[llvm] ef316da - AMDGPU: Fix dead check prefixes in test
Matt Arsenault via llvm-commits
llvm-commits at lists.llvm.org
Wed Mar 27 04:50:12 PDT 2024
Author: Matt Arsenault
Date: 2024-03-27T14:42:47+03:00
New Revision: ef316da4a2c5954a02c92707b5cb621402b76910
URL: https://github.com/llvm/llvm-project/commit/ef316da4a2c5954a02c92707b5cb621402b76910
DIFF: https://github.com/llvm/llvm-project/commit/ef316da4a2c5954a02c92707b5cb621402b76910.diff
LOG: AMDGPU: Fix dead check prefixes in test
Added:
Modified:
llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll
llvm/test/CodeGen/AMDGPU/global_atomics_i64_system.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll
index 76ec1cc84f55b2..99d02ffaa523d5 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll
@@ -358,65 +358,6 @@ define amdgpu_gfx i32 @global_atomic_xchg_i32_ret_offset_scalar(ptr addrspace(1)
; ---------------------------------------------------------------------
define void @global_atomic_xchg_f32_noret(ptr addrspace(1) %ptr, float %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_noret:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_load_dword v3, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB0_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: v_mov_b32_e32 v3, v4
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB0_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_noret:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_load_dword v3, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB0_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: v_mov_b32_e32 v3, v4
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB0_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_noret:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v3, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB0_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB0_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_noret:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -450,69 +391,6 @@ define void @global_atomic_xchg_f32_noret(ptr addrspace(1) %ptr, float %in) {
}
define void @global_atomic_xchg_f32_noret_offset(ptr addrspace(1) %out, float %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_noret_offset:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_add_f32_e32 v0, vcc, 16, v0
-; GCN1-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
-; GCN1-NEXT: global_load_dword v3, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB1_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: v_mov_b32_e32 v3, v4
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB1_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_noret_offset:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_add_u32_e32 v0, vcc, 16, v0
-; GCN2-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
-; GCN2-NEXT: global_load_dword v3, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB1_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: v_mov_b32_e32 v3, v4
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB1_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_noret_offset:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v3, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB1_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB1_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_noret_offset:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -549,71 +427,6 @@ define void @global_atomic_xchg_f32_noret_offset(ptr addrspace(1) %out, float %i
}
define float @global_atomic_xchg_f32_ret(ptr addrspace(1) %ptr, float %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_ret:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_load_dword v4, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB2_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v3, v4
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB2_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: v_mov_b32_e32 v0, v4
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_ret:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_load_dword v4, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB2_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v3, v4
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB2_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: v_mov_b32_e32 v0, v4
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_ret:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v4, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB2_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB2_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v0, v4
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_ret:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -648,73 +461,6 @@ define float @global_atomic_xchg_f32_ret(ptr addrspace(1) %ptr, float %in) {
}
define float @global_atomic_xchg_f32_ret_offset(ptr addrspace(1) %out, float %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_ret_offset:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_add_f32_e32 v4, vcc, 16, v0
-; GCN1-NEXT: v_addc_u32_e32 v5, vcc, 0, v1, vcc
-; GCN1-NEXT: global_load_dword v0, v[4:5]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB3_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v3, v0
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[4:5], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB3_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_ret_offset:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_add_u32_e32 v4, vcc, 16, v0
-; GCN2-NEXT: v_addc_u32_e32 v5, vcc, 0, v1, vcc
-; GCN2-NEXT: global_load_dword v0, v[4:5]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB3_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v3, v0
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[4:5], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB3_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_ret_offset:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v4, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB3_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB3_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v0, v4
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_ret_offset:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -752,80 +498,6 @@ define float @global_atomic_xchg_f32_ret_offset(ptr addrspace(1) %out, float %in
}
define amdgpu_gfx void @global_atomic_xchg_f32_noret_scalar(ptr addrspace(1) inreg %ptr, float inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_noret_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v0, s4
-; GCN1-NEXT: v_mov_b32_e32 v1, s5
-; GCN1-NEXT: global_load_dword v1, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[34:35], 0
-; GCN1-NEXT: .LBB4_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v2, s4
-; GCN1-NEXT: v_mov_b32_e32 v0, s6
-; GCN1-NEXT: v_mov_b32_e32 v3, s5
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN1-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN1-NEXT: v_mov_b32_e32 v1, v0
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_cbranch_execnz .LBB4_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_noret_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v0, s4
-; GCN2-NEXT: v_mov_b32_e32 v1, s5
-; GCN2-NEXT: global_load_dword v1, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[34:35], 0
-; GCN2-NEXT: .LBB4_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v2, s4
-; GCN2-NEXT: v_mov_b32_e32 v0, s6
-; GCN2-NEXT: v_mov_b32_e32 v3, s5
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN2-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN2-NEXT: v_mov_b32_e32 v1, v0
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_cbranch_execnz .LBB4_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_noret_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v1, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB4_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v2, s4
-; GCN3-NEXT: v_mov_b32_e32 v0, s6
-; GCN3-NEXT: v_mov_b32_e32 v3, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: v_mov_b32_e32 v1, v0
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB4_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_noret_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -876,84 +548,6 @@ define amdgpu_gfx void @global_atomic_xchg_f32_noret_scalar(ptr addrspace(1) inr
}
define amdgpu_gfx void @global_atomic_xchg_f32_noret_offset_scalar(ptr addrspace(1) inreg %out, float inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_noret_offset_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: s_add_u32 s34, s4, 16
-; GCN1-NEXT: s_addc_u32 s35, s5, 0
-; GCN1-NEXT: v_mov_b32_e32 v0, s34
-; GCN1-NEXT: v_mov_b32_e32 v1, s35
-; GCN1-NEXT: global_load_dword v1, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[36:37], 0
-; GCN1-NEXT: .LBB5_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v2, s34
-; GCN1-NEXT: v_mov_b32_e32 v0, s6
-; GCN1-NEXT: v_mov_b32_e32 v3, s35
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN1-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN1-NEXT: v_mov_b32_e32 v1, v0
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_cbranch_execnz .LBB5_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_noret_offset_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: s_add_u32 s34, s4, 16
-; GCN2-NEXT: s_addc_u32 s35, s5, 0
-; GCN2-NEXT: v_mov_b32_e32 v0, s34
-; GCN2-NEXT: v_mov_b32_e32 v1, s35
-; GCN2-NEXT: global_load_dword v1, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[36:37], 0
-; GCN2-NEXT: .LBB5_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v2, s34
-; GCN2-NEXT: v_mov_b32_e32 v0, s6
-; GCN2-NEXT: v_mov_b32_e32 v3, s35
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN2-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN2-NEXT: v_mov_b32_e32 v1, v0
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_cbranch_execnz .LBB5_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_noret_offset_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v1, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB5_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v2, s4
-; GCN3-NEXT: v_mov_b32_e32 v0, s6
-; GCN3-NEXT: v_mov_b32_e32 v3, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: v_mov_b32_e32 v1, v0
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB5_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_noret_offset_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -1007,83 +601,6 @@ define amdgpu_gfx void @global_atomic_xchg_f32_noret_offset_scalar(ptr addrspace
}
define amdgpu_gfx float @global_atomic_xchg_f32_ret_scalar(ptr addrspace(1) inreg %ptr, float inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_ret_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v0, s4
-; GCN1-NEXT: v_mov_b32_e32 v1, s5
-; GCN1-NEXT: global_load_dword v0, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[34:35], 0
-; GCN1-NEXT: .LBB6_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v3, s4
-; GCN1-NEXT: v_mov_b32_e32 v1, s6
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v2, v0
-; GCN1-NEXT: v_mov_b32_e32 v4, s5
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN1-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_cbranch_execnz .LBB6_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_ret_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v0, s4
-; GCN2-NEXT: v_mov_b32_e32 v1, s5
-; GCN2-NEXT: global_load_dword v0, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[34:35], 0
-; GCN2-NEXT: .LBB6_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v3, s4
-; GCN2-NEXT: v_mov_b32_e32 v1, s6
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v2, v0
-; GCN2-NEXT: v_mov_b32_e32 v4, s5
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN2-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_cbranch_execnz .LBB6_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_ret_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v0, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB6_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v3, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s6
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v2, v0
-; GCN3-NEXT: v_mov_b32_e32 v4, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB6_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_ret_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -1134,87 +651,6 @@ define amdgpu_gfx float @global_atomic_xchg_f32_ret_scalar(ptr addrspace(1) inre
}
define amdgpu_gfx float @global_atomic_xchg_f32_ret_offset_scalar(ptr addrspace(1) inreg %out, float inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f32_ret_offset_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: s_add_u32 s34, s4, 16
-; GCN1-NEXT: s_addc_u32 s35, s5, 0
-; GCN1-NEXT: v_mov_b32_e32 v0, s34
-; GCN1-NEXT: v_mov_b32_e32 v1, s35
-; GCN1-NEXT: global_load_dword v0, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[36:37], 0
-; GCN1-NEXT: .LBB7_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v3, s34
-; GCN1-NEXT: v_mov_b32_e32 v1, s6
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v2, v0
-; GCN1-NEXT: v_mov_b32_e32 v4, s35
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN1-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_cbranch_execnz .LBB7_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f32_ret_offset_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: s_add_u32 s34, s4, 16
-; GCN2-NEXT: s_addc_u32 s35, s5, 0
-; GCN2-NEXT: v_mov_b32_e32 v0, s34
-; GCN2-NEXT: v_mov_b32_e32 v1, s35
-; GCN2-NEXT: global_load_dword v0, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[36:37], 0
-; GCN2-NEXT: .LBB7_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v3, s34
-; GCN2-NEXT: v_mov_b32_e32 v1, s6
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v2, v0
-; GCN2-NEXT: v_mov_b32_e32 v4, s35
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN2-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_cbranch_execnz .LBB7_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f32_ret_offset_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v0, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB7_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v3, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s6
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v2, v0
-; GCN3-NEXT: v_mov_b32_e32 v4, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB7_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f32_ret_offset_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_i64_system.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_i64_system.ll
index d137f471910dc6..380ce7f3b93988 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_i64_system.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_i64_system.ll
@@ -372,65 +372,6 @@ define amdgpu_gfx i64 @global_atomic_xchg_i64_ret_offset_scalar(ptr addrspace(1)
; ---------------------------------------------------------------------
define void @global_atomic_xchg_f64_noret(ptr addrspace(1) %ptr, double %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_noret:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_load_dword v3, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB0_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: v_mov_b32_e32 v3, v4
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB0_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_noret:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_load_dword v3, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB0_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: v_mov_b32_e32 v3, v4
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB0_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_noret:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v3, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB0_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB0_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_noret:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -464,69 +405,6 @@ define void @global_atomic_xchg_f64_noret(ptr addrspace(1) %ptr, double %in) {
}
define void @global_atomic_xchg_f64_noret_offset(ptr addrspace(1) %out, double %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_noret_offset:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_add_f64_e32 v0, vcc, 16, v0
-; GCN1-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
-; GCN1-NEXT: global_load_dword v3, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB1_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: v_mov_b32_e32 v3, v4
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB1_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_noret_offset:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_add_u32_e32 v0, vcc, 16, v0
-; GCN2-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
-; GCN2-NEXT: global_load_dword v3, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB1_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: v_mov_b32_e32 v3, v4
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB1_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_noret_offset:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v3, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB1_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB1_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_noret_offset:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -563,71 +441,6 @@ define void @global_atomic_xchg_f64_noret_offset(ptr addrspace(1) %out, double %
}
define double @global_atomic_xchg_f64_ret(ptr addrspace(1) %ptr, double %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_ret:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_load_dword v4, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB2_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v3, v4
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB2_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: v_mov_b32_e32 v0, v4
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_ret:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_load_dword v4, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB2_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v3, v4
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB2_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: v_mov_b32_e32 v0, v4
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_ret:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v4, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB2_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB2_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v0, v4
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_ret:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -663,73 +476,6 @@ define double @global_atomic_xchg_f64_ret(ptr addrspace(1) %ptr, double %in) {
}
define double @global_atomic_xchg_f64_ret_offset(ptr addrspace(1) %out, double %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_ret_offset:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_add_f64_e32 v4, vcc, 16, v0
-; GCN1-NEXT: v_addc_u32_e32 v5, vcc, 0, v1, vcc
-; GCN1-NEXT: global_load_dword v0, v[4:5]
-; GCN1-NEXT: s_mov_b64 s[4:5], 0
-; GCN1-NEXT: .LBB3_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v3, v0
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[4:5], v[2:3] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v3
-; GCN1-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_cbranch_execnz .LBB3_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_ret_offset:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_add_u32_e32 v4, vcc, 16, v0
-; GCN2-NEXT: v_addc_u32_e32 v5, vcc, 0, v1, vcc
-; GCN2-NEXT: global_load_dword v0, v[4:5]
-; GCN2-NEXT: s_mov_b64 s[4:5], 0
-; GCN2-NEXT: .LBB3_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v3, v0
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[4:5], v[2:3] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v3
-; GCN2-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_cbranch_execnz .LBB3_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_ret_offset:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_load_dword v4, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[4:5], 0
-; GCN3-NEXT: .LBB3_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v3, v4
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v4, v[0:1], v[2:3] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v4, v3
-; GCN3-NEXT: s_or_b64 s[4:5], vcc, s[4:5]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[4:5]
-; GCN3-NEXT: s_cbranch_execnz .LBB3_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[4:5]
-; GCN3-NEXT: v_mov_b32_e32 v0, v4
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_ret_offset:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -768,80 +514,6 @@ define double @global_atomic_xchg_f64_ret_offset(ptr addrspace(1) %out, double %
}
define amdgpu_gfx void @global_atomic_xchg_f64_noret_scalar(ptr addrspace(1) inreg %ptr, double inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_noret_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v0, s4
-; GCN1-NEXT: v_mov_b32_e32 v1, s5
-; GCN1-NEXT: global_load_dword v1, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[34:35], 0
-; GCN1-NEXT: .LBB4_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v2, s4
-; GCN1-NEXT: v_mov_b32_e32 v0, s6
-; GCN1-NEXT: v_mov_b32_e32 v3, s5
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN1-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN1-NEXT: v_mov_b32_e32 v1, v0
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_cbranch_execnz .LBB4_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_noret_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v0, s4
-; GCN2-NEXT: v_mov_b32_e32 v1, s5
-; GCN2-NEXT: global_load_dword v1, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[34:35], 0
-; GCN2-NEXT: .LBB4_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v2, s4
-; GCN2-NEXT: v_mov_b32_e32 v0, s6
-; GCN2-NEXT: v_mov_b32_e32 v3, s5
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN2-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN2-NEXT: v_mov_b32_e32 v1, v0
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_cbranch_execnz .LBB4_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_noret_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v1, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB4_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v2, s4
-; GCN3-NEXT: v_mov_b32_e32 v0, s6
-; GCN3-NEXT: v_mov_b32_e32 v3, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: v_mov_b32_e32 v1, v0
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB4_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_noret_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -896,84 +568,6 @@ define amdgpu_gfx void @global_atomic_xchg_f64_noret_scalar(ptr addrspace(1) inr
}
define amdgpu_gfx void @global_atomic_xchg_f64_noret_offset_scalar(ptr addrspace(1) inreg %out, double inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_noret_offset_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: s_add_u32 s34, s4, 16
-; GCN1-NEXT: s_addc_u32 s35, s5, 0
-; GCN1-NEXT: v_mov_b32_e32 v0, s34
-; GCN1-NEXT: v_mov_b32_e32 v1, s35
-; GCN1-NEXT: global_load_dword v1, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[36:37], 0
-; GCN1-NEXT: .LBB5_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v2, s34
-; GCN1-NEXT: v_mov_b32_e32 v0, s6
-; GCN1-NEXT: v_mov_b32_e32 v3, s35
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN1-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN1-NEXT: v_mov_b32_e32 v1, v0
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_cbranch_execnz .LBB5_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_noret_offset_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: s_add_u32 s34, s4, 16
-; GCN2-NEXT: s_addc_u32 s35, s5, 0
-; GCN2-NEXT: v_mov_b32_e32 v0, s34
-; GCN2-NEXT: v_mov_b32_e32 v1, s35
-; GCN2-NEXT: global_load_dword v1, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[36:37], 0
-; GCN2-NEXT: .LBB5_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v2, s34
-; GCN2-NEXT: v_mov_b32_e32 v0, s6
-; GCN2-NEXT: v_mov_b32_e32 v3, s35
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN2-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN2-NEXT: v_mov_b32_e32 v1, v0
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_cbranch_execnz .LBB5_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_noret_offset_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v1, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB5_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v2, s4
-; GCN3-NEXT: v_mov_b32_e32 v0, s6
-; GCN3-NEXT: v_mov_b32_e32 v3, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[2:3], v[0:1] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v1
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: v_mov_b32_e32 v1, v0
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB5_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_noret_offset_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -1029,83 +623,6 @@ define amdgpu_gfx void @global_atomic_xchg_f64_noret_offset_scalar(ptr addrspace
}
define amdgpu_gfx double @global_atomic_xchg_f64_ret_scalar(ptr addrspace(1) inreg %ptr, double inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_ret_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v0, s4
-; GCN1-NEXT: v_mov_b32_e32 v1, s5
-; GCN1-NEXT: global_load_dword v0, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[34:35], 0
-; GCN1-NEXT: .LBB6_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v3, s4
-; GCN1-NEXT: v_mov_b32_e32 v1, s6
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v2, v0
-; GCN1-NEXT: v_mov_b32_e32 v4, s5
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN1-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_cbranch_execnz .LBB6_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_ret_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v0, s4
-; GCN2-NEXT: v_mov_b32_e32 v1, s5
-; GCN2-NEXT: global_load_dword v0, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[34:35], 0
-; GCN2-NEXT: .LBB6_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v3, s4
-; GCN2-NEXT: v_mov_b32_e32 v1, s6
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v2, v0
-; GCN2-NEXT: v_mov_b32_e32 v4, s5
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN2-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_cbranch_execnz .LBB6_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_ret_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v0, v[0:1]
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB6_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v3, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s6
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v2, v0
-; GCN3-NEXT: v_mov_b32_e32 v4, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB6_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_ret_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -1160,87 +677,6 @@ define amdgpu_gfx double @global_atomic_xchg_f64_ret_scalar(ptr addrspace(1) inr
}
define amdgpu_gfx double @global_atomic_xchg_f64_ret_offset_scalar(ptr addrspace(1) inreg %out, double inreg %in) {
-; GCN1-LABEL: global_atomic_xchg_f64_ret_offset_scalar:
-; GCN1: ; %bb.0:
-; GCN1-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN1-NEXT: s_add_u32 s34, s4, 16
-; GCN1-NEXT: s_addc_u32 s35, s5, 0
-; GCN1-NEXT: v_mov_b32_e32 v0, s34
-; GCN1-NEXT: v_mov_b32_e32 v1, s35
-; GCN1-NEXT: global_load_dword v0, v[0:1]
-; GCN1-NEXT: s_mov_b64 s[36:37], 0
-; GCN1-NEXT: .LBB7_1: ; %atomicrmw.start
-; GCN1-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN1-NEXT: v_mov_b32_e32 v3, s34
-; GCN1-NEXT: v_mov_b32_e32 v1, s6
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: v_mov_b32_e32 v2, v0
-; GCN1-NEXT: v_mov_b32_e32 v4, s35
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN1-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN1-NEXT: buffer_wbinvl1_vol
-; GCN1-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN1-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN1-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_cbranch_execnz .LBB7_1
-; GCN1-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN1-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN1-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN2-LABEL: global_atomic_xchg_f64_ret_offset_scalar:
-; GCN2: ; %bb.0:
-; GCN2-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN2-NEXT: s_add_u32 s34, s4, 16
-; GCN2-NEXT: s_addc_u32 s35, s5, 0
-; GCN2-NEXT: v_mov_b32_e32 v0, s34
-; GCN2-NEXT: v_mov_b32_e32 v1, s35
-; GCN2-NEXT: global_load_dword v0, v[0:1]
-; GCN2-NEXT: s_mov_b64 s[36:37], 0
-; GCN2-NEXT: .LBB7_1: ; %atomicrmw.start
-; GCN2-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN2-NEXT: v_mov_b32_e32 v3, s34
-; GCN2-NEXT: v_mov_b32_e32 v1, s6
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: v_mov_b32_e32 v2, v0
-; GCN2-NEXT: v_mov_b32_e32 v4, s35
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] glc
-; GCN2-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN2-NEXT: buffer_wbinvl1_vol
-; GCN2-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN2-NEXT: s_or_b64 s[36:37], vcc, s[36:37]
-; GCN2-NEXT: s_andn2_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_cbranch_execnz .LBB7_1
-; GCN2-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN2-NEXT: s_or_b64 exec, exec, s[36:37]
-; GCN2-NEXT: s_setpc_b64 s[30:31]
-;
-; GCN3-LABEL: global_atomic_xchg_f64_ret_offset_scalar:
-; GCN3: ; %bb.0:
-; GCN3-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v0, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s5
-; GCN3-NEXT: global_load_dword v0, v[0:1] offset:16
-; GCN3-NEXT: s_mov_b64 s[34:35], 0
-; GCN3-NEXT: .LBB7_1: ; %atomicrmw.start
-; GCN3-NEXT: ; =>This Inner Loop Header: Depth=1
-; GCN3-NEXT: v_mov_b32_e32 v3, s4
-; GCN3-NEXT: v_mov_b32_e32 v1, s6
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: v_mov_b32_e32 v2, v0
-; GCN3-NEXT: v_mov_b32_e32 v4, s5
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: global_atomic_cmpswap v0, v[3:4], v[1:2] offset:16 glc
-; GCN3-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
-; GCN3-NEXT: buffer_wbinvl1_vol
-; GCN3-NEXT: v_cmp_eq_u32_e32 vcc, v0, v2
-; GCN3-NEXT: s_or_b64 s[34:35], vcc, s[34:35]
-; GCN3-NEXT: s_andn2_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_cbranch_execnz .LBB7_1
-; GCN3-NEXT: ; %bb.2: ; %atomicrmw.end
-; GCN3-NEXT: s_or_b64 exec, exec, s[34:35]
-; GCN3-NEXT: s_setpc_b64 s[30:31]
; SI-LABEL: global_atomic_xchg_f64_ret_offset_scalar:
; SI: ; %bb.0:
; SI-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
More information about the llvm-commits
mailing list