83
|
1 ; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
|
|
2 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=VI -check-prefix=FUNC %s
|
77
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
3
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
4 declare double @llvm.AMDGPU.rsq.clamped.f64(double) nounwind readnone
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
5
|
83
|
6 ; FUNC-LABEL: {{^}}rsq_clamped_f64:
|
|
7 ; SI: v_rsq_clamp_f64_e32
|
|
8
|
|
9 ; VI: v_rsq_f64_e32 [[RSQ:v\[[0-9]+:[0-9]+\]]], s[2:3]
|
|
10 ; TODO: this constant should be folded:
|
|
11 ; VI: s_mov_b32 s[[ALLBITS:[0-9+]]], -1
|
|
12 ; VI: s_mov_b32 s[[HIGH1:[0-9+]]], 0x7fefffff
|
|
13 ; VI: s_mov_b32 s[[LOW1:[0-9+]]], s[[ALLBITS]]
|
|
14 ; VI: v_min_f64 v[0:1], [[RSQ]], s{{\[}}[[LOW1]]:[[HIGH1]]]
|
|
15 ; VI: s_mov_b32 s[[HIGH2:[0-9+]]], 0xffefffff
|
|
16 ; VI: s_mov_b32 s[[LOW2:[0-9+]]], s[[ALLBITS]]
|
|
17 ; VI: v_max_f64 v[0:1], v[0:1], s{{\[}}[[LOW2]]:[[HIGH2]]]
|
|
18
|
77
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
19 define void @rsq_clamped_f64(double addrspace(1)* %out, double %src) nounwind {
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
20 %rsq_clamped = call double @llvm.AMDGPU.rsq.clamped.f64(double %src) nounwind readnone
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
21 store double %rsq_clamped, double addrspace(1)* %out, align 8
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
22 ret void
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
23 }
|