Analysis/VectorUtils.cpp for Intrinsic::clmul. Fixes #179106
51 lines
1.7 KiB
LLVM
51 lines
1.7 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: opt -passes=loop-vectorize -mtriple=riscv64 -mattr=+v -S %s | FileCheck %s
|
|
|
|
declare i64 @llvm.clmul.i64(i64 %a, i64 %b)
|
|
|
|
define void @clmul_loop(ptr %a, ptr %b, ptr %c, i64 %n) {
|
|
; CHECK-LABEL: define void @clmul_loop(
|
|
; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], ptr [[C:%.*]], i64 [[N:%.*]]) #[[ATTR1:[0-9]+]] {
|
|
; CHECK-NEXT: [[ENTRY:.*]]:
|
|
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
|
|
; CHECK: [[FOR_BODY]]:
|
|
; CHECK-NEXT: [[I:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[I_NEXT:%.*]], %[[FOR_BODY]] ]
|
|
; CHECK-NEXT: [[PA:%.*]] = getelementptr i64, ptr [[A]], i64 [[I]]
|
|
; CHECK-NEXT: [[PB:%.*]] = getelementptr i64, ptr [[B]], i64 [[I]]
|
|
; CHECK-NEXT: [[PC:%.*]] = getelementptr i64, ptr [[C]], i64 [[I]]
|
|
; CHECK-NEXT: [[VA:%.*]] = load i64, ptr [[PA]], align 8
|
|
; CHECK-NEXT: [[VB:%.*]] = load i64, ptr [[PB]], align 8
|
|
; CHECK-NEXT: [[R:%.*]] = call i64 @llvm.clmul.i64(i64 [[VA]], i64 [[VB]])
|
|
; CHECK-NEXT: store i64 [[R]], ptr [[PC]], align 8
|
|
; CHECK-NEXT: [[I_NEXT]] = add i64 [[I]], 1
|
|
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[I_NEXT]], [[N]]
|
|
; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_EXIT:.*]], label %[[FOR_BODY]]
|
|
; CHECK: [[FOR_EXIT]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %for.body
|
|
|
|
for.body:
|
|
%i = phi i64 [0, %entry], [%i.next, %for.body]
|
|
|
|
%pa = getelementptr i64, ptr %a, i64 %i
|
|
%pb = getelementptr i64, ptr %b, i64 %i
|
|
%pc = getelementptr i64, ptr %c, i64 %i
|
|
|
|
%va = load i64, ptr %pa
|
|
%vb = load i64, ptr %pb
|
|
|
|
%r = call i64 @llvm.clmul.i64(i64 %va, i64 %vb)
|
|
|
|
store i64 %r, ptr %pc
|
|
|
|
%i.next = add i64 %i, 1
|
|
%cmp = icmp eq i64 %i.next, %n
|
|
br i1 %cmp, label %for.exit, label %for.body
|
|
|
|
for.exit:
|
|
ret void
|
|
|
|
}
|