126 lines
5.5 KiB
LLVM
126 lines
5.5 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx908 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX908 %s
|
|
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX90A %s
|
|
|
|
define void @global_atomic_fadd_f32(ptr addrspace(1) %ptr, float %data) {
|
|
; GFX908-LABEL: global_atomic_fadd_f32:
|
|
; GFX908: ; %bb.0:
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX908-NEXT: global_atomic_add_f32 v[0:1], v2, off
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX908-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX90A-LABEL: global_atomic_fadd_f32:
|
|
; GFX90A: ; %bb.0:
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX90A-NEXT: global_atomic_add_f32 v[0:1], v2, off
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX90A-NEXT: s_setpc_b64 s[30:31]
|
|
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %ptr, float %data)
|
|
ret void
|
|
}
|
|
|
|
define void @global_atomic_fadd_f32_off_2048(ptr addrspace(1) %ptr, float %data) {
|
|
; GFX908-LABEL: global_atomic_fadd_f32_off_2048:
|
|
; GFX908: ; %bb.0:
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX908-NEXT: global_atomic_add_f32 v[0:1], v2, off offset:2048
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX908-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX90A-LABEL: global_atomic_fadd_f32_off_2048:
|
|
; GFX90A: ; %bb.0:
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX90A-NEXT: global_atomic_add_f32 v[0:1], v2, off offset:2048
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX90A-NEXT: s_setpc_b64 s[30:31]
|
|
%gep = getelementptr float, ptr addrspace(1) %ptr, i64 512
|
|
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %gep, float %data)
|
|
ret void
|
|
}
|
|
|
|
define void @global_atomic_fadd_f32_off_neg2047(ptr addrspace(1) %ptr, float %data) {
|
|
; GFX908-LABEL: global_atomic_fadd_f32_off_neg2047:
|
|
; GFX908: ; %bb.0:
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX908-NEXT: global_atomic_add_f32 v[0:1], v2, off offset:-2044
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX908-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX90A-LABEL: global_atomic_fadd_f32_off_neg2047:
|
|
; GFX90A: ; %bb.0:
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX90A-NEXT: global_atomic_add_f32 v[0:1], v2, off offset:-2044
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX90A-NEXT: s_setpc_b64 s[30:31]
|
|
%gep = getelementptr float, ptr addrspace(1) %ptr, i64 -511
|
|
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %gep, float %data)
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @global_atomic_fadd_f32_off_ss(ptr addrspace(1) %ptr, float %data) {
|
|
; GFX908-LABEL: global_atomic_fadd_f32_off_ss:
|
|
; GFX908: ; %bb.0:
|
|
; GFX908-NEXT: s_load_dword s2, s[4:5], 0x8
|
|
; GFX908-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
|
|
; GFX908-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX908-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX908-NEXT: v_mov_b32_e32 v0, s2
|
|
; GFX908-NEXT: global_atomic_add_f32 v1, v0, s[0:1] offset:2048
|
|
; GFX908-NEXT: s_endpgm
|
|
;
|
|
; GFX90A-LABEL: global_atomic_fadd_f32_off_ss:
|
|
; GFX90A: ; %bb.0:
|
|
; GFX90A-NEXT: s_load_dword s2, s[4:5], 0x8
|
|
; GFX90A-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
|
|
; GFX90A-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX90A-NEXT: v_mov_b32_e32 v0, s2
|
|
; GFX90A-NEXT: global_atomic_add_f32 v1, v0, s[0:1] offset:2048
|
|
; GFX90A-NEXT: s_endpgm
|
|
%gep = getelementptr float, ptr addrspace(1) %ptr, i64 512
|
|
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %gep, float %data)
|
|
ret void
|
|
}
|
|
|
|
define void @global_atomic_fadd_v2f16(ptr addrspace(1) %ptr, <2 x half> %data) {
|
|
; GFX908-LABEL: global_atomic_fadd_v2f16:
|
|
; GFX908: ; %bb.0:
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX908-NEXT: global_atomic_pk_add_f16 v[0:1], v2, off
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX908-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX90A-LABEL: global_atomic_fadd_v2f16:
|
|
; GFX90A: ; %bb.0:
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX90A-NEXT: global_atomic_pk_add_f16 v[0:1], v2, off
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX90A-NEXT: s_setpc_b64 s[30:31]
|
|
%ret = call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %ptr, <2 x half> %data)
|
|
ret void
|
|
}
|
|
|
|
define void @global_atomic_fadd_v2f16_off_neg2047(ptr addrspace(1) %ptr, <2 x half> %data) {
|
|
; GFX908-LABEL: global_atomic_fadd_v2f16_off_neg2047:
|
|
; GFX908: ; %bb.0:
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX908-NEXT: global_atomic_pk_add_f16 v[0:1], v2, off offset:-2044
|
|
; GFX908-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX908-NEXT: s_setpc_b64 s[30:31]
|
|
;
|
|
; GFX90A-LABEL: global_atomic_fadd_v2f16_off_neg2047:
|
|
; GFX90A: ; %bb.0:
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
|
|
; GFX90A-NEXT: global_atomic_pk_add_f16 v[0:1], v2, off offset:-2044
|
|
; GFX90A-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX90A-NEXT: s_setpc_b64 s[30:31]
|
|
%gep = getelementptr <2 x half>, ptr addrspace(1) %ptr, i64 -511
|
|
%ret = call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %gep, <2 x half> %data)
|
|
ret void
|
|
}
|
|
|
|
declare float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) nocapture, float) #0
|
|
declare <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) nocapture, <2 x half>) #0
|
|
|
|
attributes #0 = { argmemonly nounwind willreturn }
|