/external/llvm-project/llvm/test/CodeGen/AMDGPU/ |
D | atomic_optimizations_global_pointer.ll | 12 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 13 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 14 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 18 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 19 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} 30 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 31 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 32 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 36 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 37 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} [all …]
|
D | sgpr-spill.mir | 14 # CHECK: $sgpr12 = S_MOV_B32 $exec_lo 15 # CHECK: $exec_lo = S_MOV_B32 1 18 # CHECK: $exec_lo = S_MOV_B32 killed $sgpr12 22 # CHECK: $sgpr12 = S_MOV_B32 $exec_lo 23 # CHECK: $exec_lo = S_MOV_B32 1 31 # GCN32: $sgpr12 = S_MOV_B32 $exec_lo 33 # GCN32: $exec_lo = S_MOV_B32 3 37 # GCN32: $exec_lo = S_MOV_B32 killed $sgpr12 43 # GCN32: $sgpr12 = S_MOV_B32 $exec_lo 45 # GCN32: $exec_lo = S_MOV_B32 3 [all …]
|
D | optimize-negated-cond-exec-masking-wave32.mir | 5 # GCN-NEXT: $vcc_lo = S_ANDN2_B32 $exec_lo, %0, implicit-def dead $scc 14 $vcc_lo = S_AND_B32 $exec_lo, killed $vcc_lo, implicit-def dead $scc 27 # GCN-NEXT: $vcc_lo = S_ANDN2_B32 $exec_lo, %0, implicit-def dead $scc 36 $vcc_lo = S_AND_B32 killed %2, $exec_lo, implicit-def dead $scc 52 # GCN-NEXT: $vcc_lo = S_AND_B32 $exec_lo, $vcc_lo, implicit-def dead $scc 62 $vcc_lo = S_AND_B32 $exec_lo, killed $vcc_lo, implicit-def dead $scc 78 # GCN-NEXT: $vcc_lo = S_AND_B32 %2, $exec_lo, implicit-def dead $scc 88 $vcc_lo = S_AND_B32 killed %2, $exec_lo, implicit-def dead $scc 100 # GCN: $vcc_lo = S_AND_B32 $exec_lo, undef $vcc_lo, implicit-def dead $scc 106 $vcc_lo = S_AND_B32 $exec_lo, undef $vcc_lo, implicit-def dead $scc [all …]
|
D | vcmpx-exec-war-hazard.mir | 4 # GCN: $sgpr0 = S_MOV_B32 $exec_lo 13 $sgpr0 = S_MOV_B32 $exec_lo 40 # GCN: $vgpr0 = V_MOV_B32_e32 $exec_lo, implicit $exec 47 $vgpr0 = V_MOV_B32_e32 $exec_lo, implicit $exec 72 # GCN: $sgpr0 = S_MOV_B32 $exec_lo 81 $sgpr0 = S_MOV_B32 $exec_lo 91 # GCN: $sgpr0 = S_MOV_B32 $exec_lo 100 $sgpr0 = S_MOV_B32 $exec_lo 110 # GCN: $sgpr0 = S_MOV_B32 $exec_lo 119 $sgpr0 = S_MOV_B32 $exec_lo [all …]
|
D | atomic_optimizations_raw_buffer.ll | 14 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 15 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 16 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 20 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 21 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} 32 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 33 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 34 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 38 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 39 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} [all …]
|
D | atomic_optimizations_struct_buffer.ll | 14 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 15 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 16 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 20 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 21 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} 32 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 33 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 34 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 38 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 39 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} [all …]
|
D | llvm.amdgcn.image.gather4.a16.dim.ll | 19 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 20 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 23 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 46 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 47 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 50 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 73 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 74 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 77 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 100 ; GFX10-NEXT: s_mov_b32 s12, exec_lo [all …]
|
D | atomic_optimizations_buffer.ll | 15 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 16 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 17 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 21 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 22 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} 33 ; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo 34 ; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec 35 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0 39 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]] 40 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} [all …]
|
D | llvm.amdgcn.image.sample.dim.ll | 27 ; GFX10-NEXT: s_mov_b32 s12, exec_lo ; encoding: [0x7e,0x03,0x8c,0xbe] 28 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo ; encoding: [0x7e,0x09,0xfe,0xbe] 29 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 ; encoding: [0x7e,0x0c,0x7e,0x87] 82 ; GFX10-NEXT: s_mov_b32 s28, exec_lo ; encoding: [0x7e,0x03,0x9c,0xbe] 83 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo ; encoding: [0x7e,0x09,0xfe,0xbe] 95 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s28 ; encoding: [0x7e,0x1c,0x7e,0x87] 136 ; GFX10-NEXT: s_mov_b32 s12, exec_lo ; encoding: [0x7e,0x03,0x8c,0xbe] 137 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo ; encoding: [0x7e,0x09,0xfe,0xbe] 141 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 ; encoding: [0x7e,0x0c,0x7e,0x87] 183 ; GFX10-NEXT: s_mov_b32 s12, exec_lo ; encoding: [0x7e,0x03,0x8c,0xbe] [all …]
|
D | llvm.amdgcn.image.sample.a16.dim.ll | 17 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 18 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 19 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 42 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 43 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 46 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 69 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 70 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 73 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 96 ; GFX10-NEXT: s_mov_b32 s12, exec_lo [all …]
|
D | fold-reload-into-exec.mir | 15 ; CHECK: S_NOP 0, implicit-def $exec_lo 16 ; CHECK: $sgpr0 = S_MOV_B32 $exec_lo 21 ; CHECK: $exec_lo = S_MOV_B32 killed $sgpr0 23 S_NOP 0, implicit-def $exec_lo 24 %0:sreg_32 = COPY $exec_lo 26 $exec_lo = COPY %0 93 …: S_NOP 0, implicit-def renamable $sgpr0, implicit-def dead renamable $sgpr1, implicit-def $exec_lo 98 ; CHECK: $exec_lo = S_MOV_B32 killed $sgpr0 100 S_NOP 0, implicit-def %0:sreg_32, implicit-def %1:sreg_32, implicit-def $exec_lo 102 $exec_lo = COPY %0
|
D | read_register.ll | 18 ; CHECK: v_mov_b32_e32 v[[LO:[0-9]+]], exec_lo 56 ; CHECK: v_mov_b32_e32 [[COPY:v[0-9]+]], exec_lo 59 %exec_lo = call i32 @llvm.read_register.i32(metadata !5) 60 store i32 %exec_lo, i32 addrspace(1)* %out 80 !5 = !{!"exec_lo"}
|
D | llvm.amdgcn.image.sample.d16.dim.ll | 37 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 38 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 39 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 99 ; GFX10-NEXT: s_mov_b32 s28, exec_lo 100 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 105 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s28 238 ; GFX10-NEXT: s_mov_b32 s12, exec_lo 239 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 240 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 301 ; GFX10-NEXT: s_mov_b32 s12, exec_lo [all …]
|
D | mubuf-legalize-operands.ll | 27 ; W32: s_mov_b32 [[SAVEEXEC:s[0-9]+]], exec_lo 39 ; W32: s_xor_b32 exec_lo, exec_lo, [[SAVE]] 41 ; W32: s_mov_b32 exec_lo, [[SAVEEXEC]] 92 ; W32: s_mov_b32 [[SAVEEXEC:s[0-9]+]], exec_lo 104 ; W32: s_xor_b32 exec_lo, exec_lo, [[SAVE]] 107 ; W32: s_mov_b32 exec_lo, [[SAVEEXEC]] 109 ; W32: s_mov_b32 [[SAVEEXEC:s[0-9]+]], exec_lo 122 ; W32: s_xor_b32 exec_lo, exec_lo, [[SAVE]] 125 ; W32: s_mov_b32 exec_lo, [[SAVEEXEC]] 188 ; W32-DAG: s_mov_b32 [[SAVEEXEC:s[0-9]+]], exec_lo [all …]
|
D | atomic_optimizations_pixelshader.ll | 150 ; GFX1032-NEXT: s_mov_b32 s9, exec_lo 155 ; GFX1032-NEXT: s_mov_b32 s10, exec_lo 167 ; GFX1032-NEXT: s_or_b32 exec_lo, exec_lo, s9 172 ; GFX1032-NEXT: s_or_b32 exec_lo, exec_lo, s8 174 ; GFX1032-NEXT: s_andn2_b32 vcc_lo, exec_lo, s4 218 ; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0 273 ; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0 345 ; GFX1064-NEXT: v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0 381 ; GFX1032-NEXT: s_mov_b32 s9, exec_lo 387 ; GFX1032-NEXT: s_not_b32 exec_lo, exec_lo [all …]
|
D | vgpr-descriptor-waterfall-loop-idom-update.ll | 13 ; GCN-NEXT: s_mov_b32 s5, exec_lo 32 ; GCN-NEXT: s_xor_b32 exec_lo, exec_lo, s4 35 ; GCN-NEXT: s_mov_b32 exec_lo, s5
|
D | wave32.ll | 170 ; GFX1032: s_andn2_b32 exec_lo, exec_lo, s{{[0-9]+}} 180 ; GFX1032: s_xor_b32 s{{[0-9]+}}, exec_lo, s{{[0-9]+}} 184 ; GFX1032: s_or_b32 exec_lo, exec_lo, s{{[0-9]+}} 237 ; GFX1032: s_andn2_b32 [[MASK1:s[0-9]+]], [[MASK1]], exec_lo 239 ; GFX1032: s_and_b32 [[MASK0]], [[MASK0]], exec_lo 244 ; GFX1032: s_and_b32 [[TMP0:s[0-9]+]], exec_lo, [[MASK1]] 248 ; GFX1032: s_andn2_b32 exec_lo, exec_lo, [[ACC]] 253 ; GFX1032: s_or_b32 [[MASK1]], [[MASK1]], exec_lo 448 ; GFX1032: s_or_b32 exec_lo, exec_lo, [[SAVE]] 496 ; GFX1032: s_and_b32 vcc_lo, exec_lo, vcc_lo [all …]
|
/external/llvm-project/llvm/test/MC/AMDGPU/ |
D | reg-syntax-err.s | 75 s_mov_b64 s[10:11], [exec_lo,vcc_hi] 78 s_mov_b64 s[10:11], [exec_hi,exec_lo] 81 s_mov_b64 s[10:11], [exec_lo,exec_lo] 84 s_mov_b64 s[10:11], [exec,exec_lo] 87 s_mov_b64 s[10:11], [exec_lo,exec] 90 s_mov_b64 s[10:11], [exec_lo,s0] 93 s_mov_b64 s[10:11], [s0,exec_lo]
|
D | smem-err.s | 12 s_store_dword exec_lo, s[2:3], 0x0 24 s_buffer_store_dword exec_lo, s[0:3], 0x0 36 s_load_dword exec_lo, s[0:1], s4 48 s_buffer_load_dword exec_lo, s[0:3], s4
|
D | gfx10_asm_all.s | 10612 s_mov_b32 exec_lo, s1 10627 s_mov_b32 s0, exec_lo 10714 s_cmov_b32 exec_lo, s1 10729 s_cmov_b32 s0, exec_lo 10816 s_not_b32 exec_lo, s1 10831 s_not_b32 s0, exec_lo 10918 s_wqm_b32 exec_lo, s1 10933 s_wqm_b32 s0, exec_lo 11020 s_brev_b32 exec_lo, s1 11035 s_brev_b32 s0, exec_lo [all …]
|
/external/llvm-project/llvm/test/CodeGen/AMDGPU/GlobalISel/ |
D | llvm.amdgcn.image.gather4.dim.ll | 29 ; GFX10NSA-NEXT: s_mov_b32 s1, exec_lo 31 ; GFX10NSA-NEXT: s_wqm_b32 exec_lo, exec_lo 32 ; GFX10NSA-NEXT: s_and_b32 exec_lo, exec_lo, s1 76 ; GFX10NSA-NEXT: s_mov_b32 s1, exec_lo 78 ; GFX10NSA-NEXT: s_wqm_b32 exec_lo, exec_lo 79 ; GFX10NSA-NEXT: s_and_b32 exec_lo, exec_lo, s1 123 ; GFX10NSA-NEXT: s_mov_b32 s1, exec_lo 125 ; GFX10NSA-NEXT: s_wqm_b32 exec_lo, exec_lo 126 ; GFX10NSA-NEXT: s_and_b32 exec_lo, exec_lo, s1 170 ; GFX10NSA-NEXT: s_mov_b32 s1, exec_lo [all …]
|
D | llvm.amdgcn.image.gather4.a16.dim.ll | 32 ; GFX10NSA-NEXT: s_mov_b32 s28, exec_lo 34 ; GFX10NSA-NEXT: s_wqm_b32 exec_lo, exec_lo 48 ; GFX10NSA-NEXT: s_and_b32 exec_lo, exec_lo, s28 86 ; GFX10NSA-NEXT: s_mov_b32 s28, exec_lo 88 ; GFX10NSA-NEXT: s_wqm_b32 exec_lo, exec_lo 105 ; GFX10NSA-NEXT: s_and_b32 exec_lo, exec_lo, s28 143 ; GFX10NSA-NEXT: s_mov_b32 s28, exec_lo 145 ; GFX10NSA-NEXT: s_wqm_b32 exec_lo, exec_lo 162 ; GFX10NSA-NEXT: s_and_b32 exec_lo, exec_lo, s28 198 ; GFX10NSA-NEXT: s_mov_b32 s28, exec_lo [all …]
|
D | llvm.amdgcn.image.gather4.o.dim.ll | 29 ; GFX10-NEXT: s_mov_b32 s1, exec_lo 31 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 32 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s1 76 ; GFX10-NEXT: s_mov_b32 s1, exec_lo 78 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 79 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s1 123 ; GFX10-NEXT: s_mov_b32 s1, exec_lo 125 ; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 126 ; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s1 170 ; GFX10-NEXT: s_mov_b32 s1, exec_lo [all …]
|
D | llvm.amdgcn.intersect_ray.ll | 81 ; GCN-NEXT: s_mov_b32 s1, exec_lo 93 ; GCN-NEXT: s_xor_b32 exec_lo, exec_lo, s0 96 ; GCN-NEXT: s_mov_b32 exec_lo, s1 116 ; GCN-NEXT: s_mov_b32 s1, exec_lo 133 ; GCN-NEXT: s_xor_b32 exec_lo, exec_lo, s0 136 ; GCN-NEXT: s_mov_b32 exec_lo, s1 151 ; GCN-NEXT: s_mov_b32 s1, exec_lo 163 ; GCN-NEXT: s_xor_b32 exec_lo, exec_lo, s0 166 ; GCN-NEXT: s_mov_b32 exec_lo, s1 186 ; GCN-NEXT: s_mov_b32 s1, exec_lo [all …]
|
/external/llvm/test/CodeGen/AMDGPU/ |
D | read_register.ll | 18 ; CHECK: v_mov_b32_e32 v[[LO:[0-9]+]], exec_lo 56 ; CHECK: v_mov_b32_e32 [[COPY:v[0-9]+]], exec_lo 59 %exec_lo = call i32 @llvm.read_register.i32(metadata !5) 60 store i32 %exec_lo, i32 addrspace(1)* %out 80 !5 = !{!"exec_lo"}
|