File: amx-tf32-internal.ll

package info (click to toggle)
llvm-toolchain-20 1%3A20.1.6-1~exp1
  • links: PTS, VCS
  • area: main
  • in suites: experimental
  • size: 2,111,304 kB
  • sloc: cpp: 7,438,677; ansic: 1,393,822; asm: 1,012,926; python: 241,650; f90: 86,635; objc: 75,479; lisp: 42,144; pascal: 17,286; sh: 10,027; ml: 5,082; perl: 4,730; awk: 3,523; makefile: 3,349; javascript: 2,251; xml: 892; fortran: 672
file content (46 lines) | stat: -rw-r--r-- 2,079 bytes parent folder | download | duplicates (6)
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
46
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+amx-tile,+avx512f, \
; RUN: -mattr=+amx-tf32,+amx-transpose -verify-machineinstrs | FileCheck %s

define void @test_amx(i8* %pointer, i8* %base, i64 %stride) {
; CHECK-LABEL: test_amx:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vxorps %xmm0, %xmm0, %xmm0
; CHECK-NEXT:    vmovups %zmm0, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movb $1, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movb $8, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movw $8, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movb $8, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movw $8, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movb $8, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movw $8, -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    ldtilecfg -{{[0-9]+}}(%rsp)
; CHECK-NEXT:    movw $8, %ax
; CHECK-NEXT:    tileloadd (%rsi,%rdx), %tmm0
; CHECK-NEXT:    tilezero %tmm1
; CHECK-NEXT:    tilezero %tmm2
; CHECK-NEXT:    tmmultf32ps %tmm1, %tmm0, %tmm2
; CHECK-NEXT:    ttmmultf32ps %tmm1, %tmm0, %tmm2
; CHECK-NEXT:    tilestored %tmm2, (%rdi,%rdx)
; CHECK-NEXT:    tilerelease
; CHECK-NEXT:    vzeroupper
; CHECK-NEXT:    retq

  %a = call x86_amx @llvm.x86.tileloadd64.internal(i16 8, i16 8, i8* %base, i64 %stride)
  %b = call x86_amx @llvm.x86.tilezero.internal(i16 8, i16 8)
  %c = call x86_amx @llvm.x86.tilezero.internal(i16 8, i16 8)

  %c1 = call x86_amx @llvm.x86.tmmultf32ps.internal(i16 8, i16 8, i16 8, x86_amx %c, x86_amx %a, x86_amx %b)
  %c2 = call x86_amx @llvm.x86.ttmmultf32ps.internal(i16 8, i16 8, i16 8, x86_amx %c1, x86_amx %a, x86_amx %b)

  call void @llvm.x86.tilestored64.internal(i16 8, i16 8, i8* %pointer, i64 %stride, x86_amx %c2)
  ret void
}

declare x86_amx @llvm.x86.tilezero.internal(i16, i16)
declare x86_amx @llvm.x86.tileloadd64.internal(i16, i16, i8*, i64)
declare void @llvm.x86.tilestored64.internal(i16, i16, i8*, i64, x86_amx)


declare x86_amx @llvm.x86.tmmultf32ps.internal(i16, i16, i16, x86_amx, x86_amx, x86_amx)
declare x86_amx @llvm.x86.ttmmultf32ps.internal(i16, i16, i16, x86_amx, x86_amx, x86_amx)