/* * Copyright 2013 Advanced Micro Devices, Inc. * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * on the rights to use, copy, modify, merge, publish, distribute, sub * license, and/or sell copies of the Software, and to permit persons to whom * the Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE * USE OR OTHER DEALINGS IN THE SOFTWARE. * */ #include "tgsi/tgsi_parse.h" #include "util/u_async_debug.h" #include "util/u_memory.h" #include "util/u_upload_mgr.h" #include "amd_kernel_code_t.h" #include "si_build_pm4.h" #include "si_compute.h" #define COMPUTE_DBG(sscreen, fmt, args...) \ do { \ if ((sscreen->debug_flags & DBG(COMPUTE))) fprintf(stderr, fmt, ##args); \ } while (0); struct dispatch_packet { uint16_t header; uint16_t setup; uint16_t workgroup_size_x; uint16_t workgroup_size_y; uint16_t workgroup_size_z; uint16_t reserved0; uint32_t grid_size_x; uint32_t grid_size_y; uint32_t grid_size_z; uint32_t private_segment_size; uint32_t group_segment_size; uint64_t kernel_object; uint64_t kernarg_address; uint64_t reserved2; }; static const amd_kernel_code_t *si_compute_get_code_object( const struct si_compute *program, uint64_t symbol_offset) { if (!program->use_code_object_v2) { return NULL; } return (const amd_kernel_code_t*) (program->shader.binary.code + symbol_offset); } static void code_object_to_config(const amd_kernel_code_t *code_object, struct si_shader_config *out_config) { uint32_t rsrc1 = code_object->compute_pgm_resource_registers; uint32_t rsrc2 = code_object->compute_pgm_resource_registers >> 32; out_config->num_sgprs = code_object->wavefront_sgpr_count; out_config->num_vgprs = code_object->workitem_vgpr_count; out_config->float_mode = G_00B028_FLOAT_MODE(rsrc1); out_config->rsrc1 = rsrc1; out_config->lds_size = MAX2(out_config->lds_size, G_00B84C_LDS_SIZE(rsrc2)); out_config->rsrc2 = rsrc2; out_config->scratch_bytes_per_wave = align(code_object->workitem_private_segment_byte_size * 64, 1024); } /* Asynchronous compute shader compilation. */ static void si_create_compute_state_async(void *job, int thread_index) { struct si_compute *program = (struct si_compute *)job; struct si_shader *shader = &program->shader; struct si_shader_selector sel; struct ac_llvm_compiler *compiler; struct pipe_debug_callback *debug = &program->compiler_ctx_state.debug; struct si_screen *sscreen = program->screen; assert(!debug->debug_message || debug->async); assert(thread_index >= 0); assert(thread_index < ARRAY_SIZE(sscreen->compiler)); compiler = &sscreen->compiler[thread_index]; memset(&sel, 0, sizeof(sel)); sel.screen = sscreen; if (program->ir_type == PIPE_SHADER_IR_TGSI) { tgsi_scan_shader(program->ir.tgsi, &sel.info); sel.tokens = program->ir.tgsi; } else { assert(program->ir_type == PIPE_SHADER_IR_NIR); sel.nir = program->ir.nir; si_nir_opts(sel.nir); si_nir_scan_shader(sel.nir, &sel.info); si_lower_nir(&sel); } /* Store the declared LDS size into tgsi_shader_info for the shader * cache to include it. */ sel.info.properties[TGSI_PROPERTY_CS_LOCAL_SIZE] = program->local_size; sel.type = PIPE_SHADER_COMPUTE; si_get_active_slot_masks(&sel.info, &program->active_const_and_shader_buffers, &program->active_samplers_and_images); program->shader.selector = &sel; program->shader.is_monolithic = true; program->uses_grid_size = sel.info.uses_grid_size; program->uses_bindless_samplers = sel.info.uses_bindless_samplers; program->uses_bindless_images = sel.info.uses_bindless_images; program->reads_variable_block_size = sel.info.uses_block_size && sel.info.properties[TGSI_PROPERTY_CS_FIXED_BLOCK_WIDTH] == 0; program->num_cs_user_data_dwords = sel.info.properties[TGSI_PROPERTY_CS_USER_DATA_DWORDS]; void *ir_binary = si_get_ir_binary(&sel); /* Try to load the shader from the shader cache. */ mtx_lock(&sscreen->shader_cache_mutex); if (ir_binary && si_shader_cache_load_shader(sscreen, ir_binary, shader)) { mtx_unlock(&sscreen->shader_cache_mutex); si_shader_dump_stats_for_shader_db(shader, debug); si_shader_dump(sscreen, shader, debug, PIPE_SHADER_COMPUTE, stderr, true); if (si_shader_binary_upload(sscreen, shader)) program->shader.compilation_failed = true; } else { mtx_unlock(&sscreen->shader_cache_mutex); if (si_shader_create(sscreen, compiler, &program->shader, debug)) { program->shader.compilation_failed = true; if (program->ir_type == PIPE_SHADER_IR_TGSI) FREE(program->ir.tgsi); program->shader.selector = NULL; return; } bool scratch_enabled = shader->config.scratch_bytes_per_wave > 0; unsigned user_sgprs = SI_NUM_RESOURCE_SGPRS + (sel.info.uses_grid_size ? 3 : 0) + (program->reads_variable_block_size ? 3 : 0) + program->num_cs_user_data_dwords; shader->config.rsrc1 = S_00B848_VGPRS((shader->config.num_vgprs - 1) / 4) | S_00B848_SGPRS((shader->config.num_sgprs - 1) / 8) | S_00B848_DX10_CLAMP(1) | S_00B848_FLOAT_MODE(shader->config.float_mode); shader->config.rsrc2 = S_00B84C_USER_SGPR(user_sgprs) | S_00B84C_SCRATCH_EN(scratch_enabled) | S_00B84C_TGID_X_EN(sel.info.uses_block_id[0]) | S_00B84C_TGID_Y_EN(sel.info.uses_block_id[1]) | S_00B84C_TGID_Z_EN(sel.info.uses_block_id[2]) | S_00B84C_TIDIG_COMP_CNT(sel.info.uses_thread_id[2] ? 2 : sel.info.uses_thread_id[1] ? 1 : 0) | S_00B84C_LDS_SIZE(shader->config.lds_size); if (ir_binary) { mtx_lock(&sscreen->shader_cache_mutex); if (!si_shader_cache_insert_shader(sscreen, ir_binary, shader, true)) FREE(ir_binary); mtx_unlock(&sscreen->shader_cache_mutex); } } if (program->ir_type == PIPE_SHADER_IR_TGSI) FREE(program->ir.tgsi); program->shader.selector = NULL; } static void *si_create_compute_state( struct pipe_context *ctx, const struct pipe_compute_state *cso) { struct si_context *sctx = (struct si_context *)ctx; struct si_screen *sscreen = (struct si_screen *)ctx->screen; struct si_compute *program = CALLOC_STRUCT(si_compute); pipe_reference_init(&program->reference, 1); program->screen = (struct si_screen *)ctx->screen; program->ir_type = cso->ir_type; program->local_size = cso->req_local_mem; program->private_size = cso->req_private_mem; program->input_size = cso->req_input_mem; program->use_code_object_v2 = cso->ir_type == PIPE_SHADER_IR_NATIVE; if (cso->ir_type != PIPE_SHADER_IR_NATIVE) { if (cso->ir_type == PIPE_SHADER_IR_TGSI) { program->ir.tgsi = tgsi_dup_tokens(cso->prog); if (!program->ir.tgsi) { FREE(program); return NULL; } } else { assert(cso->ir_type == PIPE_SHADER_IR_NIR); program->ir.nir = (struct nir_shader *) cso->prog; } program->compiler_ctx_state.debug = sctx->debug; program->compiler_ctx_state.is_debug_context = sctx->is_debug; p_atomic_inc(&sscreen->num_shaders_created); si_schedule_initial_compile(sctx, PIPE_SHADER_COMPUTE, &program->ready, &program->compiler_ctx_state, program, si_create_compute_state_async); } else { const struct pipe_llvm_program_header *header; const char *code; header = cso->prog; code = cso->prog + sizeof(struct pipe_llvm_program_header); ac_elf_read(code, header->num_bytes, &program->shader.binary); if (program->use_code_object_v2) { const amd_kernel_code_t *code_object = si_compute_get_code_object(program, 0); code_object_to_config(code_object, &program->shader.config); if (program->shader.binary.reloc_count != 0) { fprintf(stderr, "Error: %d unsupported relocations\n", program->shader.binary.reloc_count); FREE(program); return NULL; } } else { si_shader_binary_read_config(&program->shader.binary, &program->shader.config, 0); } si_shader_dump(sctx->screen, &program->shader, &sctx->debug, PIPE_SHADER_COMPUTE, stderr, true); if (si_shader_binary_upload(sctx->screen, &program->shader) < 0) { fprintf(stderr, "LLVM failed to upload shader\n"); FREE(program); return NULL; } } return program; } static void si_bind_compute_state(struct pipe_context *ctx, void *state) { struct si_context *sctx = (struct si_context*)ctx; struct si_compute *program = (struct si_compute*)state; sctx->cs_shader_state.program = program; if (!program) return; /* Wait because we need active slot usage masks. */ if (program->ir_type != PIPE_SHADER_IR_NATIVE) util_queue_fence_wait(&program->ready); si_set_active_descriptors(sctx, SI_DESCS_FIRST_COMPUTE + SI_SHADER_DESCS_CONST_AND_SHADER_BUFFERS, program->active_const_and_shader_buffers); si_set_active_descriptors(sctx, SI_DESCS_FIRST_COMPUTE + SI_SHADER_DESCS_SAMPLERS_AND_IMAGES, program->active_samplers_and_images); } static void si_set_global_binding( struct pipe_context *ctx, unsigned first, unsigned n, struct pipe_resource **resources, uint32_t **handles) { unsigned i; struct si_context *sctx = (struct si_context*)ctx; struct si_compute *program = sctx->cs_shader_state.program; assert(first + n <= MAX_GLOBAL_BUFFERS); if (!resources) { for (i = 0; i < n; i++) { pipe_resource_reference(&program->global_buffers[first + i], NULL); } return; } for (i = 0; i < n; i++) { uint64_t va; uint32_t offset; pipe_resource_reference(&program->global_buffers[first + i], resources[i]); va = si_resource(resources[i])->gpu_address; offset = util_le32_to_cpu(*handles[i]); va += offset; va = util_cpu_to_le64(va); memcpy(handles[i], &va, sizeof(va)); } } static void si_initialize_compute(struct si_context *sctx) { struct radeon_cmdbuf *cs = sctx->gfx_cs; uint64_t bc_va; radeon_set_sh_reg_seq(cs, R_00B858_COMPUTE_STATIC_THREAD_MGMT_SE0, 2); /* R_00B858_COMPUTE_STATIC_THREAD_MGMT_SE0 / SE1 */ radeon_emit(cs, S_00B858_SH0_CU_EN(0xffff) | S_00B858_SH1_CU_EN(0xffff)); radeon_emit(cs, S_00B85C_SH0_CU_EN(0xffff) | S_00B85C_SH1_CU_EN(0xffff)); if (sctx->chip_class >= CIK) { /* Also set R_00B858_COMPUTE_STATIC_THREAD_MGMT_SE2 / SE3 */ radeon_set_sh_reg_seq(cs, R_00B864_COMPUTE_STATIC_THREAD_MGMT_SE2, 2); radeon_emit(cs, S_00B864_SH0_CU_EN(0xffff) | S_00B864_SH1_CU_EN(0xffff)); radeon_emit(cs, S_00B868_SH0_CU_EN(0xffff) | S_00B868_SH1_CU_EN(0xffff)); } /* This register has been moved to R_00CD20_COMPUTE_MAX_WAVE_ID * and is now per pipe, so it should be handled in the * kernel if we want to use something other than the default value, * which is now 0x22f. */ if (sctx->chip_class <= SI) { /* XXX: This should be: * (number of compute units) * 4 * (waves per simd) - 1 */ radeon_set_sh_reg(cs, R_00B82C_COMPUTE_MAX_WAVE_ID, 0x190 /* Default value */); } /* Set the pointer to border colors. */ bc_va = sctx->border_color_buffer->gpu_address; if (sctx->chip_class >= CIK) { radeon_set_uconfig_reg_seq(cs, R_030E00_TA_CS_BC_BASE_ADDR, 2); radeon_emit(cs, bc_va >> 8); /* R_030E00_TA_CS_BC_BASE_ADDR */ radeon_emit(cs, S_030E04_ADDRESS(bc_va >> 40)); /* R_030E04_TA_CS_BC_BASE_ADDR_HI */ } else { if (sctx->screen->info.si_TA_CS_BC_BASE_ADDR_allowed) { radeon_set_config_reg(cs, R_00950C_TA_CS_BC_BASE_ADDR, bc_va >> 8); } } sctx->cs_shader_state.emitted_program = NULL; sctx->cs_shader_state.initialized = true; } static bool si_setup_compute_scratch_buffer(struct si_context *sctx, struct si_shader *shader, struct si_shader_config *config) { uint64_t scratch_bo_size, scratch_needed; scratch_bo_size = 0; scratch_needed = config->scratch_bytes_per_wave * sctx->scratch_waves; if (sctx->compute_scratch_buffer) scratch_bo_size = sctx->compute_scratch_buffer->b.b.width0; if (scratch_bo_size < scratch_needed) { si_resource_reference(&sctx->compute_scratch_buffer, NULL); sctx->compute_scratch_buffer = si_aligned_buffer_create(&sctx->screen->b, SI_RESOURCE_FLAG_UNMAPPABLE, PIPE_USAGE_DEFAULT, scratch_needed, 256); if (!sctx->compute_scratch_buffer) return false; } if (sctx->compute_scratch_buffer != shader->scratch_bo && scratch_needed) { uint64_t scratch_va = sctx->compute_scratch_buffer->gpu_address; si_shader_apply_scratch_relocs(shader, scratch_va); if (si_shader_binary_upload(sctx->screen, shader)) return false; si_resource_reference(&shader->scratch_bo, sctx->compute_scratch_buffer); } return true; } static bool si_switch_compute_shader(struct si_context *sctx, struct si_compute *program, struct si_shader *shader, const amd_kernel_code_t *code_object, unsigned offset) { struct radeon_cmdbuf *cs = sctx->gfx_cs; struct si_shader_config inline_config = {0}; struct si_shader_config *config; uint64_t shader_va; if (sctx->cs_shader_state.emitted_program == program && sctx->cs_shader_state.offset == offset) return true; if (program->ir_type != PIPE_SHADER_IR_NATIVE) { config = &shader->config; } else { unsigned lds_blocks; config = &inline_config; if (code_object) { code_object_to_config(code_object, config); } else { si_shader_binary_read_config(&shader->binary, config, offset); } lds_blocks = config->lds_size; /* XXX: We are over allocating LDS. For SI, the shader reports * LDS in blocks of 256 bytes, so if there are 4 bytes lds * allocated in the shader and 4 bytes allocated by the state * tracker, then we will set LDS_SIZE to 512 bytes rather than 256. */ if (sctx->chip_class <= SI) { lds_blocks += align(program->local_size, 256) >> 8; } else { lds_blocks += align(program->local_size, 512) >> 9; } /* TODO: use si_multiwave_lds_size_workaround */ assert(lds_blocks <= 0xFF); config->rsrc2 &= C_00B84C_LDS_SIZE; config->rsrc2 |= S_00B84C_LDS_SIZE(lds_blocks); } if (!si_setup_compute_scratch_buffer(sctx, shader, config)) return false; if (shader->scratch_bo) { COMPUTE_DBG(sctx->screen, "Waves: %u; Scratch per wave: %u bytes; " "Total Scratch: %u bytes\n", sctx->scratch_waves, config->scratch_bytes_per_wave, config->scratch_bytes_per_wave * sctx->scratch_waves); radeon_add_to_buffer_list(sctx, sctx->gfx_cs, shader->scratch_bo, RADEON_USAGE_READWRITE, RADEON_PRIO_SCRATCH_BUFFER); } /* Prefetch the compute shader to TC L2. * * We should also prefetch graphics shaders if a compute dispatch was * the last command, and the compute shader if a draw call was the last * command. However, that would add more complexity and we're likely * to get a shader state change in that case anyway. */ if (sctx->chip_class >= CIK) { cik_prefetch_TC_L2_async(sctx, &program->shader.bo->b.b, 0, program->shader.bo->b.b.width0); } shader_va = shader->bo->gpu_address + offset; if (program->use_code_object_v2) { /* Shader code is placed after the amd_kernel_code_t * struct. */ shader_va += sizeof(amd_kernel_code_t); } radeon_add_to_buffer_list(sctx, sctx->gfx_cs, shader->bo, RADEON_USAGE_READ, RADEON_PRIO_SHADER_BINARY); radeon_set_sh_reg_seq(cs, R_00B830_COMPUTE_PGM_LO, 2); radeon_emit(cs, shader_va >> 8); radeon_emit(cs, S_00B834_DATA(shader_va >> 40)); radeon_set_sh_reg_seq(cs, R_00B848_COMPUTE_PGM_RSRC1, 2); radeon_emit(cs, config->rsrc1); radeon_emit(cs, config->rsrc2); COMPUTE_DBG(sctx->screen, "COMPUTE_PGM_RSRC1: 0x%08x " "COMPUTE_PGM_RSRC2: 0x%08x\n", config->rsrc1, config->rsrc2); sctx->max_seen_compute_scratch_bytes_per_wave = MAX2(sctx->max_seen_compute_scratch_bytes_per_wave, config->scratch_bytes_per_wave); radeon_set_sh_reg(cs, R_00B860_COMPUTE_TMPRING_SIZE, S_00B860_WAVES(sctx->scratch_waves) | S_00B860_WAVESIZE(sctx->max_seen_compute_scratch_bytes_per_wave >> 10)); sctx->cs_shader_state.emitted_program = program; sctx->cs_shader_state.offset = offset; sctx->cs_shader_state.uses_scratch = config->scratch_bytes_per_wave != 0; return true; } static void setup_scratch_rsrc_user_sgprs(struct si_context *sctx, const amd_kernel_code_t *code_object, unsigned user_sgpr) { struct radeon_cmdbuf *cs = sctx->gfx_cs; uint64_t scratch_va = sctx->compute_scratch_buffer->gpu_address; unsigned max_private_element_size = AMD_HSA_BITS_GET( code_object->code_properties, AMD_CODE_PROPERTY_PRIVATE_ELEMENT_SIZE); uint32_t scratch_dword0 = scratch_va & 0xffffffff; uint32_t scratch_dword1 = S_008F04_BASE_ADDRESS_HI(scratch_va >> 32) | S_008F04_SWIZZLE_ENABLE(1); /* Disable address clamping */ uint32_t scratch_dword2 = 0xffffffff; uint32_t scratch_dword3 = S_008F0C_INDEX_STRIDE(3) | S_008F0C_ADD_TID_ENABLE(1); if (sctx->chip_class >= GFX9) { assert(max_private_element_size == 1); /* always 4 bytes on GFX9 */ } else { scratch_dword3 |= S_008F0C_ELEMENT_SIZE(max_private_element_size); if (sctx->chip_class < VI) { /* BUF_DATA_FORMAT is ignored, but it cannot be * BUF_DATA_FORMAT_INVALID. */ scratch_dword3 |= S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_8); } } radeon_set_sh_reg_seq(cs, R_00B900_COMPUTE_USER_DATA_0 + (user_sgpr * 4), 4); radeon_emit(cs, scratch_dword0); radeon_emit(cs, scratch_dword1); radeon_emit(cs, scratch_dword2); radeon_emit(cs, scratch_dword3); } static void si_setup_user_sgprs_co_v2(struct si_context *sctx, const amd_kernel_code_t *code_object, const struct pipe_grid_info *info, uint64_t kernel_args_va) { struct si_compute *program = sctx->cs_shader_state.program; struct radeon_cmdbuf *cs = sctx->gfx_cs; static const enum amd_code_property_mask_t workgroup_count_masks [] = { AMD_CODE_PROPERTY_ENABLE_SGPR_GRID_WORKGROUP_COUNT_X, AMD_CODE_PROPERTY_ENABLE_SGPR_GRID_WORKGROUP_COUNT_Y, AMD_CODE_PROPERTY_ENABLE_SGPR_GRID_WORKGROUP_COUNT_Z }; unsigned i, user_sgpr = 0; if (AMD_HSA_BITS_GET(code_object->code_properties, AMD_CODE_PROPERTY_ENABLE_SGPR_PRIVATE_SEGMENT_BUFFER)) { if (code_object->workitem_private_segment_byte_size > 0) { setup_scratch_rsrc_user_sgprs(sctx, code_object, user_sgpr); } user_sgpr += 4; } if (AMD_HSA_BITS_GET(code_object->code_properties, AMD_CODE_PROPERTY_ENABLE_SGPR_DISPATCH_PTR)) { struct dispatch_packet dispatch; unsigned dispatch_offset; struct si_resource *dispatch_buf = NULL; uint64_t dispatch_va; /* Upload dispatch ptr */ memset(&dispatch, 0, sizeof(dispatch)); dispatch.workgroup_size_x = util_cpu_to_le16(info->block[0]); dispatch.workgroup_size_y = util_cpu_to_le16(info->block[1]); dispatch.workgroup_size_z = util_cpu_to_le16(info->block[2]); dispatch.grid_size_x = util_cpu_to_le32(info->grid[0] * info->block[0]); dispatch.grid_size_y = util_cpu_to_le32(info->grid[1] * info->block[1]); dispatch.grid_size_z = util_cpu_to_le32(info->grid[2] * info->block[2]); dispatch.private_segment_size = util_cpu_to_le32(program->private_size); dispatch.group_segment_size = util_cpu_to_le32(program->local_size); dispatch.kernarg_address = util_cpu_to_le64(kernel_args_va); u_upload_data(sctx->b.const_uploader, 0, sizeof(dispatch), 256, &dispatch, &dispatch_offset, (struct pipe_resource**)&dispatch_buf); if (!dispatch_buf) { fprintf(stderr, "Error: Failed to allocate dispatch " "packet."); } radeon_add_to_buffer_list(sctx, sctx->gfx_cs, dispatch_buf, RADEON_USAGE_READ, RADEON_PRIO_CONST_BUFFER); dispatch_va = dispatch_buf->gpu_address + dispatch_offset; radeon_set_sh_reg_seq(cs, R_00B900_COMPUTE_USER_DATA_0 + (user_sgpr * 4), 2); radeon_emit(cs, dispatch_va); radeon_emit(cs, S_008F04_BASE_ADDRESS_HI(dispatch_va >> 32) | S_008F04_STRIDE(0)); si_resource_reference(&dispatch_buf, NULL); user_sgpr += 2; } if (AMD_HSA_BITS_GET(code_object->code_properties, AMD_CODE_PROPERTY_ENABLE_SGPR_KERNARG_SEGMENT_PTR)) { radeon_set_sh_reg_seq(cs, R_00B900_COMPUTE_USER_DATA_0 + (user_sgpr * 4), 2); radeon_emit(cs, kernel_args_va); radeon_emit(cs, S_008F04_BASE_ADDRESS_HI (kernel_args_va >> 32) | S_008F04_STRIDE(0)); user_sgpr += 2; } for (i = 0; i < 3 && user_sgpr < 16; i++) { if (code_object->code_properties & workgroup_count_masks[i]) { radeon_set_sh_reg_seq(cs, R_00B900_COMPUTE_USER_DATA_0 + (user_sgpr * 4), 1); radeon_emit(cs, info->grid[i]); user_sgpr += 1; } } } static bool si_upload_compute_input(struct si_context *sctx, const amd_kernel_code_t *code_object, const struct pipe_grid_info *info) { struct radeon_cmdbuf *cs = sctx->gfx_cs; struct si_compute *program = sctx->cs_shader_state.program; struct si_resource *input_buffer = NULL; unsigned kernel_args_size; unsigned num_work_size_bytes = program->use_code_object_v2 ? 0 : 36; uint32_t kernel_args_offset = 0; uint32_t *kernel_args; void *kernel_args_ptr; uint64_t kernel_args_va; unsigned i; /* The extra num_work_size_bytes are for work group / work item size information */ kernel_args_size = program->input_size + num_work_size_bytes; u_upload_alloc(sctx->b.const_uploader, 0, kernel_args_size, sctx->screen->info.tcc_cache_line_size, &kernel_args_offset, (struct pipe_resource**)&input_buffer, &kernel_args_ptr); if (unlikely(!kernel_args_ptr)) return false; kernel_args = (uint32_t*)kernel_args_ptr; kernel_args_va = input_buffer->gpu_address + kernel_args_offset; if (!code_object) { for (i = 0; i < 3; i++) { kernel_args[i] = util_cpu_to_le32(info->grid[i]); kernel_args[i + 3] = util_cpu_to_le32(info->grid[i] * info->block[i]); kernel_args[i + 6] = util_cpu_to_le32(info->block[i]); } } memcpy(kernel_args + (num_work_size_bytes / 4), info->input, program->input_size); for (i = 0; i < (kernel_args_size / 4); i++) { COMPUTE_DBG(sctx->screen, "input %u : %u\n", i, kernel_args[i]); } radeon_add_to_buffer_list(sctx, sctx->gfx_cs, input_buffer, RADEON_USAGE_READ, RADEON_PRIO_CONST_BUFFER); if (code_object) { si_setup_user_sgprs_co_v2(sctx, code_object, info, kernel_args_va); } else { radeon_set_sh_reg_seq(cs, R_00B900_COMPUTE_USER_DATA_0, 2); radeon_emit(cs, kernel_args_va); radeon_emit(cs, S_008F04_BASE_ADDRESS_HI (kernel_args_va >> 32) | S_008F04_STRIDE(0)); } si_resource_reference(&input_buffer, NULL); return true; } static void si_setup_tgsi_user_data(struct si_context *sctx, const struct pipe_grid_info *info) { struct si_compute *program = sctx->cs_shader_state.program; struct radeon_cmdbuf *cs = sctx->gfx_cs; unsigned grid_size_reg = R_00B900_COMPUTE_USER_DATA_0 + 4 * SI_NUM_RESOURCE_SGPRS; unsigned block_size_reg = grid_size_reg + /* 12 bytes = 3 dwords. */ 12 * program->uses_grid_size; unsigned cs_user_data_reg = block_size_reg + 12 * program->reads_variable_block_size; if (info->indirect) { if (program->uses_grid_size) { for (unsigned i = 0; i < 3; ++i) { si_cp_copy_data(sctx, COPY_DATA_REG, NULL, (grid_size_reg >> 2) + i, COPY_DATA_SRC_MEM, si_resource(info->indirect), info->indirect_offset + 4 * i); } } } else { if (program->uses_grid_size) { radeon_set_sh_reg_seq(cs, grid_size_reg, 3); radeon_emit(cs, info->grid[0]); radeon_emit(cs, info->grid[1]); radeon_emit(cs, info->grid[2]); } if (program->reads_variable_block_size) { radeon_set_sh_reg_seq(cs, block_size_reg, 3); radeon_emit(cs, info->block[0]); radeon_emit(cs, info->block[1]); radeon_emit(cs, info->block[2]); } } if (program->num_cs_user_data_dwords) { radeon_set_sh_reg_seq(cs, cs_user_data_reg, program->num_cs_user_data_dwords); radeon_emit_array(cs, sctx->cs_user_data, program->num_cs_user_data_dwords); } } static void si_emit_dispatch_packets(struct si_context *sctx, const struct pipe_grid_info *info) { struct si_screen *sscreen = sctx->screen; struct radeon_cmdbuf *cs = sctx->gfx_cs; bool render_cond_bit = sctx->render_cond && !sctx->render_cond_force_off; unsigned waves_per_threadgroup = DIV_ROUND_UP(info->block[0] * info->block[1] * info->block[2], 64); unsigned compute_resource_limits = S_00B854_SIMD_DEST_CNTL(waves_per_threadgroup % 4 == 0); if (sctx->chip_class >= CIK) { unsigned num_cu_per_se = sscreen->info.num_good_compute_units / sscreen->info.max_se; /* Force even distribution on all SIMDs in CU if the workgroup * size is 64. This has shown some good improvements if # of CUs * per SE is not a multiple of 4. */ if (num_cu_per_se % 4 && waves_per_threadgroup == 1) compute_resource_limits |= S_00B854_FORCE_SIMD_DIST(1); compute_resource_limits |= S_00B854_WAVES_PER_SH(sctx->cs_max_waves_per_sh); } else { /* SI */ if (sctx->cs_max_waves_per_sh) { unsigned limit_div16 = DIV_ROUND_UP(sctx->cs_max_waves_per_sh, 16); compute_resource_limits |= S_00B854_WAVES_PER_SH_SI(limit_div16); } } radeon_set_sh_reg(cs, R_00B854_COMPUTE_RESOURCE_LIMITS, compute_resource_limits); unsigned dispatch_initiator = S_00B800_COMPUTE_SHADER_EN(1) | S_00B800_FORCE_START_AT_000(1) | /* If the KMD allows it (there is a KMD hw register for it), * allow launching waves out-of-order. (same as Vulkan) */ S_00B800_ORDER_MODE(sctx->chip_class >= CIK); const uint *last_block = info->last_block; bool partial_block_en = last_block[0] || last_block[1] || last_block[2]; radeon_set_sh_reg_seq(cs, R_00B81C_COMPUTE_NUM_THREAD_X, 3); if (partial_block_en) { unsigned partial[3]; /* If no partial_block, these should be an entire block size, not 0. */ partial[0] = last_block[0] ? last_block[0] : info->block[0]; partial[1] = last_block[1] ? last_block[1] : info->block[1]; partial[2] = last_block[2] ? last_block[2] : info->block[2]; radeon_emit(cs, S_00B81C_NUM_THREAD_FULL(info->block[0]) | S_00B81C_NUM_THREAD_PARTIAL(partial[0])); radeon_emit(cs, S_00B820_NUM_THREAD_FULL(info->block[1]) | S_00B820_NUM_THREAD_PARTIAL(partial[1])); radeon_emit(cs, S_00B824_NUM_THREAD_FULL(info->block[2]) | S_00B824_NUM_THREAD_PARTIAL(partial[2])); dispatch_initiator |= S_00B800_PARTIAL_TG_EN(1); } else { radeon_emit(cs, S_00B81C_NUM_THREAD_FULL(info->block[0])); radeon_emit(cs, S_00B820_NUM_THREAD_FULL(info->block[1])); radeon_emit(cs, S_00B824_NUM_THREAD_FULL(info->block[2])); } if (info->indirect) { uint64_t base_va = si_resource(info->indirect)->gpu_address; radeon_add_to_buffer_list(sctx, sctx->gfx_cs, si_resource(info->indirect), RADEON_USAGE_READ, RADEON_PRIO_DRAW_INDIRECT); radeon_emit(cs, PKT3(PKT3_SET_BASE, 2, 0) | PKT3_SHADER_TYPE_S(1)); radeon_emit(cs, 1); radeon_emit(cs, base_va); radeon_emit(cs, base_va >> 32); radeon_emit(cs, PKT3(PKT3_DISPATCH_INDIRECT, 1, render_cond_bit) | PKT3_SHADER_TYPE_S(1)); radeon_emit(cs, info->indirect_offset); radeon_emit(cs, dispatch_initiator); } else { radeon_emit(cs, PKT3(PKT3_DISPATCH_DIRECT, 3, render_cond_bit) | PKT3_SHADER_TYPE_S(1)); radeon_emit(cs, info->grid[0]); radeon_emit(cs, info->grid[1]); radeon_emit(cs, info->grid[2]); radeon_emit(cs, dispatch_initiator); } } static void si_launch_grid( struct pipe_context *ctx, const struct pipe_grid_info *info) { struct si_context *sctx = (struct si_context*)ctx; struct si_compute *program = sctx->cs_shader_state.program; const amd_kernel_code_t *code_object = si_compute_get_code_object(program, info->pc); int i; /* HW bug workaround when CS threadgroups > 256 threads and async * compute isn't used, i.e. only one compute job can run at a time. * If async compute is possible, the threadgroup size must be limited * to 256 threads on all queues to avoid the bug. * Only SI and certain CIK chips are affected. */ bool cs_regalloc_hang = (sctx->chip_class == SI || sctx->family == CHIP_BONAIRE || sctx->family == CHIP_KABINI) && info->block[0] * info->block[1] * info->block[2] > 256; if (cs_regalloc_hang) sctx->flags |= SI_CONTEXT_PS_PARTIAL_FLUSH | SI_CONTEXT_CS_PARTIAL_FLUSH; if (program->ir_type != PIPE_SHADER_IR_NATIVE && program->shader.compilation_failed) return; if (sctx->has_graphics) { if (sctx->last_num_draw_calls != sctx->num_draw_calls) { si_update_fb_dirtiness_after_rendering(sctx); sctx->last_num_draw_calls = sctx->num_draw_calls; } si_decompress_textures(sctx, 1 << PIPE_SHADER_COMPUTE); } /* Add buffer sizes for memory checking in need_cs_space. */ si_context_add_resource_size(sctx, &program->shader.bo->b.b); /* TODO: add the scratch buffer */ if (info->indirect) { si_context_add_resource_size(sctx, info->indirect); /* Indirect buffers use TC L2 on GFX9, but not older hw. */ if (sctx->chip_class <= VI && si_resource(info->indirect)->TC_L2_dirty) { sctx->flags |= SI_CONTEXT_WRITEBACK_GLOBAL_L2; si_resource(info->indirect)->TC_L2_dirty = false; } } si_need_gfx_cs_space(sctx); if (sctx->bo_list_add_all_compute_resources) si_compute_resources_add_all_to_bo_list(sctx); if (!sctx->cs_shader_state.initialized) si_initialize_compute(sctx); if (sctx->flags) si_emit_cache_flush(sctx); if (!si_switch_compute_shader(sctx, program, &program->shader, code_object, info->pc)) return; si_upload_compute_shader_descriptors(sctx); si_emit_compute_shader_pointers(sctx); if (sctx->has_graphics && si_is_atom_dirty(sctx, &sctx->atoms.s.render_cond)) { sctx->atoms.s.render_cond.emit(sctx); si_set_atom_dirty(sctx, &sctx->atoms.s.render_cond, false); } if ((program->input_size || program->ir_type == PIPE_SHADER_IR_NATIVE) && unlikely(!si_upload_compute_input(sctx, code_object, info))) { return; } /* Global buffers */ for (i = 0; i < MAX_GLOBAL_BUFFERS; i++) { struct si_resource *buffer = si_resource(program->global_buffers[i]); if (!buffer) { continue; } radeon_add_to_buffer_list(sctx, sctx->gfx_cs, buffer, RADEON_USAGE_READWRITE, RADEON_PRIO_COMPUTE_GLOBAL); } if (program->ir_type != PIPE_SHADER_IR_NATIVE) si_setup_tgsi_user_data(sctx, info); si_emit_dispatch_packets(sctx, info); if (unlikely(sctx->current_saved_cs)) { si_trace_emit(sctx); si_log_compute_state(sctx, sctx->log); } sctx->compute_is_busy = true; sctx->num_compute_calls++; if (sctx->cs_shader_state.uses_scratch) sctx->num_spill_compute_calls++; if (cs_regalloc_hang) sctx->flags |= SI_CONTEXT_CS_PARTIAL_FLUSH; } void si_destroy_compute(struct si_compute *program) { if (program->ir_type != PIPE_SHADER_IR_NATIVE) { util_queue_drop_job(&program->screen->shader_compiler_queue, &program->ready); util_queue_fence_destroy(&program->ready); } si_shader_destroy(&program->shader); FREE(program); } static void si_delete_compute_state(struct pipe_context *ctx, void* state){ struct si_compute *program = (struct si_compute *)state; struct si_context *sctx = (struct si_context*)ctx; if (!state) return; if (program == sctx->cs_shader_state.program) sctx->cs_shader_state.program = NULL; if (program == sctx->cs_shader_state.emitted_program) sctx->cs_shader_state.emitted_program = NULL; si_compute_reference(&program, NULL); } static void si_set_compute_resources(struct pipe_context * ctx_, unsigned start, unsigned count, struct pipe_surface ** surfaces) { } void si_init_compute_functions(struct si_context *sctx) { sctx->b.create_compute_state = si_create_compute_state; sctx->b.delete_compute_state = si_delete_compute_state; sctx->b.bind_compute_state = si_bind_compute_state; sctx->b.set_compute_resources = si_set_compute_resources; sctx->b.set_global_binding = si_set_global_binding; sctx->b.launch_grid = si_launch_grid; }