|
| 1 | +# RUN: llc -march=amdgcn -mcpu=gfx1010 -start-before=greedy -verify-machineinstrs -o - %s | FileCheck -check-prefix=GCN %s |
| 2 | +... |
| 3 | +# GCN-LABEL: {{^}}"subvector-basic-bb" |
| 4 | +# GCN: s_subvector_loop_begin [[RS:s[0-9]]], BB0_2 |
| 5 | +# GCN: s_subvector_loop_end [[RS]], BB0_1 |
| 6 | +name: subvector-basic-bb |
| 7 | +tracksRegLiveness: true |
| 8 | +machineFunctionInfo: |
| 9 | + scratchRSrcReg: $sgpr0_sgpr1_sgpr2_sgpr3 |
| 10 | + scratchWaveOffsetReg: $sgpr4 |
| 11 | + frameOffsetReg: $sgpr5 |
| 12 | + stackPtrOffsetReg: $sgpr32 |
| 13 | +body: | |
| 14 | + bb.0: |
| 15 | + liveins: $sgpr0_sgpr1 |
| 16 | + successors: %bb.1, %bb.2 |
| 17 | +
|
| 18 | + %1:sgpr_64 = COPY $sgpr0_sgpr1 |
| 19 | + %4:sreg_128 = S_LOAD_DWORDX4_IMM %1, 36, 0, 0 |
| 20 | + %11:sreg_64_xexec = S_LOAD_DWORDX2_IMM %4.sub2_sub3, 0, 0, 0 |
| 21 | + undef %15.sub0:vreg_64 = COPY %4.sub0 |
| 22 | + %15.sub1:vreg_64 = COPY %4.sub1 |
| 23 | + %16:vgpr_32 = COPY %1.sub0 |
| 24 | + S_SUBVECTOR_LOOP_BEGIN %bb.2, undef %19:sreg_32, implicit-def $exec, implicit $exec, implicit-def %19 |
| 25 | +
|
| 26 | + bb.1: |
| 27 | + successors: %bb.1, %bb.2 |
| 28 | +
|
| 29 | + %14:sreg_32_xm0 = S_ADD_I32 %11.sub0, %11.sub1, implicit-def dead $scc |
| 30 | + %16:vgpr_32 = COPY %14 |
| 31 | + S_SUBVECTOR_LOOP_END %bb.1, %19:sreg_32, implicit-def $exec, implicit $exec, implicit-def %19 |
| 32 | +
|
| 33 | + bb.2: |
| 34 | +
|
| 35 | + GLOBAL_STORE_DWORD %15, %16, 0, 0, 0, 0, implicit $exec |
| 36 | + S_ENDPGM 0 |
| 37 | +... |
0 commit comments