1 ; RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GCN,GFX10 %s
2 ; RUN: llc -march=amdgcn -mcpu=gfx1100 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GCN,GFX11 %s
4 ; GCN-LABEL: {{^}}smrd_imm_dlc:
5 ; GFX10: s_buffer_load_dword s0, s[0:3], 0x0 dlc ; encoding: [0x00,0x40,0x20,0xf4,0x00,0x00,0x00,0xfa]
6 ; GFX11: s_buffer_load_b32 s0, s[0:3], 0x0 dlc ; encoding: [0x00,0x20,0x20,0xf4,0x00,0x00,0x00,0xf8]
7 define amdgpu_ps float @smrd_imm_dlc(<4 x i32> inreg %desc) #0 {
9 %r = call float @llvm.amdgcn.s.buffer.load.f32(<4 x i32> %desc, i32 0, i32 4)
13 ; GCN-LABEL: {{^}}smrd_sgpr_dlc:
14 ; GFX10: s_buffer_load_dword s0, s[0:3], s4 offset:0x0 dlc ; encoding: [0x00,0x40,0x20,0xf4,0x00,0x00,0x00,0x08]
15 ; GFX11: s_buffer_load_b32 s0, s[0:3], s4 offset:0x0 dlc ; encoding: [0x00,0x20,0x20,0xf4,0x00,0x00,0x00,0x08]
16 define amdgpu_ps float @smrd_sgpr_dlc(<4 x i32> inreg %desc, i32 inreg %offset) #0 {
18 %r = call float @llvm.amdgcn.s.buffer.load.f32(<4 x i32> %desc, i32 %offset, i32 4)
22 ; GCN-LABEL: {{^}}smrd_imm_glc_dlc:
23 ; GFX10: s_buffer_load_dword s0, s[0:3], 0x0 glc dlc ; encoding: [0x00,0x40,0x21,0xf4,0x00,0x00,0x00,0xfa]
24 ; GFX11: s_buffer_load_b32 s0, s[0:3], 0x0 glc dlc ; encoding: [0x00,0x60,0x20,0xf4,0x00,0x00,0x00,0xf8]
25 define amdgpu_ps float @smrd_imm_glc_dlc(<4 x i32> inreg %desc) #0 {
27 %r = call float @llvm.amdgcn.s.buffer.load.f32(<4 x i32> %desc, i32 0, i32 5)
31 ; GCN-LABEL: {{^}}smrd_sgpr_glc_dlc:
32 ; GFX10: s_buffer_load_dword s0, s[0:3], s4 offset:0x0 glc dlc ; encoding: [0x00,0x40,0x21,0xf4,0x00,0x00,0x00,0x08]
33 ; GFX11: s_buffer_load_b32 s0, s[0:3], s4 offset:0x0 glc dlc ; encoding: [0x00,0x60,0x20,0xf4,0x00,0x00,0x00,0x08]
34 define amdgpu_ps float @smrd_sgpr_glc_dlc(<4 x i32> inreg %desc, i32 inreg %offset) #0 {
36 %r = call float @llvm.amdgcn.s.buffer.load.f32(<4 x i32> %desc, i32 %offset, i32 5)
40 declare float @llvm.amdgcn.s.buffer.load.f32(<4 x i32>, i32, i32)