1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX9 %s 3; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10 %s 4 5declare void @extern_func() #2 6 7define <4 x float> @non_preserved_vgpr_tuple8(<8 x i32> %rsrc, <4 x i32> %samp, float %bias, float %zcompare, float %s, float %t, float %clamp) { 8; The vgpr tuple8 operand in image_gather4_c_b_cl instruction needs not be 9; preserved across the call and should get 8 scratch registers. 10; GFX9-LABEL: non_preserved_vgpr_tuple8: 11; GFX9: ; %bb.0: ; %main_body 12; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) 13; GFX9-NEXT: s_or_saveexec_b64 s[4:5], -1 14; GFX9-NEXT: buffer_store_dword v40, off, s[0:3], s32 offset:16 ; 4-byte Folded Spill 15; GFX9-NEXT: s_mov_b64 exec, s[4:5] 16; GFX9-NEXT: s_mov_b32 s4, 0 17; GFX9-NEXT: v_writelane_b32 v40, s33, 2 18; GFX9-NEXT: s_mov_b32 s33, s32 19; GFX9-NEXT: v_mov_b32_e32 v36, v16 20; GFX9-NEXT: v_mov_b32_e32 v35, v15 21; GFX9-NEXT: v_mov_b32_e32 v34, v14 22; GFX9-NEXT: v_mov_b32_e32 v33, v13 23; GFX9-NEXT: v_mov_b32_e32 v32, v12 24; GFX9-NEXT: s_mov_b32 s5, s4 25; GFX9-NEXT: s_mov_b32 s6, s4 26; GFX9-NEXT: s_mov_b32 s7, s4 27; GFX9-NEXT: s_mov_b32 s8, s4 28; GFX9-NEXT: s_mov_b32 s9, s4 29; GFX9-NEXT: s_mov_b32 s10, s4 30; GFX9-NEXT: s_mov_b32 s11, s4 31; GFX9-NEXT: buffer_store_dword v41, off, s[0:3], s33 offset:12 ; 4-byte Folded Spill 32; GFX9-NEXT: buffer_store_dword v42, off, s[0:3], s33 offset:8 ; 4-byte Folded Spill 33; GFX9-NEXT: buffer_store_dword v43, off, s[0:3], s33 offset:4 ; 4-byte Folded Spill 34; GFX9-NEXT: buffer_store_dword v44, off, s[0:3], s33 ; 4-byte Folded Spill 35; GFX9-NEXT: ;;#ASMSTART 36; GFX9-NEXT: ;;#ASMEND 37; GFX9-NEXT: ;;#ASMSTART 38; GFX9-NEXT: ;;#ASMEND 39; GFX9-NEXT: ;;#ASMSTART 40; GFX9-NEXT: ;;#ASMEND 41; GFX9-NEXT: ;;#ASMSTART 42; GFX9-NEXT: ;;#ASMEND 43; GFX9-NEXT: image_gather4_c_b_cl v[41:44], v[32:36], s[4:11], s[4:7] dmask:0x1 44; GFX9-NEXT: s_addk_i32 s32, 0x800 45; GFX9-NEXT: s_getpc_b64 s[4:5] 46; GFX9-NEXT: s_add_u32 s4, s4, extern_func@gotpcrel32@lo+4 47; GFX9-NEXT: s_addc_u32 s5, s5, extern_func@gotpcrel32@hi+12 48; GFX9-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0 49; GFX9-NEXT: v_writelane_b32 v40, s30, 0 50; GFX9-NEXT: v_writelane_b32 v40, s31, 1 51; GFX9-NEXT: s_waitcnt lgkmcnt(0) 52; GFX9-NEXT: s_swappc_b64 s[30:31], s[4:5] 53; GFX9-NEXT: v_mov_b32_e32 v0, v41 54; GFX9-NEXT: v_mov_b32_e32 v1, v42 55; GFX9-NEXT: v_mov_b32_e32 v2, v43 56; GFX9-NEXT: v_mov_b32_e32 v3, v44 57; GFX9-NEXT: buffer_load_dword v44, off, s[0:3], s33 ; 4-byte Folded Reload 58; GFX9-NEXT: buffer_load_dword v43, off, s[0:3], s33 offset:4 ; 4-byte Folded Reload 59; GFX9-NEXT: buffer_load_dword v42, off, s[0:3], s33 offset:8 ; 4-byte Folded Reload 60; GFX9-NEXT: buffer_load_dword v41, off, s[0:3], s33 offset:12 ; 4-byte Folded Reload 61; GFX9-NEXT: v_readlane_b32 s4, v40, 0 62; GFX9-NEXT: v_readlane_b32 s5, v40, 1 63; GFX9-NEXT: s_addk_i32 s32, 0xf800 64; GFX9-NEXT: v_readlane_b32 s33, v40, 2 65; GFX9-NEXT: s_or_saveexec_b64 s[6:7], -1 66; GFX9-NEXT: buffer_load_dword v40, off, s[0:3], s32 offset:16 ; 4-byte Folded Reload 67; GFX9-NEXT: s_mov_b64 exec, s[6:7] 68; GFX9-NEXT: s_waitcnt vmcnt(0) 69; GFX9-NEXT: s_setpc_b64 s[4:5] 70; 71; GFX10-LABEL: non_preserved_vgpr_tuple8: 72; GFX10: ; %bb.0: ; %main_body 73; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) 74; GFX10-NEXT: s_waitcnt_vscnt null, 0x0 75; GFX10-NEXT: s_or_saveexec_b32 s4, -1 76; GFX10-NEXT: buffer_store_dword v40, off, s[0:3], s32 offset:16 ; 4-byte Folded Spill 77; GFX10-NEXT: s_waitcnt_depctr 0xffe3 78; GFX10-NEXT: s_mov_b32 exec_lo, s4 79; GFX10-NEXT: v_mov_b32_e32 v36, v16 80; GFX10-NEXT: v_mov_b32_e32 v35, v15 81; GFX10-NEXT: v_mov_b32_e32 v34, v14 82; GFX10-NEXT: v_mov_b32_e32 v33, v13 83; GFX10-NEXT: v_mov_b32_e32 v32, v12 84; GFX10-NEXT: s_mov_b32 s4, 0 85; GFX10-NEXT: v_writelane_b32 v40, s33, 2 86; GFX10-NEXT: s_mov_b32 s33, s32 87; GFX10-NEXT: s_mov_b32 s5, s4 88; GFX10-NEXT: s_mov_b32 s6, s4 89; GFX10-NEXT: s_mov_b32 s7, s4 90; GFX10-NEXT: s_mov_b32 s8, s4 91; GFX10-NEXT: s_mov_b32 s9, s4 92; GFX10-NEXT: s_mov_b32 s10, s4 93; GFX10-NEXT: s_mov_b32 s11, s4 94; GFX10-NEXT: buffer_store_dword v41, off, s[0:3], s33 offset:12 ; 4-byte Folded Spill 95; GFX10-NEXT: buffer_store_dword v42, off, s[0:3], s33 offset:8 ; 4-byte Folded Spill 96; GFX10-NEXT: buffer_store_dword v43, off, s[0:3], s33 offset:4 ; 4-byte Folded Spill 97; GFX10-NEXT: buffer_store_dword v44, off, s[0:3], s33 ; 4-byte Folded Spill 98; GFX10-NEXT: ;;#ASMSTART 99; GFX10-NEXT: ;;#ASMEND 100; GFX10-NEXT: ;;#ASMSTART 101; GFX10-NEXT: ;;#ASMEND 102; GFX10-NEXT: ;;#ASMSTART 103; GFX10-NEXT: ;;#ASMEND 104; GFX10-NEXT: ;;#ASMSTART 105; GFX10-NEXT: ;;#ASMEND 106; GFX10-NEXT: image_gather4_c_b_cl v[41:44], v[32:36], s[4:11], s[4:7] dmask:0x1 dim:SQ_RSRC_IMG_2D 107; GFX10-NEXT: s_addk_i32 s32, 0x400 108; GFX10-NEXT: s_waitcnt_depctr 0xffe3 109; GFX10-NEXT: s_getpc_b64 s[4:5] 110; GFX10-NEXT: s_add_u32 s4, s4, extern_func@gotpcrel32@lo+4 111; GFX10-NEXT: s_addc_u32 s5, s5, extern_func@gotpcrel32@hi+12 112; GFX10-NEXT: v_writelane_b32 v40, s30, 0 113; GFX10-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0 114; GFX10-NEXT: v_writelane_b32 v40, s31, 1 115; GFX10-NEXT: s_waitcnt lgkmcnt(0) 116; GFX10-NEXT: s_swappc_b64 s[30:31], s[4:5] 117; GFX10-NEXT: v_mov_b32_e32 v0, v41 118; GFX10-NEXT: v_mov_b32_e32 v1, v42 119; GFX10-NEXT: v_mov_b32_e32 v2, v43 120; GFX10-NEXT: v_mov_b32_e32 v3, v44 121; GFX10-NEXT: s_clause 0x3 122; GFX10-NEXT: buffer_load_dword v44, off, s[0:3], s33 123; GFX10-NEXT: buffer_load_dword v43, off, s[0:3], s33 offset:4 124; GFX10-NEXT: buffer_load_dword v42, off, s[0:3], s33 offset:8 125; GFX10-NEXT: buffer_load_dword v41, off, s[0:3], s33 offset:12 126; GFX10-NEXT: v_readlane_b32 s4, v40, 0 127; GFX10-NEXT: v_readlane_b32 s5, v40, 1 128; GFX10-NEXT: s_addk_i32 s32, 0xfc00 129; GFX10-NEXT: v_readlane_b32 s33, v40, 2 130; GFX10-NEXT: s_or_saveexec_b32 s6, -1 131; GFX10-NEXT: buffer_load_dword v40, off, s[0:3], s32 offset:16 ; 4-byte Folded Reload 132; GFX10-NEXT: s_waitcnt_depctr 0xffe3 133; GFX10-NEXT: s_mov_b32 exec_lo, s6 134; GFX10-NEXT: s_waitcnt vmcnt(0) 135; GFX10-NEXT: s_setpc_b64 s[4:5] 136main_body: 137 call void asm sideeffect "", "~{v0},~{v1},~{v2},~{v3},~{v4},~{v5},~{v6},~{v7}"() #0 138 call void asm sideeffect "", "~{v8},~{v9},~{v10},~{v11},~{v12},~{v13},~{v14},~{v15}"() #0 139 call void asm sideeffect "", "~{v16},~{v17},~{v18},~{v19},~{v20},~{v21},~{v22},~{v23}"() #0 140 call void asm sideeffect "", "~{v24},~{v25},~{v26},~{v27},~{v28},~{v29},~{v30},~{v31}"() #0 141 %v = call <4 x float> @llvm.amdgcn.image.gather4.c.b.cl.2d.v4f32.f32.f32(i32 1, float %bias, float %zcompare, float %s, float %t, float %clamp, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) 142 call void @extern_func() 143 ret <4 x float> %v 144} 145 146define <4 x float> @call_preserved_vgpr_tuple8(<8 x i32> %rsrc, <4 x i32> %samp, float %bias, float %zcompare, float %s, float %t, float %clamp) { 147; The vgpr tuple8 operand in image_gather4_c_b_cl instruction needs to be preserved 148; across the call and should get allcoated to 8 CSRs. 149; Only the lower 5 sub-registers of the tuple are preserved. 150; The upper 3 sub-registers are unused. 151; GFX9-LABEL: call_preserved_vgpr_tuple8: 152; GFX9: ; %bb.0: ; %main_body 153; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) 154; GFX9-NEXT: s_or_saveexec_b64 s[4:5], -1 155; GFX9-NEXT: buffer_store_dword v40, off, s[0:3], s32 offset:20 ; 4-byte Folded Spill 156; GFX9-NEXT: s_mov_b64 exec, s[4:5] 157; GFX9-NEXT: v_writelane_b32 v40, s33, 10 158; GFX9-NEXT: v_writelane_b32 v40, s36, 0 159; GFX9-NEXT: v_writelane_b32 v40, s37, 1 160; GFX9-NEXT: v_writelane_b32 v40, s38, 2 161; GFX9-NEXT: v_writelane_b32 v40, s39, 3 162; GFX9-NEXT: v_writelane_b32 v40, s40, 4 163; GFX9-NEXT: v_writelane_b32 v40, s41, 5 164; GFX9-NEXT: s_mov_b32 s33, s32 165; GFX9-NEXT: v_writelane_b32 v40, s42, 6 166; GFX9-NEXT: s_mov_b32 s36, 0 167; GFX9-NEXT: buffer_store_dword v41, off, s[0:3], s33 offset:16 ; 4-byte Folded Spill 168; GFX9-NEXT: buffer_store_dword v42, off, s[0:3], s33 offset:12 ; 4-byte Folded Spill 169; GFX9-NEXT: buffer_store_dword v43, off, s[0:3], s33 offset:8 ; 4-byte Folded Spill 170; GFX9-NEXT: buffer_store_dword v44, off, s[0:3], s33 offset:4 ; 4-byte Folded Spill 171; GFX9-NEXT: buffer_store_dword v45, off, s[0:3], s33 ; 4-byte Folded Spill 172; GFX9-NEXT: v_writelane_b32 v40, s43, 7 173; GFX9-NEXT: v_mov_b32_e32 v45, v16 174; GFX9-NEXT: v_mov_b32_e32 v44, v15 175; GFX9-NEXT: v_mov_b32_e32 v43, v14 176; GFX9-NEXT: v_mov_b32_e32 v42, v13 177; GFX9-NEXT: v_mov_b32_e32 v41, v12 178; GFX9-NEXT: s_mov_b32 s37, s36 179; GFX9-NEXT: s_mov_b32 s38, s36 180; GFX9-NEXT: s_mov_b32 s39, s36 181; GFX9-NEXT: s_mov_b32 s40, s36 182; GFX9-NEXT: s_mov_b32 s41, s36 183; GFX9-NEXT: s_mov_b32 s42, s36 184; GFX9-NEXT: s_mov_b32 s43, s36 185; GFX9-NEXT: image_gather4_c_b_cl v[0:3], v[41:45], s[36:43], s[4:7] dmask:0x1 186; GFX9-NEXT: s_addk_i32 s32, 0x800 187; GFX9-NEXT: s_getpc_b64 s[4:5] 188; GFX9-NEXT: s_add_u32 s4, s4, extern_func@gotpcrel32@lo+4 189; GFX9-NEXT: s_addc_u32 s5, s5, extern_func@gotpcrel32@hi+12 190; GFX9-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0 191; GFX9-NEXT: v_writelane_b32 v40, s30, 8 192; GFX9-NEXT: v_writelane_b32 v40, s31, 9 193; GFX9-NEXT: s_waitcnt vmcnt(0) 194; GFX9-NEXT: global_store_dwordx4 v[0:1], v[0:3], off 195; GFX9-NEXT: s_waitcnt lgkmcnt(0) 196; GFX9-NEXT: s_swappc_b64 s[30:31], s[4:5] 197; GFX9-NEXT: image_gather4_c_b_cl v[0:3], v[41:45], s[36:43], s[4:7] dmask:0x1 198; GFX9-NEXT: s_nop 0 199; GFX9-NEXT: buffer_load_dword v45, off, s[0:3], s33 ; 4-byte Folded Reload 200; GFX9-NEXT: buffer_load_dword v44, off, s[0:3], s33 offset:4 ; 4-byte Folded Reload 201; GFX9-NEXT: buffer_load_dword v43, off, s[0:3], s33 offset:8 ; 4-byte Folded Reload 202; GFX9-NEXT: buffer_load_dword v42, off, s[0:3], s33 offset:12 ; 4-byte Folded Reload 203; GFX9-NEXT: buffer_load_dword v41, off, s[0:3], s33 offset:16 ; 4-byte Folded Reload 204; GFX9-NEXT: v_readlane_b32 s4, v40, 8 205; GFX9-NEXT: v_readlane_b32 s5, v40, 9 206; GFX9-NEXT: v_readlane_b32 s43, v40, 7 207; GFX9-NEXT: v_readlane_b32 s42, v40, 6 208; GFX9-NEXT: v_readlane_b32 s41, v40, 5 209; GFX9-NEXT: v_readlane_b32 s40, v40, 4 210; GFX9-NEXT: v_readlane_b32 s39, v40, 3 211; GFX9-NEXT: v_readlane_b32 s38, v40, 2 212; GFX9-NEXT: v_readlane_b32 s37, v40, 1 213; GFX9-NEXT: v_readlane_b32 s36, v40, 0 214; GFX9-NEXT: s_addk_i32 s32, 0xf800 215; GFX9-NEXT: v_readlane_b32 s33, v40, 10 216; GFX9-NEXT: s_or_saveexec_b64 s[6:7], -1 217; GFX9-NEXT: buffer_load_dword v40, off, s[0:3], s32 offset:20 ; 4-byte Folded Reload 218; GFX9-NEXT: s_mov_b64 exec, s[6:7] 219; GFX9-NEXT: s_waitcnt vmcnt(0) 220; GFX9-NEXT: s_setpc_b64 s[4:5] 221; 222; GFX10-LABEL: call_preserved_vgpr_tuple8: 223; GFX10: ; %bb.0: ; %main_body 224; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) 225; GFX10-NEXT: s_waitcnt_vscnt null, 0x0 226; GFX10-NEXT: s_or_saveexec_b32 s4, -1 227; GFX10-NEXT: buffer_store_dword v40, off, s[0:3], s32 offset:20 ; 4-byte Folded Spill 228; GFX10-NEXT: s_waitcnt_depctr 0xffe3 229; GFX10-NEXT: s_mov_b32 exec_lo, s4 230; GFX10-NEXT: v_writelane_b32 v40, s33, 10 231; GFX10-NEXT: s_mov_b32 s33, s32 232; GFX10-NEXT: buffer_store_dword v41, off, s[0:3], s33 offset:16 ; 4-byte Folded Spill 233; GFX10-NEXT: buffer_store_dword v42, off, s[0:3], s33 offset:12 ; 4-byte Folded Spill 234; GFX10-NEXT: buffer_store_dword v43, off, s[0:3], s33 offset:8 ; 4-byte Folded Spill 235; GFX10-NEXT: buffer_store_dword v44, off, s[0:3], s33 offset:4 ; 4-byte Folded Spill 236; GFX10-NEXT: buffer_store_dword v45, off, s[0:3], s33 ; 4-byte Folded Spill 237; GFX10-NEXT: s_addk_i32 s32, 0x400 238; GFX10-NEXT: v_writelane_b32 v40, s36, 0 239; GFX10-NEXT: s_mov_b32 s36, 0 240; GFX10-NEXT: v_mov_b32_e32 v41, v16 241; GFX10-NEXT: v_mov_b32_e32 v42, v15 242; GFX10-NEXT: v_mov_b32_e32 v43, v14 243; GFX10-NEXT: v_writelane_b32 v40, s37, 1 244; GFX10-NEXT: s_mov_b32 s37, s36 245; GFX10-NEXT: v_mov_b32_e32 v44, v13 246; GFX10-NEXT: v_mov_b32_e32 v45, v12 247; GFX10-NEXT: v_writelane_b32 v40, s38, 2 248; GFX10-NEXT: s_mov_b32 s38, s36 249; GFX10-NEXT: v_writelane_b32 v40, s39, 3 250; GFX10-NEXT: s_mov_b32 s39, s36 251; GFX10-NEXT: v_writelane_b32 v40, s40, 4 252; GFX10-NEXT: s_mov_b32 s40, s36 253; GFX10-NEXT: v_writelane_b32 v40, s41, 5 254; GFX10-NEXT: s_mov_b32 s41, s36 255; GFX10-NEXT: v_writelane_b32 v40, s42, 6 256; GFX10-NEXT: s_mov_b32 s42, s36 257; GFX10-NEXT: v_writelane_b32 v40, s43, 7 258; GFX10-NEXT: s_mov_b32 s43, s36 259; GFX10-NEXT: image_gather4_c_b_cl v[0:3], v[12:16], s[36:43], s[4:7] dmask:0x1 dim:SQ_RSRC_IMG_2D 260; GFX10-NEXT: s_waitcnt_depctr 0xffe3 261; GFX10-NEXT: s_getpc_b64 s[4:5] 262; GFX10-NEXT: s_add_u32 s4, s4, extern_func@gotpcrel32@lo+4 263; GFX10-NEXT: s_addc_u32 s5, s5, extern_func@gotpcrel32@hi+12 264; GFX10-NEXT: v_writelane_b32 v40, s30, 8 265; GFX10-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0 266; GFX10-NEXT: v_writelane_b32 v40, s31, 9 267; GFX10-NEXT: s_waitcnt vmcnt(0) 268; GFX10-NEXT: global_store_dwordx4 v[0:1], v[0:3], off 269; GFX10-NEXT: s_waitcnt lgkmcnt(0) 270; GFX10-NEXT: s_swappc_b64 s[30:31], s[4:5] 271; GFX10-NEXT: image_gather4_c_b_cl v[0:3], [v45, v44, v43, v42, v41], s[36:43], s[4:7] dmask:0x1 dim:SQ_RSRC_IMG_2D 272; GFX10-NEXT: s_clause 0x4 273; GFX10-NEXT: buffer_load_dword v45, off, s[0:3], s33 274; GFX10-NEXT: buffer_load_dword v44, off, s[0:3], s33 offset:4 275; GFX10-NEXT: buffer_load_dword v43, off, s[0:3], s33 offset:8 276; GFX10-NEXT: buffer_load_dword v42, off, s[0:3], s33 offset:12 277; GFX10-NEXT: buffer_load_dword v41, off, s[0:3], s33 offset:16 278; GFX10-NEXT: v_readlane_b32 s4, v40, 8 279; GFX10-NEXT: v_readlane_b32 s5, v40, 9 280; GFX10-NEXT: v_readlane_b32 s43, v40, 7 281; GFX10-NEXT: v_readlane_b32 s42, v40, 6 282; GFX10-NEXT: v_readlane_b32 s41, v40, 5 283; GFX10-NEXT: v_readlane_b32 s40, v40, 4 284; GFX10-NEXT: v_readlane_b32 s39, v40, 3 285; GFX10-NEXT: v_readlane_b32 s38, v40, 2 286; GFX10-NEXT: v_readlane_b32 s37, v40, 1 287; GFX10-NEXT: v_readlane_b32 s36, v40, 0 288; GFX10-NEXT: s_addk_i32 s32, 0xfc00 289; GFX10-NEXT: v_readlane_b32 s33, v40, 10 290; GFX10-NEXT: s_or_saveexec_b32 s6, -1 291; GFX10-NEXT: buffer_load_dword v40, off, s[0:3], s32 offset:20 ; 4-byte Folded Reload 292; GFX10-NEXT: s_waitcnt_depctr 0xffe3 293; GFX10-NEXT: s_mov_b32 exec_lo, s6 294; GFX10-NEXT: s_waitcnt vmcnt(0) 295; GFX10-NEXT: s_setpc_b64 s[4:5] 296main_body: 297 %v = call <4 x float> @llvm.amdgcn.image.gather4.c.b.cl.2d.v4f32.f32.f32(i32 1, float %bias, float %zcompare, float %s, float %t, float %clamp, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) 298 store <4 x float> %v, <4 x float> addrspace(1)* undef 299 call void @extern_func() 300 %v1 = call <4 x float> @llvm.amdgcn.image.gather4.c.b.cl.2d.v4f32.f32.f32(i32 1, float %bias, float %zcompare, float %s, float %t, float %clamp, <8 x i32> undef, <4 x i32> undef, i1 false, i32 0, i32 0) 301 ret <4 x float> %v1 302} 303 304declare <4 x float> @llvm.amdgcn.image.gather4.c.b.cl.2d.v4f32.f32.f32(i32 immarg, float, float, float, float, float, <8 x i32>, <4 x i32>, i1 immarg, i32 immarg, i32 immarg) #1 305 306attributes #0 = { nounwind writeonly } 307attributes #1 = { nounwind readonly } 308attributes #2 = { "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" } 309