xref: /aosp_15_r20/external/llvm/test/CodeGen/AMDGPU/llvm.AMDGPU.rsq.clamped.f64.ll (revision 9880d6810fe72a1726cb53787c6711e909410d58)
1*9880d681SAndroid Build Coastguard Worker; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
2*9880d681SAndroid Build Coastguard Worker; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=VI -check-prefix=FUNC %s
3*9880d681SAndroid Build Coastguard Worker
4*9880d681SAndroid Build Coastguard Workerdeclare double @llvm.AMDGPU.rsq.clamped.f64(double) nounwind readnone
5*9880d681SAndroid Build Coastguard Worker
6*9880d681SAndroid Build Coastguard Worker; FUNC-LABEL: {{^}}rsq_clamped_f64:
7*9880d681SAndroid Build Coastguard Worker; SI: v_rsq_clamp_f64_e32
8*9880d681SAndroid Build Coastguard Worker
9*9880d681SAndroid Build Coastguard Worker; VI-DAG: v_rsq_f64_e32 [[RSQ:v\[[0-9]+:[0-9]+\]]], s[{{[0-9]+:[0-9]+}}]
10*9880d681SAndroid Build Coastguard Worker; TODO: this constant should be folded:
11*9880d681SAndroid Build Coastguard Worker; VI-DAG: s_mov_b32 s[[LOW1:[0-9+]]], -1
12*9880d681SAndroid Build Coastguard Worker; VI-DAG: s_mov_b32 s[[HIGH1:[0-9+]]], 0x7fefffff
13*9880d681SAndroid Build Coastguard Worker; VI-DAG: v_min_f64 v[0:1], [[RSQ]], s{{\[}}[[LOW1]]:[[HIGH1]]]
14*9880d681SAndroid Build Coastguard Worker; VI-DAG: s_mov_b32 s[[HIGH2:[0-9+]]], 0xffefffff
15*9880d681SAndroid Build Coastguard Worker; VI-DAG: v_max_f64 v[0:1], v[0:1], s{{\[}}[[LOW1]]:[[HIGH2]]]
16*9880d681SAndroid Build Coastguard Worker
17*9880d681SAndroid Build Coastguard Workerdefine void @rsq_clamped_f64(double addrspace(1)* %out, double %src) nounwind {
18*9880d681SAndroid Build Coastguard Worker  %rsq_clamped = call double @llvm.AMDGPU.rsq.clamped.f64(double %src) nounwind readnone
19*9880d681SAndroid Build Coastguard Worker  store double %rsq_clamped, double addrspace(1)* %out, align 8
20*9880d681SAndroid Build Coastguard Worker  ret void
21*9880d681SAndroid Build Coastguard Worker}
22