diff --git a/lgc/test/BuiltIns/cs-deviceindex.lgc b/lgc/test/BuiltIns/cs-deviceindex.lgc index f6609790c4..5864885362 100644 --- a/lgc/test/BuiltIns/cs-deviceindex.lgc +++ b/lgc/test/BuiltIns/cs-deviceindex.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=CHECK %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=CHECK %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: @@ -31,8 +31,8 @@ attributes #0 = { nounwind } ; CHECK: s_getpc_b64 s[2:3] ; CHECK-NEXT: s_mov_b32 s0, s1 ; CHECK-NEXT: s_mov_b32 s1, s3 -; CHECK-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 ; CHECK-NEXT: v_mov_b32_e32 v0, 0xbc614e +; CHECK-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 ; CHECK-NEXT: s_waitcnt lgkmcnt(0) ; CHECK-NEXT: buffer_store_dword v0, off, s[0:3], 0 ; CHECK-NEXT: s_endpgm diff --git a/lgc/test/BuiltIns/cs-globalinvocationid.lgc b/lgc/test/BuiltIns/cs-globalinvocationid.lgc index c482da2e85..d7337267da 100644 --- a/lgc/test/BuiltIns/cs-globalinvocationid.lgc +++ b/lgc/test/BuiltIns/cs-globalinvocationid.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=GFX8 %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=GFX8 %s ; RUN: lgc -mcpu=gfx1100 - < %s | FileCheck --check-prefixes=GFX11 %s ; REQUIRES: do-not-run-me diff --git a/lgc/test/BuiltIns/cs-localinvocationid.lgc b/lgc/test/BuiltIns/cs-localinvocationid.lgc index d0f44b9cfd..3f4339459a 100644 --- a/lgc/test/BuiltIns/cs-localinvocationid.lgc +++ b/lgc/test/BuiltIns/cs-localinvocationid.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=GFX8 %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=GFX8 %s ; RUN: lgc -mcpu=gfx1100 - < %s | FileCheck --check-prefixes=GFX11 %s ; REQUIRES: do-not-run-me diff --git a/lgc/test/BuiltIns/cs-localinvocationindex.lgc b/lgc/test/BuiltIns/cs-localinvocationindex.lgc index 83994ec9b7..9c84d7eb17 100644 --- a/lgc/test/BuiltIns/cs-localinvocationindex.lgc +++ b/lgc/test/BuiltIns/cs-localinvocationindex.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=GFX8 %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=GFX8 %s ; RUN: lgc -mcpu=gfx1100 - < %s | FileCheck --check-prefixes=GFX11 %s ; REQUIRES: do-not-run-me diff --git a/lgc/test/BuiltIns/cs-numsubgroups.lgc b/lgc/test/BuiltIns/cs-numsubgroups.lgc index c86ff3709a..0b41f89cc3 100644 --- a/lgc/test/BuiltIns/cs-numsubgroups.lgc +++ b/lgc/test/BuiltIns/cs-numsubgroups.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=WAVE64 %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=WAVE64 %s ; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=WAVE32 %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { @@ -32,8 +32,8 @@ attributes #0 = { nounwind } ; WAVE64: s_getpc_b64 s[2:3] ; WAVE64-NEXT: s_mov_b32 s0, s1 ; WAVE64-NEXT: s_mov_b32 s1, s3 +; WAVE64-NEXT: v_mov_b32_e32 v0, 7 ; WAVE64-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 -; WAVE64-NEXT: v_mov_b32_e32 v0, 4 ; WAVE64-NEXT: s_waitcnt lgkmcnt(0) ; WAVE64-NEXT: buffer_store_dword v0, off, s[0:3], 0 ; WAVE64-NEXT: s_endpgm diff --git a/lgc/test/BuiltIns/cs-numworkgroups.lgc b/lgc/test/BuiltIns/cs-numworkgroups.lgc index a8f0021c37..6327b0bd6a 100644 --- a/lgc/test/BuiltIns/cs-numworkgroups.lgc +++ b/lgc/test/BuiltIns/cs-numworkgroups.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc --check-pal-metadata -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=CHECK %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=CHECK %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: @@ -43,30 +43,73 @@ attributes #0 = { nounwind } ; CHECK-NEXT: --- ; CHECK-NEXT: amdpal.pipelines: ; CHECK-NEXT: - .api: !str '' +; CHECK-NEXT: .compute_registers: +; CHECK-NEXT: .tg_size_en: true +; CHECK-NEXT: .tgid_x_en: false +; CHECK-NEXT: .tgid_y_en: false +; CHECK-NEXT: .tgid_z_en: false +; CHECK-NEXT: .tidig_comp_cnt: 0 ; CHECK-NEXT: .hardware_stages: ; CHECK-NEXT: .cs: +; CHECK-NEXT: .checksum_value: 0 +; CHECK-NEXT: .debug_mode: false ; CHECK-NEXT: .entry_point: _amdgpu_cs_main +; CHECK-NEXT: .excp_en: 0 +; CHECK-NEXT: .float_mode: 0xc0 +; CHECK-NEXT: .ieee_mode: false +; CHECK-NEXT: .lds_size: 0 +; CHECK-NEXT: .mem_ordered: true +; CHECK-NEXT: .scratch_en: false ; CHECK-NEXT: .scratch_memory_size: 0 -; CHECK-NEXT: .sgpr_count: 0x60 -; CHECK-NEXT: .sgpr_limit: 0x66 +; CHECK-NEXT: .sgpr_count: 0x8 +; CHECK-NEXT: .sgpr_limit: 0x6a ; CHECK-NEXT: .threadgroup_dimensions: ; CHECK-NEXT: - 0 ; CHECK-NEXT: - 0 ; CHECK-NEXT: - 0 +; CHECK-NEXT: .trap_present: false +; CHECK-NEXT: .user_data_reg_map: +; CHECK-NEXT: - 0x10000000 +; CHECK-NEXT: - 0x10000006 +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0x2 +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: .user_sgprs: 0x4 ; CHECK-NEXT: .vgpr_count: 0x3 ; CHECK-NEXT: .vgpr_limit: 0x100 +; CHECK-NEXT: .wavefront_size: 0x20 +; CHECK-NEXT: .wgp_mode: false ; CHECK-NEXT: .internal_pipeline_hash: ; CHECK-NEXT: - 0 ; CHECK-NEXT: - 0 -; CHECK-NEXT: .registers: -; CHECK-NEXT: 0x2e07 (COMPUTE_NUM_THREAD_X): 0 -; CHECK-NEXT: 0x2e08 (COMPUTE_NUM_THREAD_Y): 0 -; CHECK-NEXT: 0x2e09 (COMPUTE_NUM_THREAD_Z): 0 -; CHECK-NEXT: 0x2e12 (COMPUTE_PGM_RSRC1): 0x2c02c0 -; CHECK-NEXT: 0x2e13 (COMPUTE_PGM_RSRC2): 0x788 -; CHECK-NEXT: 0x2e40 (COMPUTE_USER_DATA_0): 0x10000000 -; CHECK-NEXT: 0x2e41 (COMPUTE_USER_DATA_1): 0x10000006 -; CHECK-NEXT: 0x2e43 (COMPUTE_USER_DATA_3): 0x2 +; CHECK-NEXT: .registers: {} ; CHECK-NEXT: .shaders: ; CHECK-NEXT: .compute: ; CHECK-NEXT: .api_shader_hash: @@ -78,6 +121,6 @@ attributes #0 = { nounwind } ; CHECK-NEXT: .type: Cs ; CHECK-NEXT: .user_data_limit: 0x3 ; CHECK-NEXT: amdpal.version: -; CHECK-NEXT: - 0x2 -; CHECK-NEXT: - 0x6 +; CHECK-NEXT: - 0x3 +; CHECK-NEXT: - 0 ; CHECK-NEXT: ... diff --git a/lgc/test/BuiltIns/cs-subgroupid.lgc b/lgc/test/BuiltIns/cs-subgroupid.lgc index 998390d5b4..f41b1f3718 100644 --- a/lgc/test/BuiltIns/cs-subgroupid.lgc +++ b/lgc/test/BuiltIns/cs-subgroupid.lgc @@ -1,5 +1,4 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=WAVE64 %s ; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=WAVE32 %s ; RUN: lgc -mcpu=gfx1030 - < %s | FileCheck --check-prefixes=GFX1030 %s @@ -28,20 +27,6 @@ attributes #0 = { nounwind } !2 = !{!"DescriptorBuffer", i32 6, i32 0, i32 0, i32 4, i32 0, i32 0, i32 4} ; Compute mode, containing workgroup size !3 = !{i32 5, i32 6, i32 7} -; WAVE64-LABEL: amdgpu_cs_main: -; WAVE64: v_mul_lo_u32 v2, v2, 6 -; WAVE64-NEXT: s_getpc_b64 s[2:3] -; WAVE64-NEXT: s_mov_b32 s0, s1 -; WAVE64-NEXT: s_mov_b32 s1, s3 -; WAVE64-NEXT: v_add_u32_e32 v1, vcc, v2, v1 -; WAVE64-NEXT: v_mul_lo_u32 v1, v1, 5 -; WAVE64-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 -; WAVE64-NEXT: v_add_u32_e32 v0, vcc, v1, v0 -; WAVE64-NEXT: v_lshrrev_b32_e32 v0, 6, v0 -; WAVE64-NEXT: s_waitcnt lgkmcnt(0) -; WAVE64-NEXT: buffer_store_dword v0, off, s[0:3], 0 -; WAVE64-NEXT: s_endpgm -; ; WAVE32-LABEL: amdgpu_cs_main: ; WAVE32: s_getpc_b64 s[2:3] ; WAVE32-NEXT: s_mov_b32 s0, s1 diff --git a/lgc/test/BuiltIns/cs-subgrouplocalinvocationid.lgc b/lgc/test/BuiltIns/cs-subgrouplocalinvocationid.lgc index cc2bdbee8a..81686f4469 100644 --- a/lgc/test/BuiltIns/cs-subgrouplocalinvocationid.lgc +++ b/lgc/test/BuiltIns/cs-subgrouplocalinvocationid.lgc @@ -1,5 +1,4 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=WAVE64 %s ; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=WAVE32 %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { @@ -24,17 +23,6 @@ attributes #0 = { nounwind } !1 = !{!"DescriptorTableVaPtr", i32 0, i32 0, i32 2, i32 1, i32 1} ; type, offset, size, set, binding, stride !2 = !{!"DescriptorBuffer", i32 6, i32 0, i32 0, i32 4, i32 0, i32 0, i32 4} -; WAVE64-LABEL: amdgpu_cs_main: -; WAVE64: s_getpc_b64 s[2:3] -; WAVE64-NEXT: s_mov_b32 s0, s1 -; WAVE64-NEXT: s_mov_b32 s1, s3 -; WAVE64-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 -; WAVE64-NEXT: v_mbcnt_lo_u32_b32 v0, -1, 0 -; WAVE64-NEXT: v_mbcnt_hi_u32_b32 v0, -1, v0 -; WAVE64-NEXT: s_waitcnt lgkmcnt(0) -; WAVE64-NEXT: buffer_store_dword v0, off, s[0:3], 0 -; WAVE64-NEXT: s_endpgm -; ; WAVE32-LABEL: amdgpu_cs_main: ; WAVE32: s_getpc_b64 s[2:3] ; WAVE32-NEXT: s_mov_b32 s0, s1 diff --git a/lgc/test/BuiltIns/cs-subgroupsize.lgc b/lgc/test/BuiltIns/cs-subgroupsize.lgc index 97e557fbb8..77e0344996 100644 --- a/lgc/test/BuiltIns/cs-subgroupsize.lgc +++ b/lgc/test/BuiltIns/cs-subgroupsize.lgc @@ -1,5 +1,4 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=WAVE64 %s ; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=WAVE32 %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { @@ -25,16 +24,6 @@ attributes #0 = { nounwind } ; type, offset, size, set, binding, stride !2 = !{!"DescriptorBuffer", i32 6, i32 0, i32 0, i32 4, i32 0, i32 0, i32 4} -; WAVE64-LABEL: amdgpu_cs_main: -; WAVE64: s_getpc_b64 s[2:3] -; WAVE64-NEXT: s_mov_b32 s0, s1 -; WAVE64-NEXT: s_mov_b32 s1, s3 -; WAVE64-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 -; WAVE64-NEXT: v_mov_b32_e32 v0, 64 -; WAVE64-NEXT: s_waitcnt lgkmcnt(0) -; WAVE64-NEXT: buffer_store_dword v0, off, s[0:3], 0 -; WAVE64-NEXT: s_endpgm -; ; WAVE32-LABEL: amdgpu_cs_main: ; WAVE32: s_getpc_b64 s[2:3] ; WAVE32-NEXT: s_mov_b32 s0, s1 diff --git a/lgc/test/BuiltIns/cs-workgroupid.lgc b/lgc/test/BuiltIns/cs-workgroupid.lgc index 839f22feaa..abcbc4ba71 100644 --- a/lgc/test/BuiltIns/cs-workgroupid.lgc +++ b/lgc/test/BuiltIns/cs-workgroupid.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc --check-pal-metadata -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=CHECK %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=CHECK %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: @@ -28,8 +28,8 @@ attributes #0 = { nounwind } ; CHECK: s_getpc_b64 s[6:7] ; CHECK-NEXT: s_mov_b32 s0, s1 ; CHECK-NEXT: s_mov_b32 s1, s7 -; CHECK-NEXT: s_load_dwordx4 s[8:11], s[0:1], 0x0 ; CHECK-NEXT: v_mov_b32_e32 v0, s2 +; CHECK-NEXT: s_load_dwordx4 s[8:11], s[0:1], 0x0 ; CHECK-NEXT: v_mov_b32_e32 v1, s3 ; CHECK-NEXT: v_mov_b32_e32 v2, s4 ; CHECK-NEXT: s_waitcnt lgkmcnt(0) @@ -40,29 +40,73 @@ attributes #0 = { nounwind } ; CHECK-NEXT: --- ; CHECK-NEXT: amdpal.pipelines: ; CHECK-NEXT: - .api: !str '' +; CHECK-NEXT: .compute_registers: +; CHECK-NEXT: .tg_size_en: true +; CHECK-NEXT: .tgid_x_en: true +; CHECK-NEXT: .tgid_y_en: true +; CHECK-NEXT: .tgid_z_en: true +; CHECK-NEXT: .tidig_comp_cnt: 0 ; CHECK-NEXT: .hardware_stages: ; CHECK-NEXT: .cs: +; CHECK-NEXT: .checksum_value: 0 +; CHECK-NEXT: .debug_mode: false ; CHECK-NEXT: .entry_point: _amdgpu_cs_main +; CHECK-NEXT: .excp_en: 0 +; CHECK-NEXT: .float_mode: 0xc0 +; CHECK-NEXT: .ieee_mode: false +; CHECK-NEXT: .lds_size: 0 +; CHECK-NEXT: .mem_ordered: true +; CHECK-NEXT: .scratch_en: false ; CHECK-NEXT: .scratch_memory_size: 0 -; CHECK-NEXT: .sgpr_count: 0x60 -; CHECK-NEXT: .sgpr_limit: 0x66 +; CHECK-NEXT: .sgpr_count: 0xc +; CHECK-NEXT: .sgpr_limit: 0x6a ; CHECK-NEXT: .threadgroup_dimensions: ; CHECK-NEXT: - 0 ; CHECK-NEXT: - 0 ; CHECK-NEXT: - 0 +; CHECK-NEXT: .trap_present: false +; CHECK-NEXT: .user_data_reg_map: +; CHECK-NEXT: - 0x10000000 +; CHECK-NEXT: - 0x2 +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: - 0xffffffff +; CHECK-NEXT: .user_sgprs: 0x2 ; CHECK-NEXT: .vgpr_count: 0x3 ; CHECK-NEXT: .vgpr_limit: 0x100 +; CHECK-NEXT: .wavefront_size: 0x20 +; CHECK-NEXT: .wgp_mode: false ; CHECK-NEXT: .internal_pipeline_hash: ; CHECK-NEXT: - 0 ; CHECK-NEXT: - 0 -; CHECK-NEXT: .registers: -; CHECK-NEXT: 0x2e07 (COMPUTE_NUM_THREAD_X): 0 -; CHECK-NEXT: 0x2e08 (COMPUTE_NUM_THREAD_Y): 0 -; CHECK-NEXT: 0x2e09 (COMPUTE_NUM_THREAD_Z): 0 -; CHECK-NEXT: 0x2e12 (COMPUTE_PGM_RSRC1): 0x2c02c0 -; CHECK-NEXT: 0x2e13 (COMPUTE_PGM_RSRC2): 0x784 -; CHECK-NEXT: 0x2e40 (COMPUTE_USER_DATA_0): 0x10000000 -; CHECK-NEXT: 0x2e41 (COMPUTE_USER_DATA_1): 0x2 +; CHECK-NEXT: .registers: {} ; CHECK-NEXT: .shaders: ; CHECK-NEXT: .compute: ; CHECK-NEXT: .api_shader_hash: @@ -74,6 +118,6 @@ attributes #0 = { nounwind } ; CHECK-NEXT: .type: Cs ; CHECK-NEXT: .user_data_limit: 0x3 ; CHECK-NEXT: amdpal.version: -; CHECK-NEXT: - 0x2 -; CHECK-NEXT: - 0x6 +; CHECK-NEXT: - 0x3 +; CHECK-NEXT: - 0 ; CHECK-NEXT: ... diff --git a/lgc/test/BuiltIns/cs-workgroupsize.lgc b/lgc/test/BuiltIns/cs-workgroupsize.lgc index 5a6c911d24..d79a928435 100644 --- a/lgc/test/BuiltIns/cs-workgroupsize.lgc +++ b/lgc/test/BuiltIns/cs-workgroupsize.lgc @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by tool/update_llpc_test_checks.py UTC_ARGS: --tool lgc -; RUN: lgc -mcpu=gfx802 - < %s | FileCheck --check-prefixes=CHECK %s +; RUN: lgc -mcpu=gfx1010 - < %s | FileCheck --check-prefixes=CHECK %s define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: @@ -31,8 +31,8 @@ attributes #0 = { nounwind } ; CHECK: s_getpc_b64 s[2:3] ; CHECK-NEXT: s_mov_b32 s0, s1 ; CHECK-NEXT: s_mov_b32 s1, s3 -; CHECK-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 ; CHECK-NEXT: v_mov_b32_e32 v0, 5 +; CHECK-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x0 ; CHECK-NEXT: v_mov_b32_e32 v1, 6 ; CHECK-NEXT: v_mov_b32_e32 v2, 7 ; CHECK-NEXT: s_waitcnt lgkmcnt(0) diff --git a/lgc/test/CsReconfigWorkgroup.lgc b/lgc/test/CsReconfigWorkgroup.lgc index c6d2f0a91b..f4f44976ce 100644 --- a/lgc/test/CsReconfigWorkgroup.lgc +++ b/lgc/test/CsReconfigWorkgroup.lgc @@ -1,11 +1,12 @@ ; ---------------------------------------------------------------------- ; Extract 1: Reconfiguring of workgroup size disabled -; RUN: lgc -extract=1 -mcpu=gfx802 %s -o - | FileCheck --check-prefixes=CHECK1 %s +; RUN: lgc -extract=1 -mcpu=gfx1010 %s -o - | FileCheck --check-prefixes=CHECK1 %s ; CHECK1-LABEL: _amdgpu_cs_main: -; CHECK1: COMPUTE_NUM_THREAD_X): 0x5 -; CHECK1: COMPUTE_NUM_THREAD_Y): 0x6 -; CHECK1: COMPUTE_NUM_THREAD_Z): 0x7 +; CHECK1: .threadgroup_dimensions: +; CHECK1: - 0x5 +; CHECK1: - 0x6 +; CHECK1: - 0x7 define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: @@ -36,11 +37,12 @@ attributes #0 = { nounwind } ; ---------------------------------------------------------------------- ; Extract 2: Reconfiguring of workgroup size uses 8x8 -; RUN: lgc -extract=2 -mcpu=gfx802 %s -o - | FileCheck --check-prefixes=CHECK2 %s +; RUN: lgc -extract=2 -mcpu=gfx1010 %s -o - | FileCheck --check-prefixes=CHECK2 %s ; CHECK2-LABEL: _amdgpu_cs_main: -; CHECK2: COMPUTE_NUM_THREAD_X): 0x50 -; CHECK2: COMPUTE_NUM_THREAD_Y): 0x6 -; CHECK2: COMPUTE_NUM_THREAD_Z): 0x1 +; CHECK2: .threadgroup_dimensions: +; CHECK2: - 0x50 +; CHECK2: - 0x6 +; CHECK2: - 0x7 define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: @@ -83,11 +85,12 @@ attributes #0 = { nounwind } ; ---------------------------------------------------------------------- ; Extract 3: Reconfiguring of workgroup size uses 2x2 -; RUN: lgc -extract=3 -mcpu=gfx802 %s -o - | FileCheck --check-prefixes=CHECK3 %s +; RUN: lgc -extract=3 -mcpu=gfx1010 %s -o - | FileCheck --check-prefixes=CHECK3 %s ; CHECK3-LABEL: _amdgpu_cs_main: -; CHECK3: COMPUTE_NUM_THREAD_X): 0x24 -; CHECK3: COMPUTE_NUM_THREAD_Y): 0x5 -; CHECK3: COMPUTE_NUM_THREAD_Z): 0x1 +; CHECK3: .threadgroup_dimensions: +; CHECK3: - 0x24 +; CHECK3: - 0x5 +; CHECK3: - 0x1 define dllexport spir_func void @lgc.shader.CS.main() local_unnamed_addr #0 !lgc.shaderstage !0 { .entry: diff --git a/lgc/test/ElfRelocationAndNote.lgc b/lgc/test/ElfRelocationAndNote.lgc index ace614b4b3..4702256be2 100644 --- a/lgc/test/ElfRelocationAndNote.lgc +++ b/lgc/test/ElfRelocationAndNote.lgc @@ -2,16 +2,16 @@ ; is no relocation. It also checks the ISA name is correctly generated in the ; .note section. -; RUN: lgc -mcpu=gfx802 -extract=2 -filetype=obj -o %t.vs.elf %s -; RUN: lgc -mcpu=gfx802 -extract=3 -filetype=obj -o %t.fs.elf %s -; RUN: lgc -mcpu=gfx802 -extract=1 -l -o %t.pipe.elf %s %t.vs.elf %t.fs.elf +; RUN: lgc -mcpu=gfx1010 -extract=2 -filetype=obj -o %t.vs.elf %s +; RUN: lgc -mcpu=gfx1010 -extract=3 -filetype=obj -o %t.fs.elf %s +; RUN: lgc -mcpu=gfx1010 -extract=1 -l -o %t.pipe.elf %s %t.vs.elf %t.fs.elf ; RUN: llvm-readelf %t.pipe.elf --section-headers --elf-output-style=LLVM | FileCheck %s --check-prefix=EMPTY ; RUN: llvm-objdump -sj .note %t.pipe.elf | FileCheck %s --check-prefix=ISA ; EMPTY-NOT: .rel.text ; ISA: amdgcn--amdpal-- -; ISA-NEXT: gfx802 +; ISA-NEXT: gfx1010 ; ---------------------------------------------------------------------- ; Extract 1: The pipeline state with no shaders. diff --git a/lgc/test/FetchShaderSingleInput.lgc b/lgc/test/FetchShaderSingleInput.lgc index 0cb1799ccc..27c279cafe 100644 --- a/lgc/test/FetchShaderSingleInput.lgc +++ b/lgc/test/FetchShaderSingleInput.lgc @@ -1,10 +1,7 @@ ; Test building the vertex shader by itself. -; RUN: lgc -mcpu=gfx802 -extract=2 - <%s | FileCheck --check-prefixes=VS-ISA,VS-NOTNGG-ISA %s -; RUN: lgc -mcpu=gfx900 -extract=2 - <%s | FileCheck --check-prefixes=VS-ISA,VS-NOTNGG-ISA %s ; RUN: lgc -mcpu=gfx1010 -extract=2 - <%s | FileCheck --check-prefixes=VS-ISA,VS-NGG-ISA %s ; VS-ISA: .p2align 8 -; VS-NOTNGG-ISA-LABEL: _amdgpu_vs_main_fetchless: ; VS-NGG-ISA-LABEL: _amdgpu_gs_main_fetchless: ; VS-ISA: exp pos0 ; VS-ISA: s_endpgm @@ -15,13 +12,8 @@ ; Test building the vertex and fragment shaders to separate ELFs, then building the fetch shader to ISA. -; RUN: lgc -mcpu=gfx802 -extract=2 -filetype=obj -o %t.vs.elf - <%s && lgc -mcpu=gfx802 -extract=3 -filetype=obj -o %t.fs.elf - <%s && lgc -mcpu=gfx802 -extract=1 -o - -l -glue=1 %s %t.vs.elf %t.fs.elf | FileCheck -check-prefixes=FETCH-ISA,FETCH-ISA-NOTNGG %s -; RUN: lgc -mcpu=gfx900 -extract=2 -filetype=obj -o %t.vs.elf - <%s && lgc -mcpu=gfx900 -extract=3 -filetype=obj -o %t.fs.elf - <%s && lgc -mcpu=gfx900 -extract=1 -o - -l -glue=1 %s %t.vs.elf %t.fs.elf | FileCheck -check-prefixes=FETCH-ISA,FETCH-ISA-NOTNGG %s ; RUN: lgc -mcpu=gfx1010 -extract=2 -filetype=obj -o %t.vs.elf - <%s && lgc -mcpu=gfx1010 -extract=3 -filetype=obj -o %t.fs.elf - <%s && lgc -mcpu=gfx1010 -extract=1 -o - -l -glue=1 %s %t.vs.elf %t.fs.elf | FileCheck -check-prefixes=FETCH-ISA,FETCH-ISA-NGG %s ; FETCH-ISA: .p2align 8 -; FETCH-ISA-NOTNGG: _amdgpu_vs_main: -; FETCH-ISA-NOTNGG-NOT: s_bfm_b32 exec -; FETCH-ISA-NOTNGG: tbuffer_load_format_xyz v[4:6], ; FETCH-ISA-NGG: _amdgpu_gs_main: ; FETCH-ISA-NGG: s_bfe_u32 {{.*}}, s3, 0x70000 ; FETCH-ISA-NGG: s_bfm_b32 exec diff --git a/lgc/test/IntVectorVertexInput.lgc b/lgc/test/IntVectorVertexInput.lgc index 370f70b5cb..7d0aa77fe9 100644 --- a/lgc/test/IntVectorVertexInput.lgc +++ b/lgc/test/IntVectorVertexInput.lgc @@ -1,7 +1,7 @@ ; Test with a vector of integers are a vertex attribute input. -; RUN: lgc -mcpu=gfx802 -extract=2 -filetype=obj -o %t.vs.elf - <%s && lgc -mcpu=gfx802 -extract=3 -filetype=obj -o %t.fs.elf - <%s && lgc -mcpu=gfx802 -extract=1 -o - -l -glue=1 %s %t.vs.elf %t.fs.elf | FileCheck -check-prefixes=FETCH-ISA %s -; FETCH-ISA: tbuffer_load_format_xyzw v[4:7], +; RUN: lgc -mcpu=gfx1010 -extract=2 -filetype=obj -o %t.vs.elf - <%s && lgc -mcpu=gfx1010 -extract=3 -filetype=obj -o %t.fs.elf - <%s && lgc -mcpu=gfx1010 -extract=1 -o - -l -glue=1 %s %t.vs.elf %t.fs.elf | FileCheck -check-prefixes=FETCH-ISA %s +; FETCH-ISA: tbuffer_load_format_xyzw v[{{[0-9]+:[0-9]+}}], ; ---------------------------------------------------------------------- ; Extract 1: The pipeline state with no shaders. diff --git a/lgc/tool/lgc/lgc.cpp b/lgc/tool/lgc/lgc.cpp index adc2f70ed4..adc6f371a0 100644 --- a/lgc/tool/lgc/lgc.cpp +++ b/lgc/tool/lgc/lgc.cpp @@ -239,7 +239,7 @@ int main(int argc, char **argv) { auto *mcpuOpt = reinterpret_cast *>(mcpu->second); StringRef gpuName = *mcpuOpt; if (gpuName == "") - gpuName = "gfx802"; + gpuName = "gfx1010"; // Default to reading from stdin and writing to stdout if (InFiles.empty())