@@ -20,7 +20,7 @@ define amdgpu_ps void @test_swmmac_f32_16x16x32_f16_index_key(<4 x half> %A, <8
2020; GFX12-NEXT: s_wait_loadcnt 0x0
2121; GFX12-NEXT: v_swmmac_f32_16x16x32_f16 v[20:23], v[0:1], v[2:5], v10
2222; GFX12-NEXT: v_swmmac_f32_16x16x32_f16 v[24:27], v[0:1], v[2:5], v10 index_key:1
23- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
23+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
2424; GFX12-NEXT: v_swmmac_f32_16x16x32_f16 v[28:31], v[0:1], v[2:5], v10 index_key:2
2525; GFX12-NEXT: v_swmmac_f32_16x16x32_f16 v[6:9], v[0:1], v[2:5], v10 index_key:3
2626; GFX12-NEXT: global_store_b128 v[12:13], v[20:23], off
@@ -64,7 +64,7 @@ define amdgpu_ps void @test_swmmac_f32_16x16x32_bf16_index_key(<4 x i16> %A, <8
6464; GFX12-NEXT: s_wait_loadcnt 0x0
6565; GFX12-NEXT: v_swmmac_f32_16x16x32_bf16 v[20:23], v[0:1], v[2:5], v10
6666; GFX12-NEXT: v_swmmac_f32_16x16x32_bf16 v[24:27], v[0:1], v[2:5], v10 index_key:1
67- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
67+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
6868; GFX12-NEXT: v_swmmac_f32_16x16x32_bf16 v[28:31], v[0:1], v[2:5], v10 index_key:2
6969; GFX12-NEXT: v_swmmac_f32_16x16x32_bf16 v[6:9], v[0:1], v[2:5], v10 index_key:3
7070; GFX12-NEXT: global_store_b128 v[12:13], v[20:23], off
@@ -102,7 +102,7 @@ define amdgpu_ps void @test_swmmac_f16_16x16x32_f16_index_key(<4 x half> %A, <8
102102; GFX12-NEXT: s_wait_loadcnt 0x0
103103; GFX12-NEXT: v_swmmac_f16_16x16x32_f16 v[8:9], v[0:1], v[2:5], v22
104104; GFX12-NEXT: v_swmmac_f16_16x16x32_f16 v[18:19], v[0:1], v[2:5], v22 index_key:1
105- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
105+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
106106; GFX12-NEXT: v_swmmac_f16_16x16x32_f16 v[20:21], v[0:1], v[2:5], v22 index_key:2
107107; GFX12-NEXT: v_swmmac_f16_16x16x32_f16 v[6:7], v[0:1], v[2:5], v22 index_key:3
108108; GFX12-NEXT: global_store_b64 v[10:11], v[8:9], off
@@ -140,7 +140,7 @@ define amdgpu_ps void @test_swmmac_bf16_16x16x32_bf16_index_key(<4 x i16> %A, <8
140140; GFX12-NEXT: s_wait_loadcnt 0x0
141141; GFX12-NEXT: v_swmmac_bf16_16x16x32_bf16 v[8:9], v[0:1], v[2:5], v22
142142; GFX12-NEXT: v_swmmac_bf16_16x16x32_bf16 v[18:19], v[0:1], v[2:5], v22 index_key:1
143- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
143+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
144144; GFX12-NEXT: v_swmmac_bf16_16x16x32_bf16 v[20:21], v[0:1], v[2:5], v22 index_key:2
145145; GFX12-NEXT: v_swmmac_bf16_16x16x32_bf16 v[6:7], v[0:1], v[2:5], v22 index_key:3
146146; GFX12-NEXT: global_store_b64 v[10:11], v[8:9], off
@@ -184,7 +184,7 @@ define amdgpu_ps void @test_swmmac_i32_16x16x32_iu8_index_key(i32 %A, <2 x i32>
184184; GFX12-NEXT: s_wait_loadcnt 0x0
185185; GFX12-NEXT: v_swmmac_i32_16x16x32_iu8 v[17:20], v0, v[1:2], v7
186186; GFX12-NEXT: v_swmmac_i32_16x16x32_iu8 v[21:24], v0, v[1:2], v7 index_key:1
187- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
187+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
188188; GFX12-NEXT: v_swmmac_i32_16x16x32_iu8 v[25:28], v0, v[1:2], v7 index_key:2
189189; GFX12-NEXT: v_swmmac_i32_16x16x32_iu8 v[3:6], v0, v[1:2], v7 index_key:3
190190; GFX12-NEXT: global_store_b128 v[9:10], v[17:20], off
@@ -280,7 +280,7 @@ define amdgpu_ps void @test_swmmac_f32_16x16x32_fp8_fp8_index_key(i32 %A, <2 x i
280280; GFX12-NEXT: s_wait_loadcnt 0x0
281281; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_fp8 v[17:20], v0, v[1:2], v7
282282; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_fp8 v[21:24], v0, v[1:2], v7 index_key:1
283- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
283+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
284284; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_fp8 v[25:28], v0, v[1:2], v7 index_key:2
285285; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_fp8 v[3:6], v0, v[1:2], v7 index_key:3
286286; GFX12-NEXT: global_store_b128 v[9:10], v[17:20], off
@@ -324,7 +324,7 @@ define amdgpu_ps void @test_swmmac_f32_16x16x32_fp8_bf8_index_key(i32 %A, <2 x i
324324; GFX12-NEXT: s_wait_loadcnt 0x0
325325; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_bf8 v[17:20], v0, v[1:2], v7
326326; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_bf8 v[21:24], v0, v[1:2], v7 index_key:1
327- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
327+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
328328; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_bf8 v[25:28], v0, v[1:2], v7 index_key:2
329329; GFX12-NEXT: v_swmmac_f32_16x16x32_fp8_bf8 v[3:6], v0, v[1:2], v7 index_key:3
330330; GFX12-NEXT: global_store_b128 v[9:10], v[17:20], off
@@ -368,7 +368,7 @@ define amdgpu_ps void @test_swmmac_f32_16x16x32_bf8_fp8_index_key(i32 %A, <2 x i
368368; GFX12-NEXT: s_wait_loadcnt 0x0
369369; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_fp8 v[17:20], v0, v[1:2], v7
370370; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_fp8 v[21:24], v0, v[1:2], v7 index_key:1
371- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
371+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
372372; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_fp8 v[25:28], v0, v[1:2], v7 index_key:2
373373; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_fp8 v[3:6], v0, v[1:2], v7 index_key:3
374374; GFX12-NEXT: global_store_b128 v[9:10], v[17:20], off
@@ -412,7 +412,7 @@ define amdgpu_ps void @test_swmmac_f32_16x16x32_bf8_bf8_index_key(i32 %A, <2 x i
412412; GFX12-NEXT: s_wait_loadcnt 0x0
413413; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_bf8 v[17:20], v0, v[1:2], v7
414414; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_bf8 v[21:24], v0, v[1:2], v7 index_key:1
415- ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_3 )
415+ ; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1 )
416416; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_bf8 v[25:28], v0, v[1:2], v7 index_key:2
417417; GFX12-NEXT: v_swmmac_f32_16x16x32_bf8_bf8 v[3:6], v0, v[1:2], v7 index_key:3
418418; GFX12-NEXT: global_store_b128 v[9:10], v[17:20], off
0 commit comments