File: mad_int24.ll

package info (click to toggle)
llvm-toolchain-7 1%3A7.0.1-8
  • links: PTS, VCS
  • area: main
  • in suites: buster
  • size: 733,456 kB
  • sloc: cpp: 3,776,651; ansic: 633,271; asm: 350,301; python: 142,716; objc: 107,612; sh: 22,626; lisp: 11,056; perl: 7,999; pascal: 6,742; ml: 5,537; awk: 3,536; makefile: 2,557; cs: 2,027; xml: 841; ruby: 156
file content (24 lines) | stat: -rw-r--r-- 941 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
; RUN: llc < %s -march=amdgcn -verify-machineinstrs | FileCheck %s --check-prefix=SI --check-prefix=FUNC
; RUN: llc < %s -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs | FileCheck %s --check-prefix=SI --check-prefix=FUNC
; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s --check-prefix=EG --check-prefix=FUNC
; RUN: llc < %s -march=r600 -mcpu=cayman | FileCheck %s --check-prefix=CM --check-prefix=FUNC

; FUNC-LABEL: {{^}}i32_mad24:
; Signed 24-bit multiply is not supported on pre-Cayman GPUs.
; EG: MULLO_INT
; Make sure we aren't masking the inputs.
; CM-NOT: AND
; CM: MULADD_INT24
; SI-NOT: and
; SI: v_mad_i32_i24
define amdgpu_kernel void @i32_mad24(i32 addrspace(1)* %out, i32 %a, i32 %b, i32 %c) {
entry:
  %0 = shl i32 %a, 8
  %a_24 = ashr i32 %0, 8
  %1 = shl i32 %b, 8
  %b_24 = ashr i32 %1, 8
  %2 = mul i32 %a_24, %b_24
  %3 = add i32 %2, %c
  store i32 %3, i32 addrspace(1)* %out
  ret void
}