File: sink-splat-operands-i1.ll

package info (click to toggle)
llvm-toolchain-17 1%3A17.0.6-22
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 1,799,624 kB
  • sloc: cpp: 6,428,607; ansic: 1,383,196; asm: 793,408; python: 223,504; objc: 75,364; f90: 60,502; lisp: 33,869; pascal: 15,282; sh: 9,684; perl: 7,453; ml: 4,937; awk: 3,523; makefile: 2,889; javascript: 2,149; xml: 888; fortran: 619; cs: 573
file content (45 lines) | stat: -rw-r--r-- 1,923 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+v,+f -target-abi=lp64f \
; RUN:     -disable-machine-licm | FileCheck %s

; Make sure we don't unnecessrily sink i1 vector splats.

declare <8 x i1> @llvm.vp.and.v4i1(<8 x i1>, <8 x i1>, <8 x i1>, i32)

define void @sink_splat_vp_and_i1(ptr nocapture %a, i1 zeroext %x, <8 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: sink_splat_vp_and_i1:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
; CHECK-NEXT:    vmv.v.x v8, a1
; CHECK-NEXT:    vmsne.vi v8, v8, 0
; CHECK-NEXT:    li a1, 1024
; CHECK-NEXT:  .LBB0_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vlm.v v9, (a0)
; CHECK-NEXT:    vsetvli zero, a2, e8, mf2, ta, ma
; CHECK-NEXT:    vmand.mm v9, v9, v8
; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
; CHECK-NEXT:    vsm.v v9, (a0)
; CHECK-NEXT:    addi a1, a1, -1
; CHECK-NEXT:    addi a0, a0, 1
; CHECK-NEXT:    bnez a1, .LBB0_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  %broadcast.splatinsert = insertelement <8 x i1> poison, i1 %x, i32 0
  %broadcast.splat = shufflevector <8 x i1> %broadcast.splatinsert, <8 x i1> poison, <8 x i32> zeroinitializer
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds i8, ptr %a, i64 %index
  %wide.load = load <8 x i1>, ptr %0, align 4
  %1 = call <8 x i1> @llvm.vp.and.v4i1(<8 x i1> %wide.load, <8 x i1> %broadcast.splat, <8 x i1> %m, i32 %vl)
  store <8 x i1> %1, ptr %0, align 1
  %index.next = add nuw i64 %index, 1
  %2 = icmp eq i64 %index.next, 1024
  br i1 %2, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}