; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX9 %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX8 %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=hawaii -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX7 %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10 %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11 %s define amdgpu_ps i8 @extractelement_sgpr_v4i8_sgpr_idx(ptr addrspace(4) inreg %ptr, i32 inreg %idx) { ; GFX9-LABEL: extractelement_sgpr_v4i8_sgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_and_b32 s0, s4, 3 ; GFX9-NEXT: s_ashr_i32 s1, s0, 31 ; GFX9-NEXT: s_add_u32 s0, s2, s0 ; GFX9-NEXT: s_addc_u32 s1, s3, s1 ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[0:1] ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v4i8_sgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_and_b32 s0, s4, 3 ; GFX8-NEXT: s_ashr_i32 s1, s0, 31 ; GFX8-NEXT: s_add_u32 s0, s2, s0 ; GFX8-NEXT: s_addc_u32 s1, s3, s1 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v4i8_sgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_and_b32 s4, s4, 3 ; GFX7-NEXT: s_ashr_i32 s5, s4, 31 ; GFX7-NEXT: v_mov_b32_e32 v0, s4 ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: v_mov_b32_e32 v1, s5 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v4i8_sgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_and_b32 s0, s4, 3 ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: s_ashr_i32 s1, s0, 31 ; GFX10-NEXT: s_add_u32 s0, s2, s0 ; GFX10-NEXT: s_addc_u32 s1, s3, s1 ; GFX10-NEXT: global_load_ubyte v0, v0, s[0:1] ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v4i8_sgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_and_b32 s0, s4, 3 ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: s_ashr_i32 s1, s0, 31 ; GFX11-NEXT: s_add_u32 s0, s2, s0 ; GFX11-NEXT: s_addc_u32 s1, s3, s1 ; GFX11-NEXT: global_load_u8 v0, v0, s[0:1] ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(4) %ptr %element = extractelement <4 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_vgpr_v4i8_sgpr_idx(ptr addrspace(1) %ptr, i32 inreg %idx) { ; GFX9-LABEL: extractelement_vgpr_v4i8_sgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_and_b32 s0, s2, 3 ; GFX9-NEXT: s_ashr_i32 s1, s0, 31 ; GFX9-NEXT: v_mov_b32_e32 v3, s1 ; GFX9-NEXT: v_mov_b32_e32 v2, s0 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_vgpr_v4i8_sgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_and_b32 s0, s2, 3 ; GFX8-NEXT: s_ashr_i32 s1, s0, 31 ; GFX8-NEXT: v_mov_b32_e32 v3, s1 ; GFX8-NEXT: v_mov_b32_e32 v2, s0 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_vgpr_v4i8_sgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_and_b32 s0, s2, 3 ; GFX7-NEXT: s_ashr_i32 s1, s0, 31 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_vgpr_v4i8_sgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_and_b32 s0, s2, 3 ; GFX10-NEXT: s_ashr_i32 s1, s0, 31 ; GFX10-NEXT: v_mov_b32_e32 v3, s1 ; GFX10-NEXT: v_mov_b32_e32 v2, s0 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_vgpr_v4i8_sgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_and_b32 s0, s2, 3 ; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) ; GFX11-NEXT: s_ashr_i32 s1, s0, 31 ; GFX11-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(1) %ptr %element = extractelement <4 x i8> %vector, i32 %idx ret i8 %element } define i8 @extractelement_vgpr_v4i8_vgpr_idx(ptr addrspace(1) %ptr, i32 %idx) { ; GFX9-LABEL: extractelement_vgpr_v4i8_vgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: v_and_b32_e32 v2, 3, v2 ; GFX9-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v4i8_vgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_and_b32_e32 v2, 3, v2 ; GFX8-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v4i8_vgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: v_and_b32_e32 v2, 3, v2 ; GFX7-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX7-NEXT: v_add_i32_e32 v0, vcc, v0, v2 ; GFX7-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v4i8_vgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_and_b32_e32 v2, 3, v2 ; GFX10-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v4i8_vgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: v_and_b32_e32 v2, 3, v2 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <4 x i8>, ptr addrspace(1) %ptr %element = extractelement <4 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v4i8_vgpr_idx(ptr addrspace(4) inreg %ptr, i32 %idx) { ; GFX9-LABEL: extractelement_sgpr_v4i8_vgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_and_b32_e32 v2, 3, v0 ; GFX9-NEXT: v_mov_b32_e32 v0, s2 ; GFX9-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX9-NEXT: v_mov_b32_e32 v1, s3 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v4i8_vgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: v_and_b32_e32 v2, 3, v0 ; GFX8-NEXT: v_mov_b32_e32 v0, s2 ; GFX8-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX8-NEXT: v_mov_b32_e32 v1, s3 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v4i8_vgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: v_and_b32_e32 v0, 3, v0 ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: v_ashrrev_i32_e32 v1, 31, v0 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v4i8_vgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_and_b32_e32 v2, 3, v0 ; GFX10-NEXT: v_mov_b32_e32 v0, s2 ; GFX10-NEXT: v_mov_b32_e32 v1, s3 ; GFX10-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v4i8_vgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_and_b32_e32 v2, 3, v0 ; GFX11-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(4) %ptr %element = extractelement <4 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v4i8_idx0(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v4i8_idx0: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v4i8_idx0: ; GFX8: ; %bb.0: ; GFX8-NEXT: v_mov_b32_e32 v0, s2 ; GFX8-NEXT: v_mov_b32_e32 v1, s3 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v4i8_idx0: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v4i8_idx0: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v4i8_idx0: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(4) %ptr %element = extractelement <4 x i8> %vector, i32 0 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v4i8_idx1(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v4i8_idx1: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:1 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v4i8_idx1: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 1 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v4i8_idx1: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:1 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v4i8_idx1: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:1 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v4i8_idx1: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:1 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(4) %ptr %element = extractelement <4 x i8> %vector, i32 1 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v4i8_idx2(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v4i8_idx2: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:2 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v4i8_idx2: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 2 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v4i8_idx2: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:2 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v4i8_idx2: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:2 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v4i8_idx2: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:2 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(4) %ptr %element = extractelement <4 x i8> %vector, i32 2 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v4i8_idx3(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v4i8_idx3: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:3 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v4i8_idx3: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 3 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v4i8_idx3: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:3 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v4i8_idx3: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:3 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v4i8_idx3: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:3 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <4 x i8>, ptr addrspace(4) %ptr %element = extractelement <4 x i8> %vector, i32 3 ret i8 %element } define i8 @extractelement_vgpr_v4i8_idx0(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v4i8_idx0: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v4i8_idx0: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v4i8_idx0: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v4i8_idx0: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v4i8_idx0: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <4 x i8>, ptr addrspace(1) %ptr %element = extractelement <4 x i8> %vector, i32 0 ret i8 %element } define i8 @extractelement_vgpr_v4i8_idx1(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v4i8_idx1: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:1 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v4i8_idx1: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 1, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v4i8_idx1: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:1 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v4i8_idx1: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:1 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v4i8_idx1: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:1 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <4 x i8>, ptr addrspace(1) %ptr %element = extractelement <4 x i8> %vector, i32 1 ret i8 %element } define i8 @extractelement_vgpr_v4i8_idx2(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v4i8_idx2: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:2 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v4i8_idx2: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 2, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v4i8_idx2: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:2 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v4i8_idx2: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:2 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v4i8_idx2: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:2 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <4 x i8>, ptr addrspace(1) %ptr %element = extractelement <4 x i8> %vector, i32 2 ret i8 %element } define i8 @extractelement_vgpr_v4i8_idx3(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v4i8_idx3: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:3 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v4i8_idx3: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 3, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v4i8_idx3: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:3 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v4i8_idx3: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:3 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v4i8_idx3: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:3 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <4 x i8>, ptr addrspace(1) %ptr %element = extractelement <4 x i8> %vector, i32 3 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_sgpr_idx(ptr addrspace(4) inreg %ptr, i32 inreg %idx) { ; GFX9-LABEL: extractelement_sgpr_v8i8_sgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_and_b32 s0, s4, 7 ; GFX9-NEXT: s_ashr_i32 s1, s0, 31 ; GFX9-NEXT: s_add_u32 s0, s2, s0 ; GFX9-NEXT: s_addc_u32 s1, s3, s1 ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[0:1] ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_sgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_and_b32 s0, s4, 7 ; GFX8-NEXT: s_ashr_i32 s1, s0, 31 ; GFX8-NEXT: s_add_u32 s0, s2, s0 ; GFX8-NEXT: s_addc_u32 s1, s3, s1 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_sgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_and_b32 s4, s4, 7 ; GFX7-NEXT: s_ashr_i32 s5, s4, 31 ; GFX7-NEXT: v_mov_b32_e32 v0, s4 ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: v_mov_b32_e32 v1, s5 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_sgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_and_b32 s0, s4, 7 ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: s_ashr_i32 s1, s0, 31 ; GFX10-NEXT: s_add_u32 s0, s2, s0 ; GFX10-NEXT: s_addc_u32 s1, s3, s1 ; GFX10-NEXT: global_load_ubyte v0, v0, s[0:1] ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_sgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_and_b32 s0, s4, 7 ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: s_ashr_i32 s1, s0, 31 ; GFX11-NEXT: s_add_u32 s0, s2, s0 ; GFX11-NEXT: s_addc_u32 s1, s3, s1 ; GFX11-NEXT: global_load_u8 v0, v0, s[0:1] ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_vgpr_v8i8_sgpr_idx(ptr addrspace(1) %ptr, i32 inreg %idx) { ; GFX9-LABEL: extractelement_vgpr_v8i8_sgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_and_b32 s0, s2, 7 ; GFX9-NEXT: s_ashr_i32 s1, s0, 31 ; GFX9-NEXT: v_mov_b32_e32 v3, s1 ; GFX9-NEXT: v_mov_b32_e32 v2, s0 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_vgpr_v8i8_sgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_and_b32 s0, s2, 7 ; GFX8-NEXT: s_ashr_i32 s1, s0, 31 ; GFX8-NEXT: v_mov_b32_e32 v3, s1 ; GFX8-NEXT: v_mov_b32_e32 v2, s0 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_vgpr_v8i8_sgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_and_b32 s0, s2, 7 ; GFX7-NEXT: s_ashr_i32 s1, s0, 31 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_vgpr_v8i8_sgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_and_b32 s0, s2, 7 ; GFX10-NEXT: s_ashr_i32 s1, s0, 31 ; GFX10-NEXT: v_mov_b32_e32 v3, s1 ; GFX10-NEXT: v_mov_b32_e32 v2, s0 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_vgpr_v8i8_sgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_and_b32 s0, s2, 7 ; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) ; GFX11-NEXT: s_ashr_i32 s1, s0, 31 ; GFX11-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 %idx ret i8 %element } define i8 @extractelement_vgpr_v8i8_vgpr_idx(ptr addrspace(1) %ptr, i32 %idx) { ; GFX9-LABEL: extractelement_vgpr_v8i8_vgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: v_and_b32_e32 v2, 7, v2 ; GFX9-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_vgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_and_b32_e32 v2, 7, v2 ; GFX8-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_vgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: v_and_b32_e32 v2, 7, v2 ; GFX7-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX7-NEXT: v_add_i32_e32 v0, vcc, v0, v2 ; GFX7-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_vgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_and_b32_e32 v2, 7, v2 ; GFX10-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_vgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: v_and_b32_e32 v2, 7, v2 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_vgpr_idx(ptr addrspace(4) inreg %ptr, i32 %idx) { ; GFX9-LABEL: extractelement_sgpr_v8i8_vgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_and_b32_e32 v2, 7, v0 ; GFX9-NEXT: v_mov_b32_e32 v0, s2 ; GFX9-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX9-NEXT: v_mov_b32_e32 v1, s3 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_vgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: v_and_b32_e32 v2, 7, v0 ; GFX8-NEXT: v_mov_b32_e32 v0, s2 ; GFX8-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX8-NEXT: v_mov_b32_e32 v1, s3 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_vgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: v_and_b32_e32 v0, 7, v0 ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: v_ashrrev_i32_e32 v1, 31, v0 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_vgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_and_b32_e32 v2, 7, v0 ; GFX10-NEXT: v_mov_b32_e32 v0, s2 ; GFX10-NEXT: v_mov_b32_e32 v1, s3 ; GFX10-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_vgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_and_b32_e32 v2, 7, v0 ; GFX11-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx0(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx0: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx0: ; GFX8: ; %bb.0: ; GFX8-NEXT: v_mov_b32_e32 v0, s2 ; GFX8-NEXT: v_mov_b32_e32 v1, s3 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx0: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx0: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx0: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 0 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx1(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx1: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:1 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx1: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 1 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx1: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:1 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx1: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:1 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx1: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:1 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 1 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx2(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx2: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:2 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx2: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 2 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx2: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:2 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx2: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:2 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx2: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:2 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 2 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx3(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx3: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:3 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx3: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 3 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx3: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:3 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx3: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:3 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx3: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:3 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 3 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx4(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx4: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:4 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx4: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 4 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx4: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:4 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx4: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:4 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx4: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:4 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 4 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx5(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx5: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:5 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx5: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 5 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx5: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:5 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx5: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:5 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx5: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:5 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 5 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx6(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx6: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:6 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx6: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 6 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx6: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:6 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx6: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:6 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx6: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:6 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 6 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v8i8_idx7(ptr addrspace(4) inreg %ptr) { ; GFX9-LABEL: extractelement_sgpr_v8i8_idx7: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[2:3] offset:7 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v8i8_idx7: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_add_u32 s0, s2, 7 ; GFX8-NEXT: s_addc_u32 s1, s3, 0 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v8i8_idx7: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, -1 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, off, s[0:3], 0 offset:7 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v8i8_idx7: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: global_load_ubyte v0, v0, s[2:3] offset:7 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v8i8_idx7: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: global_load_u8 v0, v0, s[2:3] offset:7 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <8 x i8>, ptr addrspace(4) %ptr %element = extractelement <8 x i8> %vector, i32 7 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx0(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx0: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx0: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx0: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx0: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx0: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 0 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx1(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx1: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:1 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx1: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 1, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx1: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:1 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx1: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:1 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx1: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:1 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 1 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx2(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx2: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:2 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx2: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 2, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx2: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:2 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx2: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:2 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx2: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:2 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 2 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx3(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx3: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:3 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx3: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 3, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx3: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:3 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx3: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:3 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx3: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:3 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 3 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx4(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx4: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:4 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx4: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 4, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx4: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:4 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx4: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:4 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx4: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:4 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 4 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx5(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx5: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:5 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx5: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 5, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx5: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:5 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx5: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:5 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx5: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:5 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 5 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx6(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx6: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:6 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx6: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 6, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx6: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:6 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx6: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:6 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx6: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:6 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 6 ret i8 %element } define i8 @extractelement_vgpr_v8i8_idx7(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v8i8_idx7: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:7 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v8i8_idx7: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 7, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v8i8_idx7: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:7 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v8i8_idx7: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:7 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v8i8_idx7: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:7 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <8 x i8>, ptr addrspace(1) %ptr %element = extractelement <8 x i8> %vector, i32 7 ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v16i8_sgpr_idx(ptr addrspace(4) inreg %ptr, i32 inreg %idx) { ; GFX9-LABEL: extractelement_sgpr_v16i8_sgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_and_b32 s0, s4, 15 ; GFX9-NEXT: s_ashr_i32 s1, s0, 31 ; GFX9-NEXT: s_add_u32 s0, s2, s0 ; GFX9-NEXT: s_addc_u32 s1, s3, s1 ; GFX9-NEXT: v_mov_b32_e32 v0, 0 ; GFX9-NEXT: global_load_ubyte v0, v0, s[0:1] ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v16i8_sgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_and_b32 s0, s4, 15 ; GFX8-NEXT: s_ashr_i32 s1, s0, 31 ; GFX8-NEXT: s_add_u32 s0, s2, s0 ; GFX8-NEXT: s_addc_u32 s1, s3, s1 ; GFX8-NEXT: v_mov_b32_e32 v0, s0 ; GFX8-NEXT: v_mov_b32_e32 v1, s1 ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v16i8_sgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_and_b32 s4, s4, 15 ; GFX7-NEXT: s_ashr_i32 s5, s4, 31 ; GFX7-NEXT: v_mov_b32_e32 v0, s4 ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: v_mov_b32_e32 v1, s5 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v16i8_sgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_and_b32 s0, s4, 15 ; GFX10-NEXT: v_mov_b32_e32 v0, 0 ; GFX10-NEXT: s_ashr_i32 s1, s0, 31 ; GFX10-NEXT: s_add_u32 s0, s2, s0 ; GFX10-NEXT: s_addc_u32 s1, s3, s1 ; GFX10-NEXT: global_load_ubyte v0, v0, s[0:1] ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v16i8_sgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_and_b32 s0, s4, 15 ; GFX11-NEXT: v_mov_b32_e32 v0, 0 ; GFX11-NEXT: s_ashr_i32 s1, s0, 31 ; GFX11-NEXT: s_add_u32 s0, s2, s0 ; GFX11-NEXT: s_addc_u32 s1, s3, s1 ; GFX11-NEXT: global_load_u8 v0, v0, s[0:1] ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <16 x i8>, ptr addrspace(4) %ptr %element = extractelement <16 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_vgpr_v16i8_sgpr_idx(ptr addrspace(1) %ptr, i32 inreg %idx) { ; GFX9-LABEL: extractelement_vgpr_v16i8_sgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_and_b32 s0, s2, 15 ; GFX9-NEXT: s_ashr_i32 s1, s0, 31 ; GFX9-NEXT: v_mov_b32_e32 v3, s1 ; GFX9-NEXT: v_mov_b32_e32 v2, s0 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_vgpr_v16i8_sgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_and_b32 s0, s2, 15 ; GFX8-NEXT: s_ashr_i32 s1, s0, 31 ; GFX8-NEXT: v_mov_b32_e32 v3, s1 ; GFX8-NEXT: v_mov_b32_e32 v2, s0 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_vgpr_v16i8_sgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_and_b32 s0, s2, 15 ; GFX7-NEXT: s_ashr_i32 s1, s0, 31 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_vgpr_v16i8_sgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_and_b32 s0, s2, 15 ; GFX10-NEXT: s_ashr_i32 s1, s0, 31 ; GFX10-NEXT: v_mov_b32_e32 v3, s1 ; GFX10-NEXT: v_mov_b32_e32 v2, s0 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_vgpr_v16i8_sgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_and_b32 s0, s2, 15 ; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) ; GFX11-NEXT: s_ashr_i32 s1, s0, 31 ; GFX11-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 %idx ret i8 %element } define i8 @extractelement_vgpr_v16i8_vgpr_idx(ptr addrspace(1) %ptr, i32 %idx) { ; GFX9-LABEL: extractelement_vgpr_v16i8_vgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: v_and_b32_e32 v2, 15, v2 ; GFX9-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_vgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_and_b32_e32 v2, 15, v2 ; GFX8-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_vgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: v_and_b32_e32 v2, 15, v2 ; GFX7-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX7-NEXT: v_add_i32_e32 v0, vcc, v0, v2 ; GFX7-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_vgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_and_b32_e32 v2, 15, v2 ; GFX10-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_vgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: v_and_b32_e32 v2, 15, v2 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 %idx ret i8 %element } define amdgpu_ps i8 @extractelement_sgpr_v16i8_vgpr_idx(ptr addrspace(4) inreg %ptr, i32 %idx) { ; GFX9-LABEL: extractelement_sgpr_v16i8_vgpr_idx: ; GFX9: ; %bb.0: ; GFX9-NEXT: v_and_b32_e32 v2, 15, v0 ; GFX9-NEXT: v_mov_b32_e32 v0, s2 ; GFX9-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX9-NEXT: v_mov_b32_e32 v1, s3 ; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2 ; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: v_readfirstlane_b32 s0, v0 ; GFX9-NEXT: ; return to shader part epilog ; ; GFX8-LABEL: extractelement_sgpr_v16i8_vgpr_idx: ; GFX8: ; %bb.0: ; GFX8-NEXT: v_and_b32_e32 v2, 15, v0 ; GFX8-NEXT: v_mov_b32_e32 v0, s2 ; GFX8-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX8-NEXT: v_mov_b32_e32 v1, s3 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: v_readfirstlane_b32 s0, v0 ; GFX8-NEXT: ; return to shader part epilog ; ; GFX7-LABEL: extractelement_sgpr_v16i8_vgpr_idx: ; GFX7: ; %bb.0: ; GFX7-NEXT: v_and_b32_e32 v0, 15, v0 ; GFX7-NEXT: s_mov_b32 s0, s2 ; GFX7-NEXT: s_mov_b32 s1, s3 ; GFX7-NEXT: v_ashrrev_i32_e32 v1, 31, v0 ; GFX7-NEXT: s_mov_b32 s2, 0 ; GFX7-NEXT: s_mov_b32 s3, 0xf000 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: v_readfirstlane_b32 s0, v0 ; GFX7-NEXT: ; return to shader part epilog ; ; GFX10-LABEL: extractelement_sgpr_v16i8_vgpr_idx: ; GFX10: ; %bb.0: ; GFX10-NEXT: v_and_b32_e32 v2, 15, v0 ; GFX10-NEXT: v_mov_b32_e32 v0, s2 ; GFX10-NEXT: v_mov_b32_e32 v1, s3 ; GFX10-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: v_readfirstlane_b32 s0, v0 ; GFX10-NEXT: ; return to shader part epilog ; ; GFX11-LABEL: extractelement_sgpr_v16i8_vgpr_idx: ; GFX11: ; %bb.0: ; GFX11-NEXT: v_and_b32_e32 v2, 15, v0 ; GFX11-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX11-NEXT: v_ashrrev_i32_e32 v3, 31, v2 ; GFX11-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) ; GFX11-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v3, vcc_lo ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: v_readfirstlane_b32 s0, v0 ; GFX11-NEXT: ; return to shader part epilog %vector = load <16 x i8>, ptr addrspace(4) %ptr %element = extractelement <16 x i8> %vector, i32 %idx ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx0(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx0: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx0: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx0: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx0: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx0: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 0 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx1(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx1: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:1 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx1: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 1, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx1: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:1 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx1: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:1 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx1: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:1 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 1 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx2(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx2: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:2 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx2: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 2, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx2: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:2 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx2: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:2 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx2: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:2 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 2 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx3(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx3: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:3 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx3: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 3, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx3: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:3 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx3: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:3 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx3: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:3 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 3 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx4(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx4: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:4 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx4: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 4, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx4: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:4 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx4: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:4 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx4: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:4 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 4 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx5(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx5: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:5 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx5: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 5, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx5: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:5 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx5: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:5 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx5: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:5 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 5 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx6(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx6: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:6 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx6: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 6, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx6: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:6 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx6: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:6 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx6: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:6 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 6 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx7(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx7: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:7 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx7: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 7, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx7: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:7 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx7: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:7 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx7: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:7 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 7 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx8(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx8: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:8 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx8: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 8, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx8: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:8 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx8: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:8 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx8: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:8 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 8 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx9(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx9: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:9 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx9: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 9, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx9: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:9 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx9: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:9 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx9: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:9 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 9 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx10(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx10: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:10 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx10: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 10, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx10: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:10 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx10: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:10 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx10: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:10 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 10 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx11(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx11: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:11 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx11: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 11, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx11: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:11 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx11: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:11 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx11: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:11 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 11 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx12(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx12: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:12 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx12: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 12, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx12: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:12 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx12: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:12 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx12: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:12 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 12 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx13(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx13: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:13 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx13: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 13, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx13: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:13 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx13: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:13 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx13: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:13 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 13 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx14(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx14: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:14 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx14: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 14, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx14: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:14 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx14: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:14 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx14: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:14 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 14 ret i8 %element } define i8 @extractelement_vgpr_v16i8_idx15(ptr addrspace(1) %ptr) { ; GFX9-LABEL: extractelement_vgpr_v16i8_idx15: ; GFX9: ; %bb.0: ; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX9-NEXT: global_load_ubyte v0, v[0:1], off offset:15 ; GFX9-NEXT: s_waitcnt vmcnt(0) ; GFX9-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: extractelement_vgpr_v16i8_idx15: ; GFX8: ; %bb.0: ; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_add_u32_e32 v0, vcc, 15, v0 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; GFX8-NEXT: flat_load_ubyte v0, v[0:1] ; GFX8-NEXT: s_waitcnt vmcnt(0) ; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX7-LABEL: extractelement_vgpr_v16i8_idx15: ; GFX7: ; %bb.0: ; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX7-NEXT: s_mov_b32 s6, 0 ; GFX7-NEXT: s_mov_b32 s7, 0xf000 ; GFX7-NEXT: s_mov_b64 s[4:5], 0 ; GFX7-NEXT: buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:15 ; GFX7-NEXT: s_waitcnt vmcnt(0) ; GFX7-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: extractelement_vgpr_v16i8_idx15: ; GFX10: ; %bb.0: ; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: global_load_ubyte v0, v[0:1], off offset:15 ; GFX10-NEXT: s_waitcnt vmcnt(0) ; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX11-LABEL: extractelement_vgpr_v16i8_idx15: ; GFX11: ; %bb.0: ; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX11-NEXT: global_load_u8 v0, v[0:1], off offset:15 ; GFX11-NEXT: s_waitcnt vmcnt(0) ; GFX11-NEXT: s_setpc_b64 s[30:31] %vector = load <16 x i8>, ptr addrspace(1) %ptr %element = extractelement <16 x i8> %vector, i32 15 ret i8 %element } ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: ; GCN: {{.*}}