Skip to content

[AMDGPU] Inflate to %AV regclass #147413

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 5 commits into
base: main
Choose a base branch
from
+876 −0
Open

[AMDGPU] Inflate to %AV regclass #147413

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
3f6ecaf
[AMDGPU] Inflate to %av regclass
jrbyrnes Jul 15, 2025
6d52737
Update lit tests
jrbyrnes Sep 26, 2025
f797b9b
Revert "Update lit tests"
jrbyrnes Oct 8, 2025
4dfc7ab
Control with flag
jrbyrnes Oct 8, 2025
896fb4b
Update llvm/lib/Target/AMDGPU/AMDGPUPrepareAGPRAlloc.cpp
jrbyrnes Oct 15, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions llvm/lib/Target/AMDGPU/AMDGPUPrepareAGPRAlloc.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,6 +97,8 @@ bool AMDGPUPrepareAGPRAllocImpl::run(MachineFunction &MF) {

const MCInstrDesc &AVImmPseudo32 = TII.get(AMDGPU::AV_MOV_B32_IMM_PSEUDO);
const MCInstrDesc &AVImmPseudo64 = TII.get(AMDGPU::AV_MOV_B64_IMM_PSEUDO);
const SIRegisterInfo *TRI =
static_cast<const SIRegisterInfo *>(MRI.getTargetRegisterInfo());

bool Changed = false;
for (MachineBasicBlock &MBB : MF) {
Expand All @@ -119,6 +121,20 @@ bool AMDGPUPrepareAGPRAllocImpl::run(MachineFunction &MF) {
Changed = true;
continue;
}

for (MachineOperand &Op : MI.operands()) {
if (!Op.isReg() || !Op.isDef())
continue;

Register DefReg = Op.getReg();
if (DefReg.isPhysical())
continue;

const TargetRegisterClass *RC = MRI.getRegClass(DefReg);

if (TRI->isAGPRClass(RC) || TRI->isVGPRClass(RC))
Changed |= MRI.recomputeRegClass(DefReg);
}
}
}

Expand Down
144 changes: 72 additions & 72 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,9 +23,9 @@ define amdgpu_kernel void @raw_buffer_atomic_add_noret_f64(<4 x i32> %rsrc, doub
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 offen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 offen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: raw_buffer_atomic_add_noret_f64:
Expand All @@ -34,9 +34,9 @@ define amdgpu_kernel void @raw_buffer_atomic_add_noret_f64(<4 x i32> %rsrc, doub
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 offen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 offen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: raw_buffer_atomic_add_noret_f64:
Expand Down Expand Up @@ -142,9 +142,9 @@ define amdgpu_kernel void @raw_ptr_buffer_atomic_add_noret_f64(ptr addrspace(8)
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 offen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 offen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: raw_ptr_buffer_atomic_add_noret_f64:
Expand All @@ -153,9 +153,9 @@ define amdgpu_kernel void @raw_ptr_buffer_atomic_add_noret_f64(ptr addrspace(8)
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 offen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 offen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: raw_ptr_buffer_atomic_add_noret_f64:
Expand Down Expand Up @@ -261,9 +261,9 @@ define amdgpu_kernel void @struct_buffer_atomic_add_noret_f64(<4 x i32> %rsrc, d
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: struct_buffer_atomic_add_noret_f64:
Expand All @@ -272,9 +272,9 @@ define amdgpu_kernel void @struct_buffer_atomic_add_noret_f64(<4 x i32> %rsrc, d
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: struct_buffer_atomic_add_noret_f64:
Expand Down Expand Up @@ -379,9 +379,9 @@ define amdgpu_kernel void @struct_ptr_buffer_atomic_add_noret_f64(ptr addrspace(
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: struct_ptr_buffer_atomic_add_noret_f64:
Expand All @@ -390,9 +390,9 @@ define amdgpu_kernel void @struct_ptr_buffer_atomic_add_noret_f64(ptr addrspace(
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_add_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: struct_ptr_buffer_atomic_add_noret_f64:
Expand Down Expand Up @@ -497,9 +497,9 @@ define amdgpu_kernel void @raw_buffer_atomic_min_noret_f64(<4 x i32> %rsrc, doub
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 offen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 offen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: raw_buffer_atomic_min_noret_f64:
Expand All @@ -508,9 +508,9 @@ define amdgpu_kernel void @raw_buffer_atomic_min_noret_f64(<4 x i32> %rsrc, doub
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 offen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 offen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: raw_buffer_atomic_min_noret_f64:
Expand Down Expand Up @@ -616,9 +616,9 @@ define amdgpu_kernel void @raw_ptr_buffer_atomic_min_noret_f64(ptr addrspace(8)
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 offen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 offen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: raw_ptr_buffer_atomic_min_noret_f64:
Expand All @@ -627,9 +627,9 @@ define amdgpu_kernel void @raw_ptr_buffer_atomic_min_noret_f64(ptr addrspace(8)
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 offen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 offen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: raw_ptr_buffer_atomic_min_noret_f64:
Expand Down Expand Up @@ -735,9 +735,9 @@ define amdgpu_kernel void @struct_buffer_atomic_min_noret_f64(<4 x i32> %rsrc, d
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: struct_buffer_atomic_min_noret_f64:
Expand All @@ -746,9 +746,9 @@ define amdgpu_kernel void @struct_buffer_atomic_min_noret_f64(<4 x i32> %rsrc, d
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: struct_buffer_atomic_min_noret_f64:
Expand Down Expand Up @@ -853,9 +853,9 @@ define amdgpu_kernel void @struct_ptr_buffer_atomic_min_noret_f64(ptr addrspace(
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: struct_ptr_buffer_atomic_min_noret_f64:
Expand All @@ -864,9 +864,9 @@ define amdgpu_kernel void @struct_ptr_buffer_atomic_min_noret_f64(ptr addrspace(
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_min_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: struct_ptr_buffer_atomic_min_noret_f64:
Expand Down Expand Up @@ -971,9 +971,9 @@ define amdgpu_kernel void @raw_buffer_atomic_max_noret_f64(<4 x i32> %rsrc, doub
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 offen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 offen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: raw_buffer_atomic_max_noret_f64:
Expand All @@ -982,9 +982,9 @@ define amdgpu_kernel void @raw_buffer_atomic_max_noret_f64(<4 x i32> %rsrc, doub
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 offen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 offen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: raw_buffer_atomic_max_noret_f64:
Expand Down Expand Up @@ -1090,9 +1090,9 @@ define amdgpu_kernel void @raw_ptr_buffer_atomic_max_noret_f64(ptr addrspace(8)
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 offen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 offen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: raw_ptr_buffer_atomic_max_noret_f64:
Expand All @@ -1101,9 +1101,9 @@ define amdgpu_kernel void @raw_ptr_buffer_atomic_max_noret_f64(ptr addrspace(8)
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 offen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 offen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: raw_ptr_buffer_atomic_max_noret_f64:
Expand Down Expand Up @@ -1209,9 +1209,9 @@ define amdgpu_kernel void @struct_buffer_atomic_max_noret_f64(<4 x i32> %rsrc, d
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: struct_buffer_atomic_max_noret_f64:
Expand All @@ -1220,9 +1220,9 @@ define amdgpu_kernel void @struct_buffer_atomic_max_noret_f64(<4 x i32> %rsrc, d
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: struct_buffer_atomic_max_noret_f64:
Expand Down Expand Up @@ -1327,9 +1327,9 @@ define amdgpu_kernel void @struct_ptr_buffer_atomic_max_noret_f64(ptr addrspace(
; GFX90A-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX90A-NEXT: s_waitcnt lgkmcnt(0)
; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v2, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[6:7], s[6:7] op_sel:[0,1]
; GFX90A-NEXT: v_mov_b32_e32 v0, s8
; GFX90A-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX90A-NEXT: s_endpgm
;
; GFX942-LABEL: struct_ptr_buffer_atomic_max_noret_f64:
Expand All @@ -1338,9 +1338,9 @@ define amdgpu_kernel void @struct_ptr_buffer_atomic_max_noret_f64(ptr addrspace(
; GFX942-NEXT: s_load_dword s8, s[4:5], 0x3c
; GFX942-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX942-NEXT: s_waitcnt lgkmcnt(0)
; GFX942-NEXT: v_mov_b64_e32 v[0:1], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v2, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[0:1], v2, s[0:3], 0 idxen
; GFX942-NEXT: v_mov_b64_e32 v[2:3], s[6:7]
; GFX942-NEXT: v_mov_b32_e32 v0, s8
; GFX942-NEXT: buffer_atomic_max_f64 v[2:3], v0, s[0:3], 0 idxen
; GFX942-NEXT: s_endpgm
;
; GFX1250-LABEL: struct_ptr_buffer_atomic_max_noret_f64:
Expand Down
Loading