1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX9 %s 3; RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10 %s 4 5define amdgpu_ps <4 x float> @sample_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) { 6; GFX9-LABEL: sample_1d: 7; GFX9: ; %bb.0: ; %main_body 8; GFX9-NEXT: s_mov_b64 s[12:13], exec 9; GFX9-NEXT: s_wqm_b64 exec, exec 10; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 11; GFX9-NEXT: image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 12; GFX9-NEXT: s_waitcnt vmcnt(0) 13; GFX9-NEXT: ; return to shader part epilog 14; 15; GFX10-LABEL: sample_1d: 16; GFX10: ; %bb.0: ; %main_body 17; GFX10-NEXT: s_mov_b32 s12, exec_lo 18; GFX10-NEXT: ; implicit-def: $vcc_hi 19; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 20; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 21; GFX10-NEXT: image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 22; GFX10-NEXT: s_waitcnt vmcnt(0) 23; GFX10-NEXT: ; return to shader part epilog 24main_body: 25 %v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 26 ret <4 x float> %v 27} 28 29define amdgpu_ps <4 x float> @sample_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) { 30; GFX9-LABEL: sample_2d: 31; GFX9: ; %bb.0: ; %main_body 32; GFX9-NEXT: s_mov_b64 s[12:13], exec 33; GFX9-NEXT: s_wqm_b64 exec, exec 34; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 35; GFX9-NEXT: v_lshl_or_b32 v0, v1, 16, v0 36; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 37; GFX9-NEXT: image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 38; GFX9-NEXT: s_waitcnt vmcnt(0) 39; GFX9-NEXT: ; return to shader part epilog 40; 41; GFX10-LABEL: sample_2d: 42; GFX10: ; %bb.0: ; %main_body 43; GFX10-NEXT: s_mov_b32 s12, exec_lo 44; GFX10-NEXT: ; implicit-def: $vcc_hi 45; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 46; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 47; GFX10-NEXT: v_lshl_or_b32 v0, v1, 16, v0 48; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 49; GFX10-NEXT: image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 50; GFX10-NEXT: s_waitcnt vmcnt(0) 51; GFX10-NEXT: ; return to shader part epilog 52main_body: 53 %v = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 54 ret <4 x float> %v 55} 56 57define amdgpu_ps <4 x float> @sample_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %r) { 58; GFX9-LABEL: sample_3d: 59; GFX9: ; %bb.0: ; %main_body 60; GFX9-NEXT: s_mov_b64 s[12:13], exec 61; GFX9-NEXT: s_wqm_b64 exec, exec 62; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 63; GFX9-NEXT: v_lshl_or_b32 v1, v1, 16, v0 64; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 65; GFX9-NEXT: image_sample v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf a16 66; GFX9-NEXT: s_waitcnt vmcnt(0) 67; GFX9-NEXT: ; return to shader part epilog 68; 69; GFX10-LABEL: sample_3d: 70; GFX10: ; %bb.0: ; %main_body 71; GFX10-NEXT: s_mov_b32 s12, exec_lo 72; GFX10-NEXT: ; implicit-def: $vcc_hi 73; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 74; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 75; GFX10-NEXT: v_lshl_or_b32 v1, v1, 16, v0 76; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 77; GFX10-NEXT: image_sample v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_3D a16 78; GFX10-NEXT: s_waitcnt vmcnt(0) 79; GFX10-NEXT: ; return to shader part epilog 80main_body: 81 %v = call <4 x float> @llvm.amdgcn.image.sample.3d.v4f32.f16(i32 15, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 82 ret <4 x float> %v 83} 84 85define amdgpu_ps <4 x float> @sample_cube(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %face) { 86; GFX9-LABEL: sample_cube: 87; GFX9: ; %bb.0: ; %main_body 88; GFX9-NEXT: s_mov_b64 s[12:13], exec 89; GFX9-NEXT: s_wqm_b64 exec, exec 90; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 91; GFX9-NEXT: v_lshl_or_b32 v1, v1, 16, v0 92; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 93; GFX9-NEXT: image_sample v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf a16 da 94; GFX9-NEXT: s_waitcnt vmcnt(0) 95; GFX9-NEXT: ; return to shader part epilog 96; 97; GFX10-LABEL: sample_cube: 98; GFX10: ; %bb.0: ; %main_body 99; GFX10-NEXT: s_mov_b32 s12, exec_lo 100; GFX10-NEXT: ; implicit-def: $vcc_hi 101; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 102; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 103; GFX10-NEXT: v_lshl_or_b32 v1, v1, 16, v0 104; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 105; GFX10-NEXT: image_sample v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_CUBE a16 106; GFX10-NEXT: s_waitcnt vmcnt(0) 107; GFX10-NEXT: ; return to shader part epilog 108main_body: 109 %v = call <4 x float> @llvm.amdgcn.image.sample.cube.v4f32.f16(i32 15, half %s, half %t, half %face, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 110 ret <4 x float> %v 111} 112 113define amdgpu_ps <4 x float> @sample_1darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %slice) { 114; GFX9-LABEL: sample_1darray: 115; GFX9: ; %bb.0: ; %main_body 116; GFX9-NEXT: s_mov_b64 s[12:13], exec 117; GFX9-NEXT: s_wqm_b64 exec, exec 118; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 119; GFX9-NEXT: v_lshl_or_b32 v0, v1, 16, v0 120; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 121; GFX9-NEXT: image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 da 122; GFX9-NEXT: s_waitcnt vmcnt(0) 123; GFX9-NEXT: ; return to shader part epilog 124; 125; GFX10-LABEL: sample_1darray: 126; GFX10: ; %bb.0: ; %main_body 127; GFX10-NEXT: s_mov_b32 s12, exec_lo 128; GFX10-NEXT: ; implicit-def: $vcc_hi 129; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 130; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 131; GFX10-NEXT: v_lshl_or_b32 v0, v1, 16, v0 132; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 133; GFX10-NEXT: image_sample v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D_ARRAY a16 134; GFX10-NEXT: s_waitcnt vmcnt(0) 135; GFX10-NEXT: ; return to shader part epilog 136main_body: 137 %v = call <4 x float> @llvm.amdgcn.image.sample.1darray.v4f32.f16(i32 15, half %s, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 138 ret <4 x float> %v 139} 140 141define amdgpu_ps <4 x float> @sample_2darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %slice) { 142; GFX9-LABEL: sample_2darray: 143; GFX9: ; %bb.0: ; %main_body 144; GFX9-NEXT: s_mov_b64 s[12:13], exec 145; GFX9-NEXT: s_wqm_b64 exec, exec 146; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 147; GFX9-NEXT: v_lshl_or_b32 v1, v1, 16, v0 148; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 149; GFX9-NEXT: image_sample v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf a16 da 150; GFX9-NEXT: s_waitcnt vmcnt(0) 151; GFX9-NEXT: ; return to shader part epilog 152; 153; GFX10-LABEL: sample_2darray: 154; GFX10: ; %bb.0: ; %main_body 155; GFX10-NEXT: s_mov_b32 s12, exec_lo 156; GFX10-NEXT: ; implicit-def: $vcc_hi 157; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 158; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 159; GFX10-NEXT: v_lshl_or_b32 v1, v1, 16, v0 160; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 161; GFX10-NEXT: image_sample v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D_ARRAY a16 162; GFX10-NEXT: s_waitcnt vmcnt(0) 163; GFX10-NEXT: ; return to shader part epilog 164main_body: 165 %v = call <4 x float> @llvm.amdgcn.image.sample.2darray.v4f32.f16(i32 15, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 166 ret <4 x float> %v 167} 168 169define amdgpu_ps <4 x float> @sample_c_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) { 170; GFX9-LABEL: sample_c_1d: 171; GFX9: ; %bb.0: ; %main_body 172; GFX9-NEXT: s_mov_b64 s[12:13], exec 173; GFX9-NEXT: s_wqm_b64 exec, exec 174; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 175; GFX9-NEXT: image_sample_c v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 176; GFX9-NEXT: s_waitcnt vmcnt(0) 177; GFX9-NEXT: ; return to shader part epilog 178; 179; GFX10-LABEL: sample_c_1d: 180; GFX10: ; %bb.0: ; %main_body 181; GFX10-NEXT: s_mov_b32 s12, exec_lo 182; GFX10-NEXT: ; implicit-def: $vcc_hi 183; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 184; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 185; GFX10-NEXT: image_sample_c v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 186; GFX10-NEXT: s_waitcnt vmcnt(0) 187; GFX10-NEXT: ; return to shader part epilog 188main_body: 189 %v = call <4 x float> @llvm.amdgcn.image.sample.c.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 190 ret <4 x float> %v 191} 192 193define amdgpu_ps <4 x float> @sample_c_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) { 194; GFX9-LABEL: sample_c_2d: 195; GFX9: ; %bb.0: ; %main_body 196; GFX9-NEXT: s_mov_b64 s[12:13], exec 197; GFX9-NEXT: s_wqm_b64 exec, exec 198; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v1 199; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1 200; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 201; GFX9-NEXT: image_sample_c v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 202; GFX9-NEXT: s_waitcnt vmcnt(0) 203; GFX9-NEXT: ; return to shader part epilog 204; 205; GFX10-LABEL: sample_c_2d: 206; GFX10: ; %bb.0: ; %main_body 207; GFX10-NEXT: s_mov_b32 s12, exec_lo 208; GFX10-NEXT: ; implicit-def: $vcc_hi 209; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 210; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 211; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 212; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 213; GFX10-NEXT: image_sample_c v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 214; GFX10-NEXT: s_waitcnt vmcnt(0) 215; GFX10-NEXT: ; return to shader part epilog 216main_body: 217 %v = call <4 x float> @llvm.amdgcn.image.sample.c.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 218 ret <4 x float> %v 219} 220 221define amdgpu_ps <4 x float> @sample_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %clamp) { 222; GFX9-LABEL: sample_cl_1d: 223; GFX9: ; %bb.0: ; %main_body 224; GFX9-NEXT: s_mov_b64 s[12:13], exec 225; GFX9-NEXT: s_wqm_b64 exec, exec 226; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 227; GFX9-NEXT: v_lshl_or_b32 v0, v1, 16, v0 228; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 229; GFX9-NEXT: image_sample_cl v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 230; GFX9-NEXT: s_waitcnt vmcnt(0) 231; GFX9-NEXT: ; return to shader part epilog 232; 233; GFX10-LABEL: sample_cl_1d: 234; GFX10: ; %bb.0: ; %main_body 235; GFX10-NEXT: s_mov_b32 s12, exec_lo 236; GFX10-NEXT: ; implicit-def: $vcc_hi 237; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 238; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 239; GFX10-NEXT: v_lshl_or_b32 v0, v1, 16, v0 240; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 241; GFX10-NEXT: image_sample_cl v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 242; GFX10-NEXT: s_waitcnt vmcnt(0) 243; GFX10-NEXT: ; return to shader part epilog 244main_body: 245 %v = call <4 x float> @llvm.amdgcn.image.sample.cl.1d.v4f32.f16(i32 15, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 246 ret <4 x float> %v 247} 248 249define amdgpu_ps <4 x float> @sample_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %clamp) { 250; GFX9-LABEL: sample_cl_2d: 251; GFX9: ; %bb.0: ; %main_body 252; GFX9-NEXT: s_mov_b64 s[12:13], exec 253; GFX9-NEXT: s_wqm_b64 exec, exec 254; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 255; GFX9-NEXT: v_lshl_or_b32 v1, v1, 16, v0 256; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 257; GFX9-NEXT: image_sample_cl v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf a16 258; GFX9-NEXT: s_waitcnt vmcnt(0) 259; GFX9-NEXT: ; return to shader part epilog 260; 261; GFX10-LABEL: sample_cl_2d: 262; GFX10: ; %bb.0: ; %main_body 263; GFX10-NEXT: s_mov_b32 s12, exec_lo 264; GFX10-NEXT: ; implicit-def: $vcc_hi 265; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 266; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 267; GFX10-NEXT: v_lshl_or_b32 v1, v1, 16, v0 268; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 269; GFX10-NEXT: image_sample_cl v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 270; GFX10-NEXT: s_waitcnt vmcnt(0) 271; GFX10-NEXT: ; return to shader part epilog 272main_body: 273 %v = call <4 x float> @llvm.amdgcn.image.sample.cl.2d.v4f32.f16(i32 15, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 274 ret <4 x float> %v 275} 276 277define amdgpu_ps <4 x float> @sample_c_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %clamp) { 278; GFX9-LABEL: sample_c_cl_1d: 279; GFX9: ; %bb.0: ; %main_body 280; GFX9-NEXT: s_mov_b64 s[12:13], exec 281; GFX9-NEXT: s_wqm_b64 exec, exec 282; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v1 283; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1 284; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 285; GFX9-NEXT: image_sample_c_cl v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 286; GFX9-NEXT: s_waitcnt vmcnt(0) 287; GFX9-NEXT: ; return to shader part epilog 288; 289; GFX10-LABEL: sample_c_cl_1d: 290; GFX10: ; %bb.0: ; %main_body 291; GFX10-NEXT: s_mov_b32 s12, exec_lo 292; GFX10-NEXT: ; implicit-def: $vcc_hi 293; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 294; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 295; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 296; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 297; GFX10-NEXT: image_sample_c_cl v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 298; GFX10-NEXT: s_waitcnt vmcnt(0) 299; GFX10-NEXT: ; return to shader part epilog 300main_body: 301 %v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 302 ret <4 x float> %v 303} 304 305define amdgpu_ps <4 x float> @sample_c_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %clamp) { 306; GFX9-LABEL: sample_c_cl_2d: 307; GFX9: ; %bb.0: ; %main_body 308; GFX9-NEXT: s_mov_b64 s[12:13], exec 309; GFX9-NEXT: s_wqm_b64 exec, exec 310; GFX9-NEXT: v_mov_b32_e32 v5, v3 311; GFX9-NEXT: v_mov_b32_e32 v3, v0 312; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v1 313; GFX9-NEXT: v_lshl_or_b32 v4, v2, 16, v0 314; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 315; GFX9-NEXT: image_sample_c_cl v[0:3], v[3:5], s[0:7], s[8:11] dmask:0xf a16 316; GFX9-NEXT: s_waitcnt vmcnt(0) 317; GFX9-NEXT: ; return to shader part epilog 318; 319; GFX10-LABEL: sample_c_cl_2d: 320; GFX10: ; %bb.0: ; %main_body 321; GFX10-NEXT: s_mov_b32 s12, exec_lo 322; GFX10-NEXT: ; implicit-def: $vcc_hi 323; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 324; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 325; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 326; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 327; GFX10-NEXT: image_sample_c_cl v[0:3], [v0, v1, v3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 328; GFX10-NEXT: s_waitcnt vmcnt(0) 329; GFX10-NEXT: ; return to shader part epilog 330main_body: 331 %v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 332 ret <4 x float> %v 333} 334 335define amdgpu_ps <4 x float> @sample_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, half %s) { 336; GFX9-LABEL: sample_b_1d: 337; GFX9: ; %bb.0: ; %main_body 338; GFX9-NEXT: s_mov_b64 s[12:13], exec 339; GFX9-NEXT: s_wqm_b64 exec, exec 340; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 341; GFX9-NEXT: image_sample_b v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 342; GFX9-NEXT: s_waitcnt vmcnt(0) 343; GFX9-NEXT: ; return to shader part epilog 344; 345; GFX10-LABEL: sample_b_1d: 346; GFX10: ; %bb.0: ; %main_body 347; GFX10-NEXT: s_mov_b32 s12, exec_lo 348; GFX10-NEXT: ; implicit-def: $vcc_hi 349; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 350; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 351; GFX10-NEXT: image_sample_b v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 352; GFX10-NEXT: s_waitcnt vmcnt(0) 353; GFX10-NEXT: ; return to shader part epilog 354main_body: 355 %v = call <4 x float> @llvm.amdgcn.image.sample.b.1d.v4f32.f32.f16(i32 15, float %bias, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 356 ret <4 x float> %v 357} 358 359define amdgpu_ps <4 x float> @sample_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, half %s, half %t) { 360; GFX9-LABEL: sample_b_2d: 361; GFX9: ; %bb.0: ; %main_body 362; GFX9-NEXT: s_mov_b64 s[12:13], exec 363; GFX9-NEXT: s_wqm_b64 exec, exec 364; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v1 365; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1 366; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 367; GFX9-NEXT: image_sample_b v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 368; GFX9-NEXT: s_waitcnt vmcnt(0) 369; GFX9-NEXT: ; return to shader part epilog 370; 371; GFX10-LABEL: sample_b_2d: 372; GFX10: ; %bb.0: ; %main_body 373; GFX10-NEXT: s_mov_b32 s12, exec_lo 374; GFX10-NEXT: ; implicit-def: $vcc_hi 375; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 376; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 377; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 378; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 379; GFX10-NEXT: image_sample_b v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 380; GFX10-NEXT: s_waitcnt vmcnt(0) 381; GFX10-NEXT: ; return to shader part epilog 382main_body: 383 %v = call <4 x float> @llvm.amdgcn.image.sample.b.2d.v4f32.f32.f16(i32 15, float %bias, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 384 ret <4 x float> %v 385} 386 387define amdgpu_ps <4 x float> @sample_c_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, float %zcompare, half %s) { 388; GFX9-LABEL: sample_c_b_1d: 389; GFX9: ; %bb.0: ; %main_body 390; GFX9-NEXT: s_mov_b64 s[12:13], exec 391; GFX9-NEXT: s_wqm_b64 exec, exec 392; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 393; GFX9-NEXT: image_sample_c_b v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 394; GFX9-NEXT: s_waitcnt vmcnt(0) 395; GFX9-NEXT: ; return to shader part epilog 396; 397; GFX10-LABEL: sample_c_b_1d: 398; GFX10: ; %bb.0: ; %main_body 399; GFX10-NEXT: s_mov_b32 s12, exec_lo 400; GFX10-NEXT: ; implicit-def: $vcc_hi 401; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 402; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 403; GFX10-NEXT: image_sample_c_b v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 404; GFX10-NEXT: s_waitcnt vmcnt(0) 405; GFX10-NEXT: ; return to shader part epilog 406main_body: 407 %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.1d.v4f32.f32.f16(i32 15, float %bias, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 408 ret <4 x float> %v 409} 410 411define amdgpu_ps <4 x float> @sample_c_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, float %zcompare, half %s, half %t) { 412; GFX9-LABEL: sample_c_b_2d: 413; GFX9: ; %bb.0: ; %main_body 414; GFX9-NEXT: s_mov_b64 s[12:13], exec 415; GFX9-NEXT: s_wqm_b64 exec, exec 416; GFX9-NEXT: v_and_b32_e32 v2, 0xffff, v2 417; GFX9-NEXT: v_lshl_or_b32 v2, v3, 16, v2 418; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 419; GFX9-NEXT: image_sample_c_b v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 420; GFX9-NEXT: s_waitcnt vmcnt(0) 421; GFX9-NEXT: ; return to shader part epilog 422; 423; GFX10-LABEL: sample_c_b_2d: 424; GFX10: ; %bb.0: ; %main_body 425; GFX10-NEXT: s_mov_b32 s12, exec_lo 426; GFX10-NEXT: ; implicit-def: $vcc_hi 427; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 428; GFX10-NEXT: v_and_b32_e32 v2, 0xffff, v2 429; GFX10-NEXT: v_lshl_or_b32 v2, v3, 16, v2 430; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 431; GFX10-NEXT: image_sample_c_b v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 432; GFX10-NEXT: s_waitcnt vmcnt(0) 433; GFX10-NEXT: ; return to shader part epilog 434main_body: 435 %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.2d.v4f32.f32.f16(i32 15, float %bias, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 436 ret <4 x float> %v 437} 438 439define amdgpu_ps <4 x float> @sample_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, half %s, half %clamp) { 440; GFX9-LABEL: sample_b_cl_1d: 441; GFX9: ; %bb.0: ; %main_body 442; GFX9-NEXT: s_mov_b64 s[12:13], exec 443; GFX9-NEXT: s_wqm_b64 exec, exec 444; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v1 445; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1 446; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 447; GFX9-NEXT: image_sample_b_cl v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 448; GFX9-NEXT: s_waitcnt vmcnt(0) 449; GFX9-NEXT: ; return to shader part epilog 450; 451; GFX10-LABEL: sample_b_cl_1d: 452; GFX10: ; %bb.0: ; %main_body 453; GFX10-NEXT: s_mov_b32 s12, exec_lo 454; GFX10-NEXT: ; implicit-def: $vcc_hi 455; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 456; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 457; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 458; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 459; GFX10-NEXT: image_sample_b_cl v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 460; GFX10-NEXT: s_waitcnt vmcnt(0) 461; GFX10-NEXT: ; return to shader part epilog 462main_body: 463 %v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.1d.v4f32.f32.f16(i32 15, float %bias, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 464 ret <4 x float> %v 465} 466 467define amdgpu_ps <4 x float> @sample_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, half %s, half %t, half %clamp) { 468; GFX9-LABEL: sample_b_cl_2d: 469; GFX9: ; %bb.0: ; %main_body 470; GFX9-NEXT: s_mov_b64 s[12:13], exec 471; GFX9-NEXT: s_wqm_b64 exec, exec 472; GFX9-NEXT: v_mov_b32_e32 v5, v3 473; GFX9-NEXT: v_mov_b32_e32 v3, v0 474; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v1 475; GFX9-NEXT: v_lshl_or_b32 v4, v2, 16, v0 476; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 477; GFX9-NEXT: image_sample_b_cl v[0:3], v[3:5], s[0:7], s[8:11] dmask:0xf a16 478; GFX9-NEXT: s_waitcnt vmcnt(0) 479; GFX9-NEXT: ; return to shader part epilog 480; 481; GFX10-LABEL: sample_b_cl_2d: 482; GFX10: ; %bb.0: ; %main_body 483; GFX10-NEXT: s_mov_b32 s12, exec_lo 484; GFX10-NEXT: ; implicit-def: $vcc_hi 485; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 486; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 487; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 488; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 489; GFX10-NEXT: image_sample_b_cl v[0:3], [v0, v1, v3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 490; GFX10-NEXT: s_waitcnt vmcnt(0) 491; GFX10-NEXT: ; return to shader part epilog 492main_body: 493 %v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.2d.v4f32.f32.f16(i32 15, float %bias, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 494 ret <4 x float> %v 495} 496 497define amdgpu_ps <4 x float> @sample_c_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, float %zcompare, half %s, half %clamp) { 498; GFX9-LABEL: sample_c_b_cl_1d: 499; GFX9: ; %bb.0: ; %main_body 500; GFX9-NEXT: s_mov_b64 s[12:13], exec 501; GFX9-NEXT: s_wqm_b64 exec, exec 502; GFX9-NEXT: v_and_b32_e32 v2, 0xffff, v2 503; GFX9-NEXT: v_lshl_or_b32 v2, v3, 16, v2 504; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 505; GFX9-NEXT: image_sample_c_b_cl v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 506; GFX9-NEXT: s_waitcnt vmcnt(0) 507; GFX9-NEXT: ; return to shader part epilog 508; 509; GFX10-LABEL: sample_c_b_cl_1d: 510; GFX10: ; %bb.0: ; %main_body 511; GFX10-NEXT: s_mov_b32 s12, exec_lo 512; GFX10-NEXT: ; implicit-def: $vcc_hi 513; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 514; GFX10-NEXT: v_and_b32_e32 v2, 0xffff, v2 515; GFX10-NEXT: v_lshl_or_b32 v2, v3, 16, v2 516; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 517; GFX10-NEXT: image_sample_c_b_cl v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 518; GFX10-NEXT: s_waitcnt vmcnt(0) 519; GFX10-NEXT: ; return to shader part epilog 520main_body: 521 %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.1d.v4f32.f32.f16(i32 15, float %bias, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 522 ret <4 x float> %v 523} 524 525define amdgpu_ps <4 x float> @sample_c_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %bias, float %zcompare, half %s, half %t, half %clamp) { 526; GFX9-LABEL: sample_c_b_cl_2d: 527; GFX9: ; %bb.0: ; %main_body 528; GFX9-NEXT: s_mov_b64 s[12:13], exec 529; GFX9-NEXT: s_wqm_b64 exec, exec 530; GFX9-NEXT: v_mov_b32_e32 v7, v4 531; GFX9-NEXT: v_mov_b32_e32 v4, v0 532; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v2 533; GFX9-NEXT: v_mov_b32_e32 v5, v1 534; GFX9-NEXT: v_lshl_or_b32 v6, v3, 16, v0 535; GFX9-NEXT: s_and_b64 exec, exec, s[12:13] 536; GFX9-NEXT: image_sample_c_b_cl v[0:3], v[4:7], s[0:7], s[8:11] dmask:0xf a16 537; GFX9-NEXT: s_waitcnt vmcnt(0) 538; GFX9-NEXT: ; return to shader part epilog 539; 540; GFX10-LABEL: sample_c_b_cl_2d: 541; GFX10: ; %bb.0: ; %main_body 542; GFX10-NEXT: s_mov_b32 s12, exec_lo 543; GFX10-NEXT: ; implicit-def: $vcc_hi 544; GFX10-NEXT: s_wqm_b32 exec_lo, exec_lo 545; GFX10-NEXT: v_and_b32_e32 v2, 0xffff, v2 546; GFX10-NEXT: v_lshl_or_b32 v2, v3, 16, v2 547; GFX10-NEXT: s_and_b32 exec_lo, exec_lo, s12 548; GFX10-NEXT: image_sample_c_b_cl v[0:3], [v0, v1, v2, v4], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 549; GFX10-NEXT: s_waitcnt vmcnt(0) 550; GFX10-NEXT: ; return to shader part epilog 551main_body: 552 %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.2d.v4f32.f32.f16(i32 15, float %bias, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 553 ret <4 x float> %v 554} 555 556define amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) { 557; GFX9-LABEL: sample_d_1d: 558; GFX9: ; %bb.0: ; %main_body 559; GFX9-NEXT: image_sample_d v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 560; GFX9-NEXT: s_waitcnt vmcnt(0) 561; GFX9-NEXT: ; return to shader part epilog 562; 563; GFX10-LABEL: sample_d_1d: 564; GFX10: ; %bb.0: ; %main_body 565; GFX10-NEXT: image_sample_d v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 566; GFX10-NEXT: ; implicit-def: $vcc_hi 567; GFX10-NEXT: s_waitcnt vmcnt(0) 568; GFX10-NEXT: ; return to shader part epilog 569main_body: 570 %v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 571 ret <4 x float> %v 572} 573 574define amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) { 575; GFX9-LABEL: sample_d_2d: 576; GFX9: ; %bb.0: ; %main_body 577; GFX9-NEXT: v_mov_b32_e32 v6, 0xffff 578; GFX9-NEXT: v_and_b32_e32 v4, v6, v4 579; GFX9-NEXT: v_and_b32_e32 v2, v6, v2 580; GFX9-NEXT: v_and_b32_e32 v0, v6, v0 581; GFX9-NEXT: v_lshl_or_b32 v3, v3, 16, v2 582; GFX9-NEXT: v_lshl_or_b32 v4, v5, 16, v4 583; GFX9-NEXT: v_lshl_or_b32 v2, v1, 16, v0 584; GFX9-NEXT: image_sample_d v[0:3], v[2:4], s[0:7], s[8:11] dmask:0xf a16 585; GFX9-NEXT: s_waitcnt vmcnt(0) 586; GFX9-NEXT: ; return to shader part epilog 587; 588; GFX10-LABEL: sample_d_2d: 589; GFX10: ; %bb.0: ; %main_body 590; GFX10-NEXT: v_mov_b32_e32 v7, 0xffff 591; GFX10-NEXT: ; implicit-def: $vcc_hi 592; GFX10-NEXT: v_and_b32_e32 v4, v7, v4 593; GFX10-NEXT: v_and_b32_e32 v2, v7, v2 594; GFX10-NEXT: v_and_b32_e32 v0, v7, v0 595; GFX10-NEXT: v_lshl_or_b32 v4, v5, 16, v4 596; GFX10-NEXT: v_lshl_or_b32 v3, v3, 16, v2 597; GFX10-NEXT: v_lshl_or_b32 v2, v1, 16, v0 598; GFX10-NEXT: image_sample_d v[0:3], v[2:4], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 599; GFX10-NEXT: s_waitcnt vmcnt(0) 600; GFX10-NEXT: ; return to shader part epilog 601main_body: 602 %v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 603 ret <4 x float> %v 604} 605 606define amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) { 607; GFX9-LABEL: sample_d_3d: 608; GFX9: ; %bb.0: ; %main_body 609; GFX9-NEXT: v_mov_b32_e32 v12, v8 610; GFX9-NEXT: v_mov_b32_e32 v8, v2 611; GFX9-NEXT: v_mov_b32_e32 v2, 0xffff 612; GFX9-NEXT: v_mov_b32_e32 v10, v5 613; GFX9-NEXT: v_and_b32_e32 v5, v2, v6 614; GFX9-NEXT: v_and_b32_e32 v3, v2, v3 615; GFX9-NEXT: v_and_b32_e32 v0, v2, v0 616; GFX9-NEXT: v_lshl_or_b32 v11, v7, 16, v5 617; GFX9-NEXT: v_lshl_or_b32 v9, v4, 16, v3 618; GFX9-NEXT: v_lshl_or_b32 v7, v1, 16, v0 619; GFX9-NEXT: image_sample_d v[0:3], v[7:14], s[0:7], s[8:11] dmask:0xf a16 620; GFX9-NEXT: s_waitcnt vmcnt(0) 621; GFX9-NEXT: ; return to shader part epilog 622; 623; GFX10-LABEL: sample_d_3d: 624; GFX10: ; %bb.0: ; %main_body 625; GFX10-NEXT: v_mov_b32_e32 v9, 0xffff 626; GFX10-NEXT: ; implicit-def: $vcc_hi 627; GFX10-NEXT: v_and_b32_e32 v6, v9, v6 628; GFX10-NEXT: v_and_b32_e32 v3, v9, v3 629; GFX10-NEXT: v_and_b32_e32 v0, v9, v0 630; GFX10-NEXT: v_lshl_or_b32 v6, v7, 16, v6 631; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 632; GFX10-NEXT: v_lshl_or_b32 v0, v1, 16, v0 633; GFX10-NEXT: image_sample_d v[0:3], [v0, v2, v3, v5, v6, v8], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_3D a16 634; GFX10-NEXT: s_waitcnt vmcnt(0) 635; GFX10-NEXT: ; return to shader part epilog 636main_body: 637 %v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 638 ret <4 x float> %v 639} 640 641define amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) { 642; GFX9-LABEL: sample_c_d_1d: 643; GFX9: ; %bb.0: ; %main_body 644; GFX9-NEXT: image_sample_c_d v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf a16 645; GFX9-NEXT: s_waitcnt vmcnt(0) 646; GFX9-NEXT: ; return to shader part epilog 647; 648; GFX10-LABEL: sample_c_d_1d: 649; GFX10: ; %bb.0: ; %main_body 650; GFX10-NEXT: image_sample_c_d v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 651; GFX10-NEXT: ; implicit-def: $vcc_hi 652; GFX10-NEXT: s_waitcnt vmcnt(0) 653; GFX10-NEXT: ; return to shader part epilog 654main_body: 655 %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 656 ret <4 x float> %v 657} 658 659define amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) { 660; GFX9-LABEL: sample_c_d_2d: 661; GFX9: ; %bb.0: ; %main_body 662; GFX9-NEXT: v_mov_b32_e32 v9, 0xffff 663; GFX9-NEXT: v_mov_b32_e32 v8, v2 664; GFX9-NEXT: v_mov_b32_e32 v7, v3 665; GFX9-NEXT: v_and_b32_e32 v2, v9, v5 666; GFX9-NEXT: v_and_b32_e32 v1, v9, v1 667; GFX9-NEXT: v_lshl_or_b32 v3, v6, 16, v2 668; GFX9-NEXT: v_and_b32_e32 v2, v9, v7 669; GFX9-NEXT: v_lshl_or_b32 v2, v4, 16, v2 670; GFX9-NEXT: v_lshl_or_b32 v1, v8, 16, v1 671; GFX9-NEXT: image_sample_c_d v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf a16 672; GFX9-NEXT: s_waitcnt vmcnt(0) 673; GFX9-NEXT: ; return to shader part epilog 674; 675; GFX10-LABEL: sample_c_d_2d: 676; GFX10: ; %bb.0: ; %main_body 677; GFX10-NEXT: v_mov_b32_e32 v10, 0xffff 678; GFX10-NEXT: ; implicit-def: $vcc_hi 679; GFX10-NEXT: v_and_b32_e32 v5, v10, v5 680; GFX10-NEXT: v_and_b32_e32 v3, v10, v3 681; GFX10-NEXT: v_and_b32_e32 v1, v10, v1 682; GFX10-NEXT: v_lshl_or_b32 v5, v6, 16, v5 683; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 684; GFX10-NEXT: v_lshl_or_b32 v2, v2, 16, v1 685; GFX10-NEXT: image_sample_c_d v[0:3], [v0, v2, v3, v5], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 686; GFX10-NEXT: s_waitcnt vmcnt(0) 687; GFX10-NEXT: ; return to shader part epilog 688main_body: 689 %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 690 ret <4 x float> %v 691} 692 693define amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) { 694; GFX9-LABEL: sample_d_cl_1d: 695; GFX9: ; %bb.0: ; %main_body 696; GFX9-NEXT: v_and_b32_e32 v2, 0xffff, v2 697; GFX9-NEXT: v_lshl_or_b32 v2, v3, 16, v2 698; GFX9-NEXT: image_sample_d_cl v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 699; GFX9-NEXT: s_waitcnt vmcnt(0) 700; GFX9-NEXT: ; return to shader part epilog 701; 702; GFX10-LABEL: sample_d_cl_1d: 703; GFX10: ; %bb.0: ; %main_body 704; GFX10-NEXT: v_and_b32_e32 v2, 0xffff, v2 705; GFX10-NEXT: ; implicit-def: $vcc_hi 706; GFX10-NEXT: v_lshl_or_b32 v2, v3, 16, v2 707; GFX10-NEXT: image_sample_d_cl v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 708; GFX10-NEXT: s_waitcnt vmcnt(0) 709; GFX10-NEXT: ; return to shader part epilog 710main_body: 711 %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 712 ret <4 x float> %v 713} 714 715define amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) { 716; GFX9-LABEL: sample_d_cl_2d: 717; GFX9: ; %bb.0: ; %main_body 718; GFX9-NEXT: v_mov_b32_e32 v7, 0xffff 719; GFX9-NEXT: v_and_b32_e32 v4, v7, v4 720; GFX9-NEXT: v_and_b32_e32 v2, v7, v2 721; GFX9-NEXT: v_and_b32_e32 v0, v7, v0 722; GFX9-NEXT: v_lshl_or_b32 v5, v5, 16, v4 723; GFX9-NEXT: v_lshl_or_b32 v4, v3, 16, v2 724; GFX9-NEXT: v_lshl_or_b32 v3, v1, 16, v0 725; GFX9-NEXT: image_sample_d_cl v[0:3], v[3:6], s[0:7], s[8:11] dmask:0xf a16 726; GFX9-NEXT: s_waitcnt vmcnt(0) 727; GFX9-NEXT: ; return to shader part epilog 728; 729; GFX10-LABEL: sample_d_cl_2d: 730; GFX10: ; %bb.0: ; %main_body 731; GFX10-NEXT: v_mov_b32_e32 v7, 0xffff 732; GFX10-NEXT: ; implicit-def: $vcc_hi 733; GFX10-NEXT: v_and_b32_e32 v4, v7, v4 734; GFX10-NEXT: v_and_b32_e32 v2, v7, v2 735; GFX10-NEXT: v_and_b32_e32 v0, v7, v0 736; GFX10-NEXT: v_lshl_or_b32 v4, v5, 16, v4 737; GFX10-NEXT: v_lshl_or_b32 v5, v3, 16, v2 738; GFX10-NEXT: v_lshl_or_b32 v3, v1, 16, v0 739; GFX10-NEXT: image_sample_d_cl v[0:3], [v3, v5, v4, v6], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 740; GFX10-NEXT: s_waitcnt vmcnt(0) 741; GFX10-NEXT: ; return to shader part epilog 742main_body: 743 %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 744 ret <4 x float> %v 745} 746 747define amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) { 748; GFX9-LABEL: sample_c_d_cl_1d: 749; GFX9: ; %bb.0: ; %main_body 750; GFX9-NEXT: v_and_b32_e32 v3, 0xffff, v3 751; GFX9-NEXT: v_lshl_or_b32 v3, v4, 16, v3 752; GFX9-NEXT: image_sample_c_d_cl v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf a16 753; GFX9-NEXT: s_waitcnt vmcnt(0) 754; GFX9-NEXT: ; return to shader part epilog 755; 756; GFX10-LABEL: sample_c_d_cl_1d: 757; GFX10: ; %bb.0: ; %main_body 758; GFX10-NEXT: v_and_b32_e32 v3, 0xffff, v3 759; GFX10-NEXT: ; implicit-def: $vcc_hi 760; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 761; GFX10-NEXT: image_sample_c_d_cl v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 762; GFX10-NEXT: s_waitcnt vmcnt(0) 763; GFX10-NEXT: ; return to shader part epilog 764main_body: 765 %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 766 ret <4 x float> %v 767} 768 769define amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) { 770; GFX9-LABEL: sample_c_d_cl_2d: 771; GFX9: ; %bb.0: ; %main_body 772; GFX9-NEXT: v_mov_b32_e32 v11, v7 773; GFX9-NEXT: v_mov_b32_e32 v7, v0 774; GFX9-NEXT: v_mov_b32_e32 v0, 0xffff 775; GFX9-NEXT: v_and_b32_e32 v5, v0, v5 776; GFX9-NEXT: v_and_b32_e32 v3, v0, v3 777; GFX9-NEXT: v_and_b32_e32 v0, v0, v1 778; GFX9-NEXT: v_lshl_or_b32 v10, v6, 16, v5 779; GFX9-NEXT: v_lshl_or_b32 v9, v4, 16, v3 780; GFX9-NEXT: v_lshl_or_b32 v8, v2, 16, v0 781; GFX9-NEXT: image_sample_c_d_cl v[0:3], v[7:14], s[0:7], s[8:11] dmask:0xf a16 782; GFX9-NEXT: s_waitcnt vmcnt(0) 783; GFX9-NEXT: ; return to shader part epilog 784; 785; GFX10-LABEL: sample_c_d_cl_2d: 786; GFX10: ; %bb.0: ; %main_body 787; GFX10-NEXT: v_mov_b32_e32 v8, 0xffff 788; GFX10-NEXT: ; implicit-def: $vcc_hi 789; GFX10-NEXT: v_and_b32_e32 v5, v8, v5 790; GFX10-NEXT: v_and_b32_e32 v3, v8, v3 791; GFX10-NEXT: v_and_b32_e32 v1, v8, v1 792; GFX10-NEXT: v_lshl_or_b32 v5, v6, 16, v5 793; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 794; GFX10-NEXT: v_lshl_or_b32 v2, v2, 16, v1 795; GFX10-NEXT: image_sample_c_d_cl v[0:3], [v0, v2, v3, v5, v7], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 796; GFX10-NEXT: s_waitcnt vmcnt(0) 797; GFX10-NEXT: ; return to shader part epilog 798main_body: 799 %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 800 ret <4 x float> %v 801} 802 803define amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) { 804; GFX9-LABEL: sample_cd_1d: 805; GFX9: ; %bb.0: ; %main_body 806; GFX9-NEXT: image_sample_cd v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 807; GFX9-NEXT: s_waitcnt vmcnt(0) 808; GFX9-NEXT: ; return to shader part epilog 809; 810; GFX10-LABEL: sample_cd_1d: 811; GFX10: ; %bb.0: ; %main_body 812; GFX10-NEXT: image_sample_cd v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 813; GFX10-NEXT: ; implicit-def: $vcc_hi 814; GFX10-NEXT: s_waitcnt vmcnt(0) 815; GFX10-NEXT: ; return to shader part epilog 816main_body: 817 %v = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 818 ret <4 x float> %v 819} 820 821define amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) { 822; GFX9-LABEL: sample_cd_2d: 823; GFX9: ; %bb.0: ; %main_body 824; GFX9-NEXT: v_mov_b32_e32 v6, 0xffff 825; GFX9-NEXT: v_and_b32_e32 v4, v6, v4 826; GFX9-NEXT: v_and_b32_e32 v2, v6, v2 827; GFX9-NEXT: v_and_b32_e32 v0, v6, v0 828; GFX9-NEXT: v_lshl_or_b32 v3, v3, 16, v2 829; GFX9-NEXT: v_lshl_or_b32 v4, v5, 16, v4 830; GFX9-NEXT: v_lshl_or_b32 v2, v1, 16, v0 831; GFX9-NEXT: image_sample_cd v[0:3], v[2:4], s[0:7], s[8:11] dmask:0xf a16 832; GFX9-NEXT: s_waitcnt vmcnt(0) 833; GFX9-NEXT: ; return to shader part epilog 834; 835; GFX10-LABEL: sample_cd_2d: 836; GFX10: ; %bb.0: ; %main_body 837; GFX10-NEXT: v_mov_b32_e32 v7, 0xffff 838; GFX10-NEXT: ; implicit-def: $vcc_hi 839; GFX10-NEXT: v_and_b32_e32 v4, v7, v4 840; GFX10-NEXT: v_and_b32_e32 v2, v7, v2 841; GFX10-NEXT: v_and_b32_e32 v0, v7, v0 842; GFX10-NEXT: v_lshl_or_b32 v4, v5, 16, v4 843; GFX10-NEXT: v_lshl_or_b32 v3, v3, 16, v2 844; GFX10-NEXT: v_lshl_or_b32 v2, v1, 16, v0 845; GFX10-NEXT: image_sample_cd v[0:3], v[2:4], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 846; GFX10-NEXT: s_waitcnt vmcnt(0) 847; GFX10-NEXT: ; return to shader part epilog 848main_body: 849 %v = call <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 850 ret <4 x float> %v 851} 852 853define amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) { 854; GFX9-LABEL: sample_c_cd_1d: 855; GFX9: ; %bb.0: ; %main_body 856; GFX9-NEXT: image_sample_c_cd v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf a16 857; GFX9-NEXT: s_waitcnt vmcnt(0) 858; GFX9-NEXT: ; return to shader part epilog 859; 860; GFX10-LABEL: sample_c_cd_1d: 861; GFX10: ; %bb.0: ; %main_body 862; GFX10-NEXT: image_sample_c_cd v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 863; GFX10-NEXT: ; implicit-def: $vcc_hi 864; GFX10-NEXT: s_waitcnt vmcnt(0) 865; GFX10-NEXT: ; return to shader part epilog 866main_body: 867 %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 868 ret <4 x float> %v 869} 870 871define amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) { 872; GFX9-LABEL: sample_c_cd_2d: 873; GFX9: ; %bb.0: ; %main_body 874; GFX9-NEXT: v_mov_b32_e32 v9, 0xffff 875; GFX9-NEXT: v_mov_b32_e32 v8, v2 876; GFX9-NEXT: v_mov_b32_e32 v7, v3 877; GFX9-NEXT: v_and_b32_e32 v2, v9, v5 878; GFX9-NEXT: v_and_b32_e32 v1, v9, v1 879; GFX9-NEXT: v_lshl_or_b32 v3, v6, 16, v2 880; GFX9-NEXT: v_and_b32_e32 v2, v9, v7 881; GFX9-NEXT: v_lshl_or_b32 v2, v4, 16, v2 882; GFX9-NEXT: v_lshl_or_b32 v1, v8, 16, v1 883; GFX9-NEXT: image_sample_c_cd v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf a16 884; GFX9-NEXT: s_waitcnt vmcnt(0) 885; GFX9-NEXT: ; return to shader part epilog 886; 887; GFX10-LABEL: sample_c_cd_2d: 888; GFX10: ; %bb.0: ; %main_body 889; GFX10-NEXT: v_mov_b32_e32 v10, 0xffff 890; GFX10-NEXT: ; implicit-def: $vcc_hi 891; GFX10-NEXT: v_and_b32_e32 v5, v10, v5 892; GFX10-NEXT: v_and_b32_e32 v3, v10, v3 893; GFX10-NEXT: v_and_b32_e32 v1, v10, v1 894; GFX10-NEXT: v_lshl_or_b32 v5, v6, 16, v5 895; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 896; GFX10-NEXT: v_lshl_or_b32 v2, v2, 16, v1 897; GFX10-NEXT: image_sample_c_cd v[0:3], [v0, v2, v3, v5], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 898; GFX10-NEXT: s_waitcnt vmcnt(0) 899; GFX10-NEXT: ; return to shader part epilog 900main_body: 901 %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 902 ret <4 x float> %v 903} 904 905define amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) { 906; GFX9-LABEL: sample_cd_cl_1d: 907; GFX9: ; %bb.0: ; %main_body 908; GFX9-NEXT: v_and_b32_e32 v2, 0xffff, v2 909; GFX9-NEXT: v_lshl_or_b32 v2, v3, 16, v2 910; GFX9-NEXT: image_sample_cd_cl v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf a16 911; GFX9-NEXT: s_waitcnt vmcnt(0) 912; GFX9-NEXT: ; return to shader part epilog 913; 914; GFX10-LABEL: sample_cd_cl_1d: 915; GFX10: ; %bb.0: ; %main_body 916; GFX10-NEXT: v_and_b32_e32 v2, 0xffff, v2 917; GFX10-NEXT: ; implicit-def: $vcc_hi 918; GFX10-NEXT: v_lshl_or_b32 v2, v3, 16, v2 919; GFX10-NEXT: image_sample_cd_cl v[0:3], v[0:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 920; GFX10-NEXT: s_waitcnt vmcnt(0) 921; GFX10-NEXT: ; return to shader part epilog 922main_body: 923 %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 924 ret <4 x float> %v 925} 926 927define amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) { 928; GFX9-LABEL: sample_cd_cl_2d: 929; GFX9: ; %bb.0: ; %main_body 930; GFX9-NEXT: v_mov_b32_e32 v7, 0xffff 931; GFX9-NEXT: v_and_b32_e32 v4, v7, v4 932; GFX9-NEXT: v_and_b32_e32 v2, v7, v2 933; GFX9-NEXT: v_and_b32_e32 v0, v7, v0 934; GFX9-NEXT: v_lshl_or_b32 v5, v5, 16, v4 935; GFX9-NEXT: v_lshl_or_b32 v4, v3, 16, v2 936; GFX9-NEXT: v_lshl_or_b32 v3, v1, 16, v0 937; GFX9-NEXT: image_sample_cd_cl v[0:3], v[3:6], s[0:7], s[8:11] dmask:0xf a16 938; GFX9-NEXT: s_waitcnt vmcnt(0) 939; GFX9-NEXT: ; return to shader part epilog 940; 941; GFX10-LABEL: sample_cd_cl_2d: 942; GFX10: ; %bb.0: ; %main_body 943; GFX10-NEXT: v_mov_b32_e32 v7, 0xffff 944; GFX10-NEXT: ; implicit-def: $vcc_hi 945; GFX10-NEXT: v_and_b32_e32 v4, v7, v4 946; GFX10-NEXT: v_and_b32_e32 v2, v7, v2 947; GFX10-NEXT: v_and_b32_e32 v0, v7, v0 948; GFX10-NEXT: v_lshl_or_b32 v4, v5, 16, v4 949; GFX10-NEXT: v_lshl_or_b32 v5, v3, 16, v2 950; GFX10-NEXT: v_lshl_or_b32 v3, v1, 16, v0 951; GFX10-NEXT: image_sample_cd_cl v[0:3], [v3, v5, v4, v6], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 952; GFX10-NEXT: s_waitcnt vmcnt(0) 953; GFX10-NEXT: ; return to shader part epilog 954main_body: 955 %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 956 ret <4 x float> %v 957} 958 959define amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) { 960; GFX9-LABEL: sample_c_cd_cl_1d: 961; GFX9: ; %bb.0: ; %main_body 962; GFX9-NEXT: v_and_b32_e32 v3, 0xffff, v3 963; GFX9-NEXT: v_lshl_or_b32 v3, v4, 16, v3 964; GFX9-NEXT: image_sample_c_cd_cl v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf a16 965; GFX9-NEXT: s_waitcnt vmcnt(0) 966; GFX9-NEXT: ; return to shader part epilog 967; 968; GFX10-LABEL: sample_c_cd_cl_1d: 969; GFX10: ; %bb.0: ; %main_body 970; GFX10-NEXT: v_and_b32_e32 v3, 0xffff, v3 971; GFX10-NEXT: ; implicit-def: $vcc_hi 972; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 973; GFX10-NEXT: image_sample_c_cd_cl v[0:3], v[0:3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 974; GFX10-NEXT: s_waitcnt vmcnt(0) 975; GFX10-NEXT: ; return to shader part epilog 976main_body: 977 %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 978 ret <4 x float> %v 979} 980 981define amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) { 982; GFX9-LABEL: sample_c_cd_cl_2d: 983; GFX9: ; %bb.0: ; %main_body 984; GFX9-NEXT: v_mov_b32_e32 v11, v7 985; GFX9-NEXT: v_mov_b32_e32 v7, v0 986; GFX9-NEXT: v_mov_b32_e32 v0, 0xffff 987; GFX9-NEXT: v_and_b32_e32 v5, v0, v5 988; GFX9-NEXT: v_and_b32_e32 v3, v0, v3 989; GFX9-NEXT: v_and_b32_e32 v0, v0, v1 990; GFX9-NEXT: v_lshl_or_b32 v10, v6, 16, v5 991; GFX9-NEXT: v_lshl_or_b32 v9, v4, 16, v3 992; GFX9-NEXT: v_lshl_or_b32 v8, v2, 16, v0 993; GFX9-NEXT: image_sample_c_cd_cl v[0:3], v[7:14], s[0:7], s[8:11] dmask:0xf a16 994; GFX9-NEXT: s_waitcnt vmcnt(0) 995; GFX9-NEXT: ; return to shader part epilog 996; 997; GFX10-LABEL: sample_c_cd_cl_2d: 998; GFX10: ; %bb.0: ; %main_body 999; GFX10-NEXT: v_mov_b32_e32 v8, 0xffff 1000; GFX10-NEXT: ; implicit-def: $vcc_hi 1001; GFX10-NEXT: v_and_b32_e32 v5, v8, v5 1002; GFX10-NEXT: v_and_b32_e32 v3, v8, v3 1003; GFX10-NEXT: v_and_b32_e32 v1, v8, v1 1004; GFX10-NEXT: v_lshl_or_b32 v5, v6, 16, v5 1005; GFX10-NEXT: v_lshl_or_b32 v3, v4, 16, v3 1006; GFX10-NEXT: v_lshl_or_b32 v2, v2, 16, v1 1007; GFX10-NEXT: image_sample_c_cd_cl v[0:3], [v0, v2, v3, v5, v7], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 1008; GFX10-NEXT: s_waitcnt vmcnt(0) 1009; GFX10-NEXT: ; return to shader part epilog 1010main_body: 1011 %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1012 ret <4 x float> %v 1013} 1014 1015define amdgpu_ps <4 x float> @sample_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %lod) { 1016; GFX9-LABEL: sample_l_1d: 1017; GFX9: ; %bb.0: ; %main_body 1018; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 1019; GFX9-NEXT: v_lshl_or_b32 v0, v1, 16, v0 1020; GFX9-NEXT: image_sample_l v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 1021; GFX9-NEXT: s_waitcnt vmcnt(0) 1022; GFX9-NEXT: ; return to shader part epilog 1023; 1024; GFX10-LABEL: sample_l_1d: 1025; GFX10: ; %bb.0: ; %main_body 1026; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 1027; GFX10-NEXT: ; implicit-def: $vcc_hi 1028; GFX10-NEXT: v_lshl_or_b32 v0, v1, 16, v0 1029; GFX10-NEXT: image_sample_l v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 1030; GFX10-NEXT: s_waitcnt vmcnt(0) 1031; GFX10-NEXT: ; return to shader part epilog 1032main_body: 1033 %v = call <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f16(i32 15, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1034 ret <4 x float> %v 1035} 1036 1037define amdgpu_ps <4 x float> @sample_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %lod) { 1038; GFX9-LABEL: sample_l_2d: 1039; GFX9: ; %bb.0: ; %main_body 1040; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 1041; GFX9-NEXT: v_lshl_or_b32 v1, v1, 16, v0 1042; GFX9-NEXT: image_sample_l v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf a16 1043; GFX9-NEXT: s_waitcnt vmcnt(0) 1044; GFX9-NEXT: ; return to shader part epilog 1045; 1046; GFX10-LABEL: sample_l_2d: 1047; GFX10: ; %bb.0: ; %main_body 1048; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 1049; GFX10-NEXT: ; implicit-def: $vcc_hi 1050; GFX10-NEXT: v_lshl_or_b32 v1, v1, 16, v0 1051; GFX10-NEXT: image_sample_l v[0:3], v[1:2], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 1052; GFX10-NEXT: s_waitcnt vmcnt(0) 1053; GFX10-NEXT: ; return to shader part epilog 1054main_body: 1055 %v = call <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f16(i32 15, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1056 ret <4 x float> %v 1057} 1058 1059define amdgpu_ps <4 x float> @sample_c_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %lod) { 1060; GFX9-LABEL: sample_c_l_1d: 1061; GFX9: ; %bb.0: ; %main_body 1062; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v1 1063; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1 1064; GFX9-NEXT: image_sample_c_l v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 1065; GFX9-NEXT: s_waitcnt vmcnt(0) 1066; GFX9-NEXT: ; return to shader part epilog 1067; 1068; GFX10-LABEL: sample_c_l_1d: 1069; GFX10: ; %bb.0: ; %main_body 1070; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 1071; GFX10-NEXT: ; implicit-def: $vcc_hi 1072; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 1073; GFX10-NEXT: image_sample_c_l v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 1074; GFX10-NEXT: s_waitcnt vmcnt(0) 1075; GFX10-NEXT: ; return to shader part epilog 1076main_body: 1077 %v = call <4 x float> @llvm.amdgcn.image.sample.c.l.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1078 ret <4 x float> %v 1079} 1080 1081define amdgpu_ps <4 x float> @sample_c_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %lod) { 1082; GFX9-LABEL: sample_c_l_2d: 1083; GFX9: ; %bb.0: ; %main_body 1084; GFX9-NEXT: v_mov_b32_e32 v5, v3 1085; GFX9-NEXT: v_mov_b32_e32 v3, v0 1086; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v1 1087; GFX9-NEXT: v_lshl_or_b32 v4, v2, 16, v0 1088; GFX9-NEXT: image_sample_c_l v[0:3], v[3:5], s[0:7], s[8:11] dmask:0xf a16 1089; GFX9-NEXT: s_waitcnt vmcnt(0) 1090; GFX9-NEXT: ; return to shader part epilog 1091; 1092; GFX10-LABEL: sample_c_l_2d: 1093; GFX10: ; %bb.0: ; %main_body 1094; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 1095; GFX10-NEXT: ; implicit-def: $vcc_hi 1096; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 1097; GFX10-NEXT: image_sample_c_l v[0:3], [v0, v1, v3], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 1098; GFX10-NEXT: s_waitcnt vmcnt(0) 1099; GFX10-NEXT: ; return to shader part epilog 1100main_body: 1101 %v = call <4 x float> @llvm.amdgcn.image.sample.c.l.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1102 ret <4 x float> %v 1103} 1104 1105define amdgpu_ps <4 x float> @sample_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) { 1106; GFX9-LABEL: sample_lz_1d: 1107; GFX9: ; %bb.0: ; %main_body 1108; GFX9-NEXT: image_sample_lz v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 1109; GFX9-NEXT: s_waitcnt vmcnt(0) 1110; GFX9-NEXT: ; return to shader part epilog 1111; 1112; GFX10-LABEL: sample_lz_1d: 1113; GFX10: ; %bb.0: ; %main_body 1114; GFX10-NEXT: image_sample_lz v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 1115; GFX10-NEXT: ; implicit-def: $vcc_hi 1116; GFX10-NEXT: s_waitcnt vmcnt(0) 1117; GFX10-NEXT: ; return to shader part epilog 1118main_body: 1119 %v = call <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1120 ret <4 x float> %v 1121} 1122 1123define amdgpu_ps <4 x float> @sample_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) { 1124; GFX9-LABEL: sample_lz_2d: 1125; GFX9: ; %bb.0: ; %main_body 1126; GFX9-NEXT: v_and_b32_e32 v0, 0xffff, v0 1127; GFX9-NEXT: v_lshl_or_b32 v0, v1, 16, v0 1128; GFX9-NEXT: image_sample_lz v[0:3], v0, s[0:7], s[8:11] dmask:0xf a16 1129; GFX9-NEXT: s_waitcnt vmcnt(0) 1130; GFX9-NEXT: ; return to shader part epilog 1131; 1132; GFX10-LABEL: sample_lz_2d: 1133; GFX10: ; %bb.0: ; %main_body 1134; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0 1135; GFX10-NEXT: ; implicit-def: $vcc_hi 1136; GFX10-NEXT: v_lshl_or_b32 v0, v1, 16, v0 1137; GFX10-NEXT: image_sample_lz v[0:3], v0, s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 1138; GFX10-NEXT: s_waitcnt vmcnt(0) 1139; GFX10-NEXT: ; return to shader part epilog 1140main_body: 1141 %v = call <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1142 ret <4 x float> %v 1143} 1144 1145define amdgpu_ps <4 x float> @sample_c_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) { 1146; GFX9-LABEL: sample_c_lz_1d: 1147; GFX9: ; %bb.0: ; %main_body 1148; GFX9-NEXT: image_sample_c_lz v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 1149; GFX9-NEXT: s_waitcnt vmcnt(0) 1150; GFX9-NEXT: ; return to shader part epilog 1151; 1152; GFX10-LABEL: sample_c_lz_1d: 1153; GFX10: ; %bb.0: ; %main_body 1154; GFX10-NEXT: image_sample_c_lz v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_1D a16 1155; GFX10-NEXT: ; implicit-def: $vcc_hi 1156; GFX10-NEXT: s_waitcnt vmcnt(0) 1157; GFX10-NEXT: ; return to shader part epilog 1158main_body: 1159 %v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1160 ret <4 x float> %v 1161} 1162 1163define amdgpu_ps <4 x float> @sample_c_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) { 1164; GFX9-LABEL: sample_c_lz_2d: 1165; GFX9: ; %bb.0: ; %main_body 1166; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v1 1167; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1 1168; GFX9-NEXT: image_sample_c_lz v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf a16 1169; GFX9-NEXT: s_waitcnt vmcnt(0) 1170; GFX9-NEXT: ; return to shader part epilog 1171; 1172; GFX10-LABEL: sample_c_lz_2d: 1173; GFX10: ; %bb.0: ; %main_body 1174; GFX10-NEXT: v_and_b32_e32 v1, 0xffff, v1 1175; GFX10-NEXT: ; implicit-def: $vcc_hi 1176; GFX10-NEXT: v_lshl_or_b32 v1, v2, 16, v1 1177; GFX10-NEXT: image_sample_c_lz v[0:3], v[0:1], s[0:7], s[8:11] dmask:0xf dim:SQ_RSRC_IMG_2D a16 1178; GFX10-NEXT: s_waitcnt vmcnt(0) 1179; GFX10-NEXT: ; return to shader part epilog 1180main_body: 1181 %v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1182 ret <4 x float> %v 1183} 1184 1185define amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) { 1186; GFX9-LABEL: sample_c_d_o_2darray_V1: 1187; GFX9: ; %bb.0: ; %main_body 1188; GFX9-NEXT: v_mov_b32_e32 v13, v8 1189; GFX9-NEXT: v_mov_b32_e32 v8, v0 1190; GFX9-NEXT: v_mov_b32_e32 v0, 0xffff 1191; GFX9-NEXT: v_mov_b32_e32 v9, v1 1192; GFX9-NEXT: v_and_b32_e32 v1, v0, v6 1193; GFX9-NEXT: v_lshl_or_b32 v12, v7, 16, v1 1194; GFX9-NEXT: v_and_b32_e32 v1, v0, v4 1195; GFX9-NEXT: v_and_b32_e32 v0, v0, v2 1196; GFX9-NEXT: v_lshl_or_b32 v11, v5, 16, v1 1197; GFX9-NEXT: v_lshl_or_b32 v10, v3, 16, v0 1198; GFX9-NEXT: image_sample_c_d_o v0, v[8:15], s[0:7], s[8:11] dmask:0x4 a16 da 1199; GFX9-NEXT: s_waitcnt vmcnt(0) 1200; GFX9-NEXT: ; return to shader part epilog 1201; 1202; GFX10-LABEL: sample_c_d_o_2darray_V1: 1203; GFX10: ; %bb.0: ; %main_body 1204; GFX10-NEXT: v_mov_b32_e32 v9, 0xffff 1205; GFX10-NEXT: ; implicit-def: $vcc_hi 1206; GFX10-NEXT: v_and_b32_e32 v6, v9, v6 1207; GFX10-NEXT: v_and_b32_e32 v4, v9, v4 1208; GFX10-NEXT: v_and_b32_e32 v2, v9, v2 1209; GFX10-NEXT: v_lshl_or_b32 v6, v7, 16, v6 1210; GFX10-NEXT: v_lshl_or_b32 v4, v5, 16, v4 1211; GFX10-NEXT: v_lshl_or_b32 v3, v3, 16, v2 1212; GFX10-NEXT: image_sample_c_d_o v0, [v0, v1, v3, v4, v6, v8], s[0:7], s[8:11] dmask:0x4 dim:SQ_RSRC_IMG_2D_ARRAY a16 1213; GFX10-NEXT: s_waitcnt vmcnt(0) 1214; GFX10-NEXT: ; return to shader part epilog 1215main_body: 1216 %v = call float @llvm.amdgcn.image.sample.c.d.o.2darray.f32.f16.f16(i32 4, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1217 ret float %v 1218} 1219 1220define amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) { 1221; GFX9-LABEL: sample_c_d_o_2darray_V2: 1222; GFX9: ; %bb.0: ; %main_body 1223; GFX9-NEXT: v_mov_b32_e32 v13, v8 1224; GFX9-NEXT: v_mov_b32_e32 v8, v0 1225; GFX9-NEXT: v_mov_b32_e32 v0, 0xffff 1226; GFX9-NEXT: v_mov_b32_e32 v9, v1 1227; GFX9-NEXT: v_and_b32_e32 v1, v0, v6 1228; GFX9-NEXT: v_lshl_or_b32 v12, v7, 16, v1 1229; GFX9-NEXT: v_and_b32_e32 v1, v0, v4 1230; GFX9-NEXT: v_and_b32_e32 v0, v0, v2 1231; GFX9-NEXT: v_lshl_or_b32 v11, v5, 16, v1 1232; GFX9-NEXT: v_lshl_or_b32 v10, v3, 16, v0 1233; GFX9-NEXT: image_sample_c_d_o v[0:1], v[8:15], s[0:7], s[8:11] dmask:0x6 a16 da 1234; GFX9-NEXT: s_waitcnt vmcnt(0) 1235; GFX9-NEXT: ; return to shader part epilog 1236; 1237; GFX10-LABEL: sample_c_d_o_2darray_V2: 1238; GFX10: ; %bb.0: ; %main_body 1239; GFX10-NEXT: v_mov_b32_e32 v9, 0xffff 1240; GFX10-NEXT: ; implicit-def: $vcc_hi 1241; GFX10-NEXT: v_and_b32_e32 v6, v9, v6 1242; GFX10-NEXT: v_and_b32_e32 v4, v9, v4 1243; GFX10-NEXT: v_and_b32_e32 v2, v9, v2 1244; GFX10-NEXT: v_lshl_or_b32 v6, v7, 16, v6 1245; GFX10-NEXT: v_lshl_or_b32 v4, v5, 16, v4 1246; GFX10-NEXT: v_lshl_or_b32 v3, v3, 16, v2 1247; GFX10-NEXT: image_sample_c_d_o v[0:1], [v0, v1, v3, v4, v6, v8], s[0:7], s[8:11] dmask:0x6 dim:SQ_RSRC_IMG_2D_ARRAY a16 1248; GFX10-NEXT: s_waitcnt vmcnt(0) 1249; GFX10-NEXT: ; return to shader part epilog 1250main_body: 1251 %v = call <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f32.f16(i32 6, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) 1252 ret <2 x float> %v 1253} 1254 1255declare <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f16(i32, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1256declare <8 x float> @llvm.amdgcn.image.sample.1d.v8f32.f16(i32, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1257declare <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1258declare <4 x float> @llvm.amdgcn.image.sample.3d.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1259declare <4 x float> @llvm.amdgcn.image.sample.cube.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1260declare <4 x float> @llvm.amdgcn.image.sample.1darray.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1261declare <4 x float> @llvm.amdgcn.image.sample.2darray.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1262 1263declare <4 x float> @llvm.amdgcn.image.sample.c.1d.v4f32.f16(i32, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1264declare <4 x float> @llvm.amdgcn.image.sample.c.2d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1265declare <4 x float> @llvm.amdgcn.image.sample.cl.1d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1266declare <4 x float> @llvm.amdgcn.image.sample.cl.2d.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1267declare <4 x float> @llvm.amdgcn.image.sample.c.cl.1d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1268declare <4 x float> @llvm.amdgcn.image.sample.c.cl.2d.v4f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1269 1270declare <4 x float> @llvm.amdgcn.image.sample.b.1d.v4f32.f32.f16(i32, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1271declare <4 x float> @llvm.amdgcn.image.sample.b.2d.v4f32.f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1272declare <4 x float> @llvm.amdgcn.image.sample.c.b.1d.v4f32.f32.f16(i32, float, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1273declare <4 x float> @llvm.amdgcn.image.sample.c.b.2d.v4f32.f32.f16(i32, float, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1274declare <4 x float> @llvm.amdgcn.image.sample.b.cl.1d.v4f32.f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1275declare <4 x float> @llvm.amdgcn.image.sample.b.cl.2d.v4f32.f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1276declare <4 x float> @llvm.amdgcn.image.sample.c.b.cl.1d.v4f32.f32.f16(i32, float, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1277declare <4 x float> @llvm.amdgcn.image.sample.c.b.cl.2d.v4f32.f32.f16(i32, float, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1278 1279declare <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1280declare <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1281declare <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1282declare <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1283declare <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1284declare <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f16(i32, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1285declare <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1286declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f32.f16(i32, float, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1287declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1288 1289declare <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1290declare <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1291declare <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1292declare <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1293declare <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f16(i32, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1294declare <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1295declare <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f32.f16(i32, float, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1296declare <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1297 1298declare <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1299declare <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1300declare <4 x float> @llvm.amdgcn.image.sample.c.l.1d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1301declare <4 x float> @llvm.amdgcn.image.sample.c.l.2d.v4f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1302 1303declare <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f16(i32, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1304declare <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1305declare <4 x float> @llvm.amdgcn.image.sample.c.lz.1d.v4f32.f16(i32, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1306declare <4 x float> @llvm.amdgcn.image.sample.c.lz.2d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1307 1308declare float @llvm.amdgcn.image.sample.c.d.o.2darray.f32.f16.f16(i32, i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1309declare <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f32.f16(i32, i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1 1310 1311attributes #0 = { nounwind } 1312attributes #1 = { nounwind readonly } 1313attributes #2 = { nounwind readnone } 1314