diff options
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll | 45 |
1 files changed, 24 insertions, 21 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll index 7bd1ff220197..9a90faf72346 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll @@ -78,12 +78,13 @@ define amdgpu_kernel void @v_mul_i64_zext_src1(ptr addrspace(1) %out, ptr addrsp ; GFX11-NEXT: v_lshlrev_b32_e32 v0, 2, v0 ; GFX11-NEXT: s_waitcnt lgkmcnt(0) ; GFX11-NEXT: global_load_b64 v[2:3], v1, s[2:3] -; GFX11-NEXT: global_load_b32 v6, v0, s[4:5] +; GFX11-NEXT: global_load_b32 v5, v0, s[4:5] ; GFX11-NEXT: s_waitcnt vmcnt(0) -; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v2, v6, 0 +; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v2, v5, 0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX11-NEXT: v_mad_u64_u32 v[4:5], null, v3, v6, v[1:2] -; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, v4 +; GFX11-NEXT: v_mov_b32_e32 v4, v1 +; GFX11-NEXT: v_mad_u64_u32 v[1:2], null, v3, v5, v[4:5] +; GFX11-NEXT: v_mov_b32_e32 v2, 0 ; GFX11-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX11-NEXT: s_endpgm %tid = call i32 @llvm.amdgcn.workitem.id.x() @@ -126,13 +127,14 @@ define amdgpu_kernel void @v_mul_i64_zext_src0(ptr addrspace(1) %out, ptr addrsp ; GFX11-NEXT: v_lshlrev_b32_e32 v1, 2, v0 ; GFX11-NEXT: v_lshlrev_b32_e32 v0, 3, v0 ; GFX11-NEXT: s_waitcnt lgkmcnt(0) -; GFX11-NEXT: global_load_b32 v6, v1, s[2:3] +; GFX11-NEXT: global_load_b32 v5, v1, s[2:3] ; GFX11-NEXT: global_load_b64 v[2:3], v0, s[4:5] ; GFX11-NEXT: s_waitcnt vmcnt(0) -; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v6, v2, 0 +; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v5, v2, 0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX11-NEXT: v_mad_u64_u32 v[4:5], null, v6, v3, v[1:2] -; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, v4 +; GFX11-NEXT: v_mov_b32_e32 v4, v1 +; GFX11-NEXT: v_mad_u64_u32 v[1:2], null, v5, v3, v[4:5] +; GFX11-NEXT: v_mov_b32_e32 v2, 0 ; GFX11-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX11-NEXT: s_endpgm %tid = call i32 @llvm.amdgcn.workitem.id.x() @@ -222,13 +224,14 @@ define amdgpu_kernel void @v_mul_i64_masked_src0_hi(ptr addrspace(1) %out, ptr a ; GFX11-NEXT: v_lshlrev_b32_e32 v0, 3, v0 ; GFX11-NEXT: s_waitcnt lgkmcnt(0) ; GFX11-NEXT: s_clause 0x1 -; GFX11-NEXT: global_load_b32 v6, v0, s[2:3] +; GFX11-NEXT: global_load_b32 v5, v0, s[2:3] ; GFX11-NEXT: global_load_b64 v[2:3], v0, s[4:5] ; GFX11-NEXT: s_waitcnt vmcnt(0) -; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v6, v2, 0 +; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v5, v2, 0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX11-NEXT: v_mad_u64_u32 v[4:5], null, v6, v3, v[1:2] -; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, v4 +; GFX11-NEXT: v_mov_b32_e32 v4, v1 +; GFX11-NEXT: v_mad_u64_u32 v[1:2], null, v5, v3, v[4:5] +; GFX11-NEXT: v_mov_b32_e32 v2, 0 ; GFX11-NEXT: global_store_b64 v2, v[0:1], s[0:1] ; GFX11-NEXT: s_endpgm %tid = call i32 @llvm.amdgcn.workitem.id.x() @@ -520,28 +523,28 @@ define amdgpu_kernel void @v_mul64_masked_before_and_in_branch(ptr addrspace(1) ; GFX11-NEXT: v_lshlrev_b32_e32 v0, 3, v0 ; GFX11-NEXT: s_waitcnt lgkmcnt(0) ; GFX11-NEXT: s_clause 0x1 -; GFX11-NEXT: global_load_b64 v[2:3], v0, s[2:3] -; GFX11-NEXT: global_load_b64 v[4:5], v0, s[4:5] +; GFX11-NEXT: global_load_b64 v[3:4], v0, s[2:3] +; GFX11-NEXT: global_load_b64 v[5:6], v0, s[4:5] ; GFX11-NEXT: s_mov_b32 s2, exec_lo ; GFX11-NEXT: ; implicit-def: $vgpr0_vgpr1 ; GFX11-NEXT: s_waitcnt vmcnt(1) -; GFX11-NEXT: v_cmpx_ge_u64_e32 0, v[2:3] +; GFX11-NEXT: v_cmpx_ge_u64_e32 0, v[3:4] ; GFX11-NEXT: s_xor_b32 s2, exec_lo, s2 ; GFX11-NEXT: s_cbranch_execz .LBB10_2 ; GFX11-NEXT: ; %bb.1: ; %else ; GFX11-NEXT: s_waitcnt vmcnt(0) -; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v2, v4, 0 +; GFX11-NEXT: v_mad_u64_u32 v[0:1], null, v3, v5, 0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) -; GFX11-NEXT: v_mad_u64_u32 v[3:4], null, v2, v5, v[1:2] -; GFX11-NEXT: ; implicit-def: $vgpr4_vgpr5 -; GFX11-NEXT: v_mov_b32_e32 v1, v3 -; GFX11-NEXT: ; implicit-def: $vgpr2_vgpr3 +; GFX11-NEXT: v_mov_b32_e32 v4, v1 +; GFX11-NEXT: v_mad_u64_u32 v[1:2], null, v3, v6, v[4:5] +; GFX11-NEXT: ; implicit-def: $vgpr3_vgpr4 +; GFX11-NEXT: ; implicit-def: $vgpr5_vgpr6 ; GFX11-NEXT: .LBB10_2: ; %Flow ; GFX11-NEXT: s_and_not1_saveexec_b32 s2, s2 ; GFX11-NEXT: s_cbranch_execz .LBB10_4 ; GFX11-NEXT: ; %bb.3: ; %if ; GFX11-NEXT: s_waitcnt vmcnt(0) -; GFX11-NEXT: v_mul_lo_u32 v1, v2, v5 +; GFX11-NEXT: v_mul_lo_u32 v1, v3, v6 ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: .LBB10_4: ; %endif ; GFX11-NEXT: s_or_b32 exec_lo, exec_lo, s2 |