/third_party/mesa3d/src/amd/vulkan/ |
D | radv_llvm_helper.cpp | 31 : family(arg_family), tm_options(arg_tm_options), wave_size(arg_wave_size), passes(NULL) in radv_llvm_per_thread_info() 60 if (arg_family == family && arg_tm_options == tm_options && arg_wave_size == wave_size) in is_same() 69 unsigned wave_size; member in radv_llvm_per_thread_info 101 enum ac_target_machine_options tm_options, unsigned wave_size) in radv_init_llvm_compiler() argument 104 if (I.is_same(family, tm_options, wave_size)) { in radv_init_llvm_compiler() 110 radv_llvm_per_thread_list.emplace_back(family, tm_options, wave_size); in radv_init_llvm_compiler()
|
D | radv_shader.c | 1009 info->wave_size, in radv_lower_ngg() 1020 nir, info->wave_size, info->workgroup_size, in radv_lower_ngg() 1367 config_out->rsrc1 = S_00B848_VGPRS((num_vgprs - 1) / (info->wave_size == 32 ? 8 : 4)) | in radv_postprocess_config() 1601 .wave_size = binary->info.wave_size, in radv_shader_variant_create() 1906 info.wave_size = 64; 1918 upload_vs_prolog(struct radv_device *device, struct radv_prolog_binary *bin, unsigned wave_size) argument 1935 prolog->rsrc1 = S_00B848_VGPRS((bin->num_vgprs - 1) / (wave_size == 32 ? 8 : 4)) | 1954 info.wave_size = key->wave32 ? 32 : 64; 1977 struct radv_shader_prolog *prolog = upload_vs_prolog(device, binary, info.wave_size); 2084 uint8_t wave_size = variant->info.wave_size; local [all …]
|
D | radv_llvm_helper.h | 33 enum ac_target_machine_options tm_options, unsigned wave_size);
|
D | radv_pipeline.c | 1942 pipeline->device->physical_device->rad_info.chip_class, infos[es_stage].wave_size, in gfx9_get_gs_info() 2116 wavesize = gs_info->wave_size; in gfx10_get_ngg_info() 2118 wavesize = nir[MESA_SHADER_TESS_CTRL] ? infos[MESA_SHADER_TESS_EVAL].wave_size in gfx10_get_ngg_info() 2119 : infos[MESA_SHADER_VERTEX].wave_size; in gfx10_get_ngg_info() 2223 unsigned wave_size = 64; in radv_pipeline_init_gs_ring_state() local 2237 align(gs->vgt_esgs_ring_itemsize * 4 * gs_vertex_reuse * wave_size, alignment); in radv_pipeline_init_gs_ring_state() 2240 max_gs_waves * 2 * wave_size * gs->vgt_esgs_ring_itemsize * 4 * gs_info->gs.vertices_in; in radv_pipeline_init_gs_ring_state() 2241 unsigned gsvs_ring_size = max_gs_waves * 2 * wave_size * gs_info->gs.max_gsvs_emit_size; in radv_pipeline_init_gs_ring_state() 2974 infos[i].wave_size = radv_get_wave_size(pipeline->device, pStages[i], i, &infos[i]); in radv_fill_shader_info() 2982 infos[MESA_SHADER_FRAGMENT].workgroup_size = infos[MESA_SHADER_FRAGMENT].wave_size; in radv_fill_shader_info() [all …]
|
/third_party/mesa3d/src/amd/common/ |
D | ac_binary.c | 39 void ac_parse_shader_binary_config(const char *data, size_t nbytes, unsigned wave_size, in ac_parse_shader_binary_config() argument 54 if (wave_size == 32 || info->wave64_vgpr_alloc_granularity == 8) in ac_parse_shader_binary_config() 136 conf->num_vgprs = align(conf->num_vgprs, wave_size == 32 ? 16 : 8); in ac_parse_shader_binary_config()
|
D | ac_rtld.h | 59 unsigned wave_size; member 96 unsigned wave_size; member
|
D | ac_nir.h | 99 unsigned wave_size, 110 unsigned wave_size,
|
D | ac_nir_lower_ngg.c | 55 unsigned wave_size; member 84 unsigned wave_size; member 210 unsigned wave_size) in repack_invocations_in_workgroup() argument 220 nir_ssa_def *input_mask = nir_build_ballot(b, 1, wave_size, input_bool); in repack_invocations_in_workgroup() 1196 … nogs_state->max_num_waves, nogs_state->wave_size); in add_deferred_attribute_culling() 1268 unsigned wave_size, in ac_nir_lower_ngg_nogs() argument 1279 assert(max_num_es_vertices && max_workgroup_size && wave_size); in ac_nir_lower_ngg_nogs() 1298 .max_num_waves = DIV_ROUND_UP(max_workgroup_size, wave_size), in ac_nir_lower_ngg_nogs() 1300 .wave_size = wave_size, in ac_nir_lower_ngg_nogs() 1486 …nir_ssa_def *num_threads = nir_bit_count(b, nir_build_ballot(b, 1, s->wave_size, nir_imm_bool(b, t… in ngg_gs_shader_query() [all …]
|
D | ac_binary.h | 53 void ac_parse_shader_binary_config(const char *data, size_t nbytes, unsigned wave_size,
|
D | ac_shader_util.h | 115 unsigned ac_compute_esgs_workgroup_size(enum chip_class chip_class, unsigned wave_size,
|
D | ac_nir_lower_esgs_io_to_mem.c | 219 unsigned wave_size = 64u; /* GFX6-8 only support wave64 */ in lower_gs_per_vertex_input_load() local 221 return emit_split_buffer_load(b, ring, off, nir_imm_zero(b, 1, 32), 4u * wave_size, in lower_gs_per_vertex_input_load()
|
D | ac_shader_util.c | 547 unsigned ac_compute_esgs_workgroup_size(enum chip_class chip_class, unsigned wave_size, in ac_compute_esgs_workgroup_size() argument 558 return wave_size; in ac_compute_esgs_workgroup_size()
|
D | ac_rtld.c | 259 binary->wave_size = i.wave_size; in ac_rtld_open() 529 ac_parse_shader_binary_config(config_data, config_nbytes, binary->wave_size, true, info, &c); in ac_rtld_read_config()
|
/third_party/mesa3d/src/amd/compiler/ |
D | aco_spill.cpp | 83 unsigned wave_size; member 90 processed(program->blocks.size(), false), wave_size(program->wave_size) in spill_ctx() 1414 S_008F0C_ADD_TID_ENABLE(1) | S_008F0C_INDEX_STRIDE(ctx.program->wave_size == 64 ? 3 : 2); in load_scratch_resource() 1447 find_available_slot(std::vector<bool>& used, unsigned wave_size, unsigned size, bool is_sgpr, in find_available_slot() argument 1450 unsigned wave_size_minus_one = wave_size - 1; in find_available_slot() 1466 if (is_sgpr && ((slot & wave_size_minus_one) > wave_size - size)) { in find_available_slot() 1467 slot = align(slot, wave_size); in find_available_slot() 1499 find_available_slot(slots_used, ctx.wave_size, ctx.interferences[vec[0]].first.size(), in assign_spill_slots_helper() 1520 find_available_slot(slots_used, ctx.wave_size, ctx.interferences[id].first.size(), in assign_spill_slots_helper() 1574 std::vector<Temp> vgpr_spill_temps((sgpr_spill_slots + ctx.wave_size - 1) / ctx.wave_size); in assign_spill_slots() [all …]
|
D | aco_live_var_analysis.cpp | 271 program->workgroup_size == UINT_MAX ? program->wave_size : program->workgroup_size; in calc_waves_per_workgroup() 273 return align(workgroup_size, program->wave_size) / program->wave_size; in calc_waves_per_workgroup() 355 unsigned max_waves_per_simd = program->dev.max_wave64_per_simd * (64 / program->wave_size); in update_vgpr_sgpr_demand()
|
D | aco_ir.cpp | 88 program->wave_size = info->wave_size; in init_program() 89 program->lane_mask = program->wave_size == 32 ? s1 : s2; in init_program() 104 program->dev.physical_vgprs = program->wave_size == 32 ? 1024 : 512; in init_program() 109 program->dev.vgpr_alloc_granule = program->wave_size == 32 ? 16 : 8; in init_program() 111 program->dev.vgpr_alloc_granule = program->wave_size == 32 ? 8 : 4; in init_program()
|
D | aco_statistics.cpp | 360 bool dual_issue = program->chip_class >= GFX10 && program->wave_size == 64 && in add() 539 double wave64_per_cycle = waves_per_cycle * (program->wave_size / 64.0); in collect_preasm_stats() 544 program->workgroup_size / (double)align(program->workgroup_size, program->wave_size); in collect_preasm_stats()
|
/third_party/mesa3d/src/amd/compiler/tests/ |
D | helpers.h | 74 unsigned wave_size=64, enum radeon_family family=CHIP_UNKNOWN); 77 unsigned wave_size=64);
|
D | helpers.cpp | 75 void create_program(enum chip_class chip_class, Stage stage, unsigned wave_size, enum radeon_family… in create_program() argument 78 info.wave_size = wave_size; in create_program() 103 unsigned wave_size) in setup_cs() argument 113 create_program(chip_class, compute_cs, wave_size, family); in setup_cs() 180 program->workgroup_size = program->wave_size; in finish_ra_test()
|
/third_party/mesa3d/src/gallium/drivers/radeonsi/ |
D | si_debug.c | 854 gl_shader_stage stage, unsigned wave_size) in si_add_split_disasm() argument 859 .wave_size = wave_size, in si_add_split_disasm() 928 unsigned wave_size = si_get_shader_wave_size(shader); in si_print_annotated_shader() local 935 instructions, stage, wave_size); in si_print_annotated_shader() 939 &num_inst, instructions, stage, wave_size); in si_print_annotated_shader() 943 &num_inst, instructions, stage, wave_size); in si_print_annotated_shader() 946 instructions, stage, wave_size); in si_print_annotated_shader() 949 instructions, stage, wave_size); in si_print_annotated_shader()
|
D | si_compute_blit.c | 175 unsigned wave_size = sctx->screen->compute_wave_size; in si_compute_clear_buffer_rmw() local 176 unsigned dwords_per_wave = dwords_per_instruction * wave_size; in si_compute_clear_buffer_rmw() 182 info.block[0] = MIN2(wave_size, num_instructions); in si_compute_clear_buffer_rmw() 260 unsigned wave_size = sctx->screen->compute_wave_size; in si_compute_do_clear_or_copy() local 261 unsigned dwords_per_wave = dwords_per_thread * wave_size; in si_compute_do_clear_or_copy() 267 info.block[0] = MIN2(wave_size, num_instructions); in si_compute_do_clear_or_copy()
|
D | si_shader.c | 810 .wave_size = si_get_shader_wave_size(shader), in si_shader_binary_open() 896 gl_shader_stage stage, unsigned wave_size, in si_shader_dump_disassembly() argument 905 .wave_size = wave_size, in si_shader_dump_disassembly() 1114 unsigned wave_size = si_get_shader_wave_size(shader); in si_shader_dump() local 1119 si_shader_dump_disassembly(sscreen, &shader->prolog->binary, stage, wave_size, debug, in si_shader_dump() 1123 wave_size, debug, "previous stage", file); in si_shader_dump() 1125 si_shader_dump_disassembly(sscreen, &shader->prolog2->binary, stage, wave_size, in si_shader_dump() 1128 si_shader_dump_disassembly(sscreen, &shader->binary, stage, wave_size, debug, "main", in si_shader_dump() 1132 si_shader_dump_disassembly(sscreen, &shader->epilog->binary, stage, wave_size, debug, in si_shader_dump() 1479 unsigned wave_size = sscreen->compute_wave_size; in si_compile_shader() local [all …]
|
D | gfx10_shader_ngg.c | 46 LLVMConstInt(ctx->ac.i32, ctx->ac.wave_size, false), ""); in get_thread_id_in_tg() 481 primemit_scan[stream].maxwaves = DIV_ROUND_UP(256, ctx->ac.wave_size); in build_streamout() 484 ctx->ac.wave_size); in build_streamout() 777 LLVMConstInt(ctx->ac.i32, ctx->ac.wave_size, 0), ""); in update_thread_counts() 781 ac_build_imin(&ctx->ac, *new_num_threads, LLVMConstInt(ctx->ac.i32, ctx->ac.wave_size, 0)); in update_thread_counts() 824 unsigned max_waves = DIV_ROUND_UP(ctx->screen->ngg_subgroup_size, ctx->ac.wave_size); in gfx10_emit_ngg_culling_epilogue() 1129 LLVMConstInt(ctx->ac.i32, ctx->ac.wave_size, 0), ""), in gfx10_emit_ngg_culling_epilogue() 1763 numprims = ac_build_reduce(&ctx->ac, numprims, nir_op_iadd, ctx->ac.wave_size); in gfx10_ngg_gs_emit_epilogue() 1883 vertlive_scan.maxwaves = DIV_ROUND_UP(256, ctx->ac.wave_size); in gfx10_ngg_gs_emit_epilogue()
|
D | si_shader_llvm_gs.c | 152 LLVMConstInt(ctx->ac.i32, ctx->ac.wave_size, false), ""), in si_llvm_emit_es_epilogue() 369 num_records = ctx->ac.wave_size; in si_preload_gs_rings() 374 stream_offset += stride * ctx->ac.wave_size; in si_preload_gs_rings()
|
/third_party/mesa3d/src/amd/llvm/ |
D | ac_llvm_util.c | 324 ctx->chip_class >= GFX10 && ctx->wave_size == 64 ? in ac_llvm_set_target_features()
|