mirror of
https://github.com/RPCS3/llvm.git
synced 2024-11-24 12:20:00 +00:00
49683f3c96
The new target machines are: nvptx (old ptx32) => 32-bit PTX nvptx64 (old ptx64) => 64-bit PTX The sources are based on the internal NVIDIA NVPTX back-end, and contain more functionality than the current PTX back-end currently provides. NV_CONTRIB git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@156196 91177308-0d34-0410-b5e6-96231b3b80d8
25 lines
785 B
LLVM
25 lines
785 B
LLVM
; RUN: llc < %s -march=nvptx -mcpu=sm_20 -nvptx-fma-level=1 | FileCheck %s -check-prefix=FMA
|
|
; RUN: llc < %s -march=nvptx -mcpu=sm_20 -nvptx-fma-level=0 | FileCheck %s -check-prefix=MUL
|
|
; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 -nvptx-fma-level=1 | FileCheck %s -check-prefix=FMA
|
|
; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 -nvptx-fma-level=0 | FileCheck %s -check-prefix=MUL
|
|
|
|
define ptx_device float @test_mul_add_f(float %x, float %y, float %z) {
|
|
entry:
|
|
; FMA: fma.rn.f32
|
|
; MUL: mul.rn.f32
|
|
; MUL: add.rn.f32
|
|
%a = fmul float %x, %y
|
|
%b = fadd float %a, %z
|
|
ret float %b
|
|
}
|
|
|
|
define ptx_device double @test_mul_add_d(double %x, double %y, double %z) {
|
|
entry:
|
|
; FMA: fma.rn.f64
|
|
; MUL: mul.rn.f64
|
|
; MUL: add.rn.f64
|
|
%a = fmul double %x, %y
|
|
%b = fadd double %a, %z
|
|
ret double %b
|
|
}
|