diff options
Diffstat (limited to 'servers/rendering/rendering_device.cpp')
| -rw-r--r-- | servers/rendering/rendering_device.cpp | 729 |
1 files changed, 480 insertions, 249 deletions
diff --git a/servers/rendering/rendering_device.cpp b/servers/rendering/rendering_device.cpp index 5d65118159..3290738b40 100644 --- a/servers/rendering/rendering_device.cpp +++ b/servers/rendering/rendering_device.cpp @@ -35,7 +35,71 @@ #include "core/config/project_settings.h" #include "core/io/dir_access.h" -#include "servers/rendering/renderer_rd/api_context_rd.h" + +#define FORCE_SEPARATE_PRESENT_QUEUE 0 + +/**************************/ +/**** HELPER FUNCTIONS ****/ +/**************************/ + +static String _get_device_vendor_name(const RenderingContextDriver::Device &p_device) { + switch (p_device.vendor) { + case RenderingContextDriver::VENDOR_AMD: + return "AMD"; + case RenderingContextDriver::VENDOR_IMGTEC: + return "ImgTec"; + case RenderingContextDriver::VENDOR_APPLE: + return "Apple"; + case RenderingContextDriver::VENDOR_NVIDIA: + return "NVIDIA"; + case RenderingContextDriver::VENDOR_ARM: + return "ARM"; + case RenderingContextDriver::VENDOR_MICROSOFT: + return "Microsoft"; + case RenderingContextDriver::VENDOR_QUALCOMM: + return "Qualcomm"; + case RenderingContextDriver::VENDOR_INTEL: + return "Intel"; + default: + return "Unknown"; + } +} + +static String _get_device_type_name(const RenderingContextDriver::Device &p_device) { + switch (p_device.type) { + case RenderingContextDriver::DEVICE_TYPE_INTEGRATED_GPU: + return "Integrated"; + case RenderingContextDriver::DEVICE_TYPE_DISCRETE_GPU: + return "Discrete"; + case RenderingContextDriver::DEVICE_TYPE_VIRTUAL_GPU: + return "Virtual"; + case RenderingContextDriver::DEVICE_TYPE_CPU: + return "CPU"; + case RenderingContextDriver::DEVICE_TYPE_OTHER: + default: + return "Other"; + } +} + +static uint32_t _get_device_type_score(const RenderingContextDriver::Device &p_device) { + switch (p_device.type) { + case RenderingContextDriver::DEVICE_TYPE_INTEGRATED_GPU: + return 4; + case RenderingContextDriver::DEVICE_TYPE_DISCRETE_GPU: + return 5; + case RenderingContextDriver::DEVICE_TYPE_VIRTUAL_GPU: + return 3; + case RenderingContextDriver::DEVICE_TYPE_CPU: + return 2; + case RenderingContextDriver::DEVICE_TYPE_OTHER: + default: + return 1; + } +} + +/**************************/ +/**** RENDERING DEVICE ****/ +/**************************/ // When true, the command graph will attempt to reorder the rendering commands submitted by the user based on the dependencies detected from // the commands automatically. This should improve rendering performance in most scenarios at the cost of some extra CPU overhead. @@ -240,7 +304,7 @@ Error RenderingDevice::_staging_buffer_allocate(uint32_t p_amount, uint32_t p_re // and this frame is not even done. // If this is the main thread, it means the user is likely loading a lot of resources at once,. // Otherwise, the thread should just be blocked until the next frame (currently unimplemented). - r_required_action = STAGING_REQUIRED_ACTION_FLUSH_CURRENT; + r_required_action = STAGING_REQUIRED_ACTION_FLUSH_AND_STALL_ALL; } } else { @@ -249,7 +313,7 @@ Error RenderingDevice::_staging_buffer_allocate(uint32_t p_amount, uint32_t p_re } } - } else if (staging_buffer_blocks[staging_buffer_current].frame_used <= frames_drawn - frame_count) { + } else if (staging_buffer_blocks[staging_buffer_current].frame_used <= frames_drawn - frames.size()) { // This is an old block, which was already processed, let's reuse. staging_buffer_blocks.write[staging_buffer_current].frame_used = frames_drawn; staging_buffer_blocks.write[staging_buffer_current].fill_amount = 0; @@ -268,7 +332,7 @@ Error RenderingDevice::_staging_buffer_allocate(uint32_t p_amount, uint32_t p_re // Let's flush older frames. // The logic here is that if a game is loading a lot of data from the main thread, it will need to be stalled anyway. // If loading from a separate thread, we can block that thread until next frame when more room is made (not currently implemented, though). - r_required_action = STAGING_REQUIRED_ACTION_FLUSH_OLDER; + r_required_action = STAGING_REQUIRED_ACTION_STALL_PREVIOUS; } } @@ -286,9 +350,8 @@ void RenderingDevice::_staging_buffer_execute_required_action(StagingRequiredAct case STAGING_REQUIRED_ACTION_NONE: { // Do nothing. } break; - case STAGING_REQUIRED_ACTION_FLUSH_CURRENT: { - // Flush EVERYTHING including setup commands. IF not immediate, also need to flush the draw commands. - _flush(true); + case STAGING_REQUIRED_ACTION_FLUSH_AND_STALL_ALL: { + _flush_and_stall_for_all_frames(); // Clear the whole staging buffer. for (int i = 0; i < staging_buffer_blocks.size(); i++) { @@ -299,8 +362,8 @@ void RenderingDevice::_staging_buffer_execute_required_action(StagingRequiredAct // Claim for current frame. staging_buffer_blocks.write[staging_buffer_current].frame_used = frames_drawn; } break; - case STAGING_REQUIRED_ACTION_FLUSH_OLDER: { - _flush(false); + case STAGING_REQUIRED_ACTION_STALL_PREVIOUS: { + _stall_for_previous_frames(); for (int i = 0; i < staging_buffer_blocks.size(); i++) { // Clear all blocks but the ones from this frame. @@ -340,7 +403,7 @@ Error RenderingDevice::_buffer_update(Buffer *p_buffer, RID p_buffer_id, size_t return err; } - if (p_use_draw_queue && !command_buffer_copies_vector.is_empty() && required_action == STAGING_REQUIRED_ACTION_FLUSH_CURRENT) { + if (p_use_draw_queue && !command_buffer_copies_vector.is_empty() && required_action == STAGING_REQUIRED_ACTION_FLUSH_AND_STALL_ALL) { if (_buffer_make_mutable(p_buffer, p_buffer_id)) { // The buffer must be mutable to be used as a copy destination. draw_graph.add_synchronization(); @@ -507,7 +570,7 @@ Vector<uint8_t> RenderingDevice::buffer_get_data(RID p_buffer, uint32_t p_offset draw_graph.add_buffer_get_data(buffer->driver_id, buffer->draw_tracker, tmp_buffer, region); // Flush everything so memory can be safely mapped. - _flush(true); + _flush_and_stall_for_all_frames(); uint8_t *buffer_mem = driver->buffer_map(tmp_buffer); ERR_FAIL_NULL_V(buffer_mem, Vector<uint8_t>()); @@ -1078,7 +1141,7 @@ Error RenderingDevice::_texture_update(RID p_texture, uint32_t p_layer, const Ve Error err = _staging_buffer_allocate(to_allocate, required_align, alloc_offset, alloc_size, required_action, false); ERR_FAIL_COND_V(err, ERR_CANT_CREATE); - if (!p_use_setup_queue && !command_buffer_to_texture_copies_vector.is_empty() && required_action == STAGING_REQUIRED_ACTION_FLUSH_CURRENT) { + if (!p_use_setup_queue && !command_buffer_to_texture_copies_vector.is_empty() && required_action == STAGING_REQUIRED_ACTION_FLUSH_AND_STALL_ALL) { if (_texture_make_mutable(texture, p_texture)) { // The texture must be mutable to be used as a copy destination. draw_graph.add_synchronization(); @@ -1314,7 +1377,8 @@ Vector<uint8_t> RenderingDevice::texture_get_data(RID p_texture, uint32_t p_laye draw_graph.add_texture_get_data(tex->driver_id, tex->draw_tracker, tmp_buffer, command_buffer_texture_copy_regions_vector); - _flush(true); + // Flush everything so memory can be safely mapped. + _flush_and_stall_for_all_frames(); const uint8_t *read_ptr = driver->buffer_map(tmp_buffer); ERR_FAIL_NULL_V(read_ptr, Vector<uint8_t>()); @@ -2976,7 +3040,7 @@ RID RenderingDevice::render_pipeline_create(RID p_shader, FramebufferFormatID p_ p_specialization_constants); ERR_FAIL_COND_V(!pipeline.driver_id, RID()); - if (pipelines_cache_enabled) { + if (pipeline_cache_enabled) { _update_pipeline_cache(); } @@ -3053,7 +3117,7 @@ RID RenderingDevice::compute_pipeline_create(RID p_shader, const Vector<Pipeline pipeline.driver_id = driver->compute_pipeline_create(shader->driver_id, p_specialization_constants); ERR_FAIL_COND_V(!pipeline.driver_id, RID()); - if (pipelines_cache_enabled) { + if (pipeline_cache_enabled) { _update_pipeline_cache(); } @@ -3084,23 +3148,95 @@ bool RenderingDevice::compute_pipeline_is_valid(RID p_pipeline) { /**** SCREEN ****/ /****************/ +uint32_t RenderingDevice::_get_swap_chain_desired_count() const { + return MAX(2U, uint32_t(GLOBAL_GET("rendering/rendering_device/vsync/swapchain_image_count"))); +} + +Error RenderingDevice::screen_create(DisplayServer::WindowID p_screen) { + _THREAD_SAFE_METHOD_ + + RenderingContextDriver::SurfaceID surface = context->surface_get_from_window(p_screen); + ERR_FAIL_COND_V_MSG(surface == 0, ERR_CANT_CREATE, "A surface was not created for the screen."); + + HashMap<DisplayServer::WindowID, RDD::SwapChainID>::ConstIterator it = screen_swap_chains.find(p_screen); + ERR_FAIL_COND_V_MSG(it != screen_swap_chains.end(), ERR_CANT_CREATE, "A swap chain was already created for the screen."); + + RDD::SwapChainID swap_chain = driver->swap_chain_create(surface); + ERR_FAIL_COND_V_MSG(swap_chain.id == 0, ERR_CANT_CREATE, "Unable to create swap chain."); + + Error err = driver->swap_chain_resize(main_queue, swap_chain, _get_swap_chain_desired_count()); + ERR_FAIL_COND_V_MSG(err != OK, ERR_CANT_CREATE, "Unable to resize the new swap chain."); + + screen_swap_chains[p_screen] = swap_chain; + + return OK; +} + +Error RenderingDevice::screen_prepare_for_drawing(DisplayServer::WindowID p_screen) { + _THREAD_SAFE_METHOD_ + + HashMap<DisplayServer::WindowID, RDD::SwapChainID>::ConstIterator it = screen_swap_chains.find(p_screen); + ERR_FAIL_COND_V_MSG(it == screen_swap_chains.end(), ERR_CANT_CREATE, "A swap chain was not created for the screen."); + + // Erase the framebuffer corresponding to this screen from the map in case any of the operations fail. + screen_framebuffers.erase(p_screen); + + // If this frame has already queued this swap chain for presentation, we present it and remove it from the pending list. + uint32_t to_present_index = 0; + while (to_present_index < frames[frame].swap_chains_to_present.size()) { + if (frames[frame].swap_chains_to_present[to_present_index] == it->value) { + driver->command_queue_present(present_queue, it->value, {}); + frames[frame].swap_chains_to_present.remove_at(to_present_index); + } else { + to_present_index++; + } + } + + bool resize_required = false; + RDD::FramebufferID framebuffer = driver->swap_chain_acquire_framebuffer(main_queue, it->value, resize_required); + if (resize_required) { + // Flush everything so nothing can be using the swap chain before resizing it. + _flush_and_stall_for_all_frames(); + + Error err = driver->swap_chain_resize(main_queue, it->value, _get_swap_chain_desired_count()); + if (err != OK) { + // Resize is allowed to fail silently because the window can be minimized. + return err; + } + + framebuffer = driver->swap_chain_acquire_framebuffer(main_queue, it->value, resize_required); + } + + ERR_FAIL_COND_V_MSG(framebuffer.id == 0, FAILED, "Unable to acquire framebuffer."); + + // Store the framebuffer that will be used next to draw to this screen. + screen_framebuffers[p_screen] = framebuffer; + frames[frame].swap_chains_to_present.push_back(it->value); + + return OK; +} + int RenderingDevice::screen_get_width(DisplayServer::WindowID p_screen) const { _THREAD_SAFE_METHOD_ - ERR_FAIL_COND_V_MSG(local_device.is_valid(), -1, "Local devices have no screen"); - return context->window_get_width(p_screen); + RenderingContextDriver::SurfaceID surface = context->surface_get_from_window(p_screen); + ERR_FAIL_COND_V_MSG(surface == 0, 0, "A surface was not created for the screen."); + return context->surface_get_width(surface); } int RenderingDevice::screen_get_height(DisplayServer::WindowID p_screen) const { _THREAD_SAFE_METHOD_ - ERR_FAIL_COND_V_MSG(local_device.is_valid(), -1, "Local devices have no screen"); - return context->window_get_height(p_screen); + RenderingContextDriver::SurfaceID surface = context->surface_get_from_window(p_screen); + ERR_FAIL_COND_V_MSG(surface == 0, 0, "A surface was not created for the screen."); + return context->surface_get_height(surface); } -RenderingDevice::FramebufferFormatID RenderingDevice::screen_get_framebuffer_format() const { +RenderingDevice::FramebufferFormatID RenderingDevice::screen_get_framebuffer_format(DisplayServer::WindowID p_screen) const { _THREAD_SAFE_METHOD_ - ERR_FAIL_COND_V_MSG(local_device.is_valid(), INVALID_ID, "Local devices have no screen"); - DataFormat format = driver->screen_get_format(); + HashMap<DisplayServer::WindowID, RDD::SwapChainID>::ConstIterator it = screen_swap_chains.find(p_screen); + ERR_FAIL_COND_V_MSG(it == screen_swap_chains.end(), FAILED, "Screen was never prepared."); + + DataFormat format = driver->swap_chain_get_format(it->value); ERR_FAIL_COND_V(format == DATA_FORMAT_MAX, INVALID_ID); AttachmentFormat attachment; @@ -3112,33 +3248,54 @@ RenderingDevice::FramebufferFormatID RenderingDevice::screen_get_framebuffer_for return const_cast<RenderingDevice *>(this)->framebuffer_format_create(screen_attachment); } +Error RenderingDevice::screen_free(DisplayServer::WindowID p_screen) { + _THREAD_SAFE_METHOD_ + + HashMap<DisplayServer::WindowID, RDD::SwapChainID>::ConstIterator it = screen_swap_chains.find(p_screen); + ERR_FAIL_COND_V_MSG(it == screen_swap_chains.end(), FAILED, "Screen was never created."); + + // Flush everything so nothing can be using the swap chain before erasing it. + _flush_and_stall_for_all_frames(); + + const DisplayServer::WindowID screen = it->key; + const RDD::SwapChainID swap_chain = it->value; + driver->swap_chain_free(swap_chain); + screen_framebuffers.erase(screen); + screen_swap_chains.erase(screen); + + return OK; +} + /*******************/ /**** DRAW LIST ****/ /*******************/ RenderingDevice::DrawListID RenderingDevice::draw_list_begin_for_screen(DisplayServer::WindowID p_screen, const Color &p_clear_color) { _THREAD_SAFE_METHOD_ - ERR_FAIL_COND_V_MSG(local_device.is_valid(), INVALID_ID, "Local devices have no screen"); ERR_FAIL_COND_V_MSG(draw_list != nullptr, INVALID_ID, "Only one draw list can be active at the same time."); ERR_FAIL_COND_V_MSG(compute_list != nullptr, INVALID_ID, "Only one draw/compute list can be active at the same time."); - if (!context->window_is_valid_swapchain(p_screen)) { - return INVALID_ID; - } + RenderingContextDriver::SurfaceID surface = context->surface_get_from_window(p_screen); + HashMap<DisplayServer::WindowID, RDD::SwapChainID>::ConstIterator sc_it = screen_swap_chains.find(p_screen); + HashMap<DisplayServer::WindowID, RDD::FramebufferID>::ConstIterator fb_it = screen_framebuffers.find(p_screen); + ERR_FAIL_COND_V_MSG(surface == 0, 0, "A surface was not created for the screen."); + ERR_FAIL_COND_V_MSG(sc_it == screen_swap_chains.end(), INVALID_ID, "Screen was never prepared."); + ERR_FAIL_COND_V_MSG(fb_it == screen_framebuffers.end(), INVALID_ID, "Framebuffer was never prepared."); - Rect2i viewport = Rect2i(0, 0, context->window_get_width(p_screen), context->window_get_height(p_screen)); + Rect2i viewport = Rect2i(0, 0, context->surface_get_width(surface), context->surface_get_height(surface)); _draw_list_allocate(viewport, 0); #ifdef DEBUG_ENABLED - draw_list_framebuffer_format = screen_get_framebuffer_format(); + draw_list_framebuffer_format = screen_get_framebuffer_format(p_screen); #endif draw_list_subpass_count = 1; RDD::RenderPassClearValue clear_value; clear_value.color = p_clear_color; - draw_graph.add_draw_list_begin(context->window_get_render_pass(p_screen), context->window_get_framebuffer(p_screen), viewport, clear_value, true, false); + RDD::RenderPassID render_pass = driver->swap_chain_get_render_pass(sc_it->value); + draw_graph.add_draw_list_begin(render_pass, fb_it->value, viewport, clear_value, true, false); _draw_list_set_viewport(viewport); _draw_list_set_scissor(viewport); @@ -3775,7 +3932,6 @@ Error RenderingDevice::_draw_list_allocate(const Rect2i &p_viewport, uint32_t p_ draw_list = memnew(DrawList); draw_list->viewport = p_viewport; - draw_list_count = 0; return OK; } @@ -4533,130 +4689,50 @@ void RenderingDevice::draw_command_end_label() { } String RenderingDevice::get_device_vendor_name() const { - return context->get_device_vendor_name(); + return _get_device_vendor_name(device); } String RenderingDevice::get_device_name() const { - return context->get_device_name(); + return device.name; } RenderingDevice::DeviceType RenderingDevice::get_device_type() const { - return context->get_device_type(); + return DeviceType(device.type); } -String RenderingDevice::get_device_api_version() const { - return context->get_device_api_version(); -} - -String RenderingDevice::get_device_pipeline_cache_uuid() const { - return context->get_device_pipeline_cache_uuid(); +String RenderingDevice::get_device_api_name() const { + return driver->get_api_name(); } -void RenderingDevice::_finalize_command_buffers(bool p_postpare) { - if (draw_list) { - ERR_PRINT("Found open draw list at the end of the frame, this should never happen (further drawing will likely not work)."); - } - - if (compute_list) { - ERR_PRINT("Found open compute list at the end of the frame, this should never happen (further compute will likely not work)."); - } - - { // Complete the setup buffer (that needs to be processed before anything else). - draw_graph.end(frames[frame].draw_command_buffer, RENDER_GRAPH_REORDER, RENDER_GRAPH_FULL_BARRIERS); - - if (p_postpare) { - context->postpare_buffers(frames[frame].draw_command_buffer); - } - - driver->end_segment(); - driver->command_buffer_end(frames[frame].setup_command_buffer); - driver->command_buffer_end(frames[frame].draw_command_buffer); - } +String RenderingDevice::get_device_api_version() const { + return driver->get_api_version(); } -void RenderingDevice::_begin_frame() { - draw_graph.begin(); - - // Erase pending resources. - _free_pending_resources(frame); - - // Create setup command buffer and set as the setup buffer. - - { - bool ok = driver->command_buffer_begin(frames[frame].setup_command_buffer); - ERR_FAIL_COND(!ok); - ok = driver->command_buffer_begin(frames[frame].draw_command_buffer); - ERR_FAIL_COND(!ok); - - if (local_device.is_null()) { - context->append_command_buffer(frames[frame].draw_command_buffer); - context->set_setup_buffer(frames[frame].setup_command_buffer); // Append now so it's added before everything else. - } - - driver->begin_segment(frames[frame].draw_command_buffer, frame, frames_drawn); - } - - // Advance current frame. - frames_drawn++; - // Advance staging buffer if used. - if (staging_buffer_used) { - staging_buffer_current = (staging_buffer_current + 1) % staging_buffer_blocks.size(); - staging_buffer_used = false; - } - - if (frames[frame].timestamp_count) { - driver->timestamp_query_pool_get_results(frames[frame].timestamp_pool, frames[frame].timestamp_count, frames[frame].timestamp_result_values.ptr()); - driver->command_timestamp_query_pool_reset(frames[frame].setup_command_buffer, frames[frame].timestamp_pool, frames[frame].timestamp_count); - SWAP(frames[frame].timestamp_names, frames[frame].timestamp_result_names); - SWAP(frames[frame].timestamp_cpu_values, frames[frame].timestamp_cpu_result_values); - } - - frames[frame].timestamp_result_count = frames[frame].timestamp_count; - frames[frame].timestamp_count = 0; - frames[frame].index = Engine::get_singleton()->get_frames_drawn(); +String RenderingDevice::get_device_pipeline_cache_uuid() const { + return driver->get_pipeline_cache_uuid(); } void RenderingDevice::swap_buffers() { - ERR_FAIL_COND_MSG(local_device.is_valid(), "Local devices can't swap buffers."); _THREAD_SAFE_METHOD_ - _finalize_command_buffers(true); - - // Swap buffers. - if (!screen_prepared) { - context->flush(true, true, false); - } else { - screen_prepared = false; - context->swap_buffers(); - } - - frame = (frame + 1) % frame_count; + _end_frame(); + _execute_frame(true); + _present_frame(); + // Advance to the next frame and begin recording again. + frame = (frame + 1) % frames.size(); _begin_frame(); } void RenderingDevice::submit() { _THREAD_SAFE_METHOD_ - - ERR_FAIL_COND_MSG(local_device.is_null(), "Only local devices can submit and sync."); - ERR_FAIL_COND_MSG(local_device_processing, "device already submitted, call sync to wait until done."); - - _finalize_command_buffers(false); - - RDD::CommandBufferID command_buffers[2] = { frames[frame].setup_command_buffer, frames[frame].draw_command_buffer }; - context->local_device_push_command_buffers(local_device, command_buffers, 2); - local_device_processing = true; + _end_frame(); + _execute_frame(false); } void RenderingDevice::sync() { _THREAD_SAFE_METHOD_ - - ERR_FAIL_COND_MSG(local_device.is_null(), "Only local devices can submit and sync."); - ERR_FAIL_COND_MSG(!local_device_processing, "sync can only be called after a submit"); - - context->local_device_sync(local_device); _begin_frame(); - local_device_processing = false; } void RenderingDevice::_free_pending_resources(int p_frame) { @@ -4741,14 +4817,8 @@ void RenderingDevice::_free_pending_resources(int p_frame) { } } -void RenderingDevice::prepare_screen_for_drawing() { - _THREAD_SAFE_METHOD_ - context->prepare_buffers(frames[frame].draw_command_buffer); - screen_prepared = true; -} - uint32_t RenderingDevice::get_frame_delay() const { - return frame_count; + return frames.size(); } uint64_t RenderingDevice::get_memory_usage(MemoryType p_type) const { @@ -4769,113 +4839,239 @@ uint64_t RenderingDevice::get_memory_usage(MemoryType p_type) const { } } -void RenderingDevice::_flush(bool p_current_frame) { - if (local_device.is_valid() && !p_current_frame) { - return; // Flushing previous frames has no effect with local device. +void RenderingDevice::_begin_frame() { + // Before beginning this frame, wait on the fence if it was signaled to make sure its work is finished. + if (frames[frame].draw_fence_signaled) { + driver->fence_wait(frames[frame].draw_fence); + frames[frame].draw_fence_signaled = false; + } + + // Begin recording on the frame's command buffers. + driver->begin_segment(frame, frames_drawn++); + driver->command_buffer_begin(frames[frame].setup_command_buffer); + driver->command_buffer_begin(frames[frame].draw_command_buffer); + + // Reset the graph. + draw_graph.begin(); + + // Erase pending resources. + _free_pending_resources(frame); + + // Advance staging buffer if used. + if (staging_buffer_used) { + staging_buffer_current = (staging_buffer_current + 1) % staging_buffer_blocks.size(); + staging_buffer_used = false; + } + + if (frames[frame].timestamp_count) { + driver->timestamp_query_pool_get_results(frames[frame].timestamp_pool, frames[frame].timestamp_count, frames[frame].timestamp_result_values.ptr()); + driver->command_timestamp_query_pool_reset(frames[frame].setup_command_buffer, frames[frame].timestamp_pool, frames[frame].timestamp_count); + SWAP(frames[frame].timestamp_names, frames[frame].timestamp_result_names); + SWAP(frames[frame].timestamp_cpu_values, frames[frame].timestamp_cpu_result_values); + } + + frames[frame].timestamp_result_count = frames[frame].timestamp_count; + frames[frame].timestamp_count = 0; + frames[frame].index = Engine::get_singleton()->get_frames_drawn(); +} + +void RenderingDevice::_end_frame() { + if (draw_list) { + ERR_PRINT("Found open draw list at the end of the frame, this should never happen (further drawing will likely not work)."); } - // Not doing this crashes RADV (undefined behavior). - if (p_current_frame) { - draw_graph.end(frames[frame].draw_command_buffer, RENDER_GRAPH_REORDER, RENDER_GRAPH_FULL_BARRIERS); - driver->end_segment(); - driver->command_buffer_end(frames[frame].setup_command_buffer); - driver->command_buffer_end(frames[frame].draw_command_buffer); - draw_graph.begin(); + if (compute_list) { + ERR_PRINT("Found open compute list at the end of the frame, this should never happen (further compute will likely not work)."); } - if (local_device.is_valid()) { - RDD::CommandBufferID command_buffers[2] = { frames[frame].setup_command_buffer, frames[frame].draw_command_buffer }; - context->local_device_push_command_buffers(local_device, command_buffers, 2); - context->local_device_sync(local_device); + draw_graph.end(frames[frame].draw_command_buffer, RENDER_GRAPH_REORDER, RENDER_GRAPH_FULL_BARRIERS); + driver->command_buffer_end(frames[frame].setup_command_buffer); + driver->command_buffer_end(frames[frame].draw_command_buffer); + driver->end_segment(); +} - bool ok = driver->command_buffer_begin(frames[frame].setup_command_buffer); - ERR_FAIL_COND(!ok); - ok = driver->command_buffer_begin(frames[frame].draw_command_buffer); - ERR_FAIL_COND(!ok); +void RenderingDevice::_execute_frame(bool p_signal_for_present) { + const bool frame_can_present = !frames[frame].swap_chains_to_present.is_empty(); + const VectorView<RDD::SemaphoreID> execute_draw_semaphore = p_signal_for_present && frame_can_present ? frames[frame].draw_semaphore : VectorView<RDD::SemaphoreID>(); + driver->command_queue_execute(main_queue, frames[frame].setup_command_buffer, {}, frames[frame].setup_semaphore, {}); + driver->command_queue_execute(main_queue, frames[frame].draw_command_buffer, frames[frame].setup_semaphore, execute_draw_semaphore, frames[frame].draw_fence); + frames[frame].draw_fence_signaled = true; +} - driver->begin_segment(frames[frame].draw_command_buffer, frame, frames_drawn); - } else { - context->flush(p_current_frame, p_current_frame); - // Re-create the setup command. - if (p_current_frame) { - bool ok = driver->command_buffer_begin(frames[frame].setup_command_buffer); - ERR_FAIL_COND(!ok); - - context->set_setup_buffer(frames[frame].setup_command_buffer); // Append now so it's added before everything else. - ok = driver->command_buffer_begin(frames[frame].draw_command_buffer); - ERR_FAIL_COND(!ok); - context->append_command_buffer(frames[frame].draw_command_buffer); - - driver->begin_segment(frames[frame].draw_command_buffer, frame, frames_drawn); +void RenderingDevice::_present_frame() { + if (!frames[frame].swap_chains_to_present.is_empty()) { + driver->command_queue_present(present_queue, frames[frame].swap_chains_to_present, frames[frame].draw_semaphore); + frames[frame].swap_chains_to_present.clear(); + } +} + +void RenderingDevice::_stall_for_previous_frames() { + for (uint32_t i = 0; i < frames.size(); i++) { + if (frames[i].draw_fence_signaled) { + driver->fence_wait(frames[i].draw_fence); + frames[i].draw_fence_signaled = false; } } } -void RenderingDevice::initialize(ApiContextRD *p_context, bool p_local_device) { +void RenderingDevice::_flush_and_stall_for_all_frames() { + _stall_for_previous_frames(); + _end_frame(); + _execute_frame(false); + _begin_frame(); +} + +Error RenderingDevice::initialize(RenderingContextDriver *p_context, DisplayServer::WindowID p_main_window) { + Error err; + + RenderingContextDriver::SurfaceID main_surface = 0; + const bool main_instance = (singleton == this) && (p_main_window != DisplayServer::INVALID_WINDOW_ID); + if (p_main_window != DisplayServer::INVALID_WINDOW_ID) { + // Retrieve the surface from the main window if it was specified. + main_surface = p_context->surface_get_from_window(p_main_window); + ERR_FAIL_COND_V(main_surface == 0, FAILED); + } + context = p_context; + driver = context->driver_create(); + + print_verbose("Devices:"); + int32_t device_index = Engine::get_singleton()->get_gpu_index(); + const uint32_t device_count = context->device_get_count(); + const bool detect_device = (device_index < 0) || (device_index >= int32_t(device_count)); + uint32_t device_type_score = 0; + for (uint32_t i = 0; i < device_count; i++) { + RenderingContextDriver::Device device_option = context->device_get(i); + String name = device_option.name; + String vendor = _get_device_vendor_name(device_option); + String type = _get_device_type_name(device_option); + bool present_supported = main_surface != 0 ? context->device_supports_present(i, main_surface) : false; + print_verbose(" #" + itos(i) + ": " + vendor + " " + name + " - " + (present_supported ? "Supported" : "Unsupported") + ", " + type); + if (detect_device && (present_supported || main_surface == 0)) { + // If a window was specified, present must be supported by the device to be available as an option. + // Assign a score for each type of device and prefer the device with the higher score. + uint32_t option_score = _get_device_type_score(device_option); + if (option_score > device_type_score) { + device_index = i; + device_type_score = option_score; + } + } + } - device_capabilities = p_context->get_device_capabilities(); + ERR_FAIL_COND_V_MSG((device_index < 0) || (device_index >= int32_t(device_count)), ERR_CANT_CREATE, "None of the devices supports both graphics and present queues."); - if (p_local_device) { - frame_count = 1; - local_device = context->local_device_create(); - } else { - frame_count = context->get_swapchain_image_count() + 1; // Always need one extra to ensure it's unused at any time, without having to use a fence for this. + uint32_t frame_count = 1; + if (main_surface != 0) { + frame_count = MAX(2U, uint32_t(GLOBAL_GET("rendering/rendering_device/vsync/frame_queue_size"))); } - driver = context->get_driver(local_device); - max_timestamp_query_elements = 256; - frames.resize(frame_count); frame = 0; - // Create setup and frame buffers. - for (int i = 0; i < frame_count; i++) { - frames[i].index = 0; - - // Create command pool, one per frame is recommended. - frames[i].command_pool = driver->command_pool_create(RDD::COMMAND_BUFFER_TYPE_PRIMARY); - ERR_FAIL_COND(!frames[i].command_pool); + frames.resize(frame_count); + max_timestamp_query_elements = 256; - // Create command buffers. - frames[i].setup_command_buffer = driver->command_buffer_create(RDD::COMMAND_BUFFER_TYPE_PRIMARY, frames[i].command_pool); - ERR_CONTINUE(!frames[i].setup_command_buffer); - frames[i].draw_command_buffer = driver->command_buffer_create(RDD::COMMAND_BUFFER_TYPE_PRIMARY, frames[i].command_pool); - ERR_CONTINUE(!frames[i].draw_command_buffer); + device = context->device_get(device_index); + err = driver->initialize(device_index, frame_count); + ERR_FAIL_COND_V_MSG(err != OK, FAILED, "Failed to initialize driver for device."); - { - // Create query pool. - frames[i].timestamp_pool = driver->timestamp_query_pool_create(max_timestamp_query_elements); - frames[i].timestamp_names.resize(max_timestamp_query_elements); - frames[i].timestamp_cpu_values.resize(max_timestamp_query_elements); - frames[i].timestamp_count = 0; - frames[i].timestamp_result_names.resize(max_timestamp_query_elements); - frames[i].timestamp_cpu_result_values.resize(max_timestamp_query_elements); - frames[i].timestamp_result_values.resize(max_timestamp_query_elements); - frames[i].timestamp_result_count = 0; + if (main_instance) { + // Only the singleton instance with a display should print this information. + String rendering_method; + if (OS::get_singleton()->get_current_rendering_method() == "mobile") { + rendering_method = "Forward Mobile"; + } else { + rendering_method = "Forward+"; } + + // Output our device version. + print_line(vformat("%s %s - %s - Using Device #%d: %s - %s", get_device_api_name(), get_device_api_version(), rendering_method, device_index, _get_device_vendor_name(device), device.name)); } + // Pick the main queue family. It is worth noting we explicitly do not request the transfer bit, as apparently the specification defines + // that the existence of either the graphics or compute bit implies that the queue can also do transfer operations, but it is optional + // to indicate whether it supports them or not with the dedicated transfer bit if either is set. + BitField<RDD::CommandQueueFamilyBits> main_queue_bits; + main_queue_bits.set_flag(RDD::COMMAND_QUEUE_FAMILY_GRAPHICS_BIT); + main_queue_bits.set_flag(RDD::COMMAND_QUEUE_FAMILY_COMPUTE_BIT); + +#if !FORCE_SEPARATE_PRESENT_QUEUE + // Needing to use a separate queue for presentation is an edge case that remains to be seen what hardware triggers it at all. + main_queue_family = driver->command_queue_family_get(main_queue_bits, main_surface); + if (!main_queue_family && (main_surface != 0)) +#endif { - // Begin the first command buffer for the first frame, so - // setting up things can be done in the meantime until swap_buffers(), which is called before advance. - bool ok = driver->command_buffer_begin(frames[0].setup_command_buffer); - ERR_FAIL_COND(!ok); - - ok = driver->command_buffer_begin(frames[0].draw_command_buffer); - ERR_FAIL_COND(!ok); - if (local_device.is_null()) { - context->set_setup_buffer(frames[0].setup_command_buffer); // Append now so it's added before everything else. - context->append_command_buffer(frames[0].draw_command_buffer); - } + // If it was not possible to find a main queue that supports the surface, we attempt to get two different queues instead. + main_queue_family = driver->command_queue_family_get(main_queue_bits); + present_queue_family = driver->command_queue_family_get(BitField<RDD::CommandQueueFamilyBits>(), main_surface); + ERR_FAIL_COND_V(!present_queue_family, FAILED); } - for (int i = 0; i < frame_count; i++) { - // Reset all queries in a query pool before doing any operations with them. + ERR_FAIL_COND_V(!main_queue_family, FAILED); + + // Create the main queue. + main_queue = driver->command_queue_create(main_queue_family, true); + ERR_FAIL_COND_V(!main_queue, FAILED); + + if (present_queue_family) { + // Create the presentation queue. + present_queue = driver->command_queue_create(present_queue_family); + ERR_FAIL_COND_V(!present_queue, FAILED); + } else { + present_queue = main_queue; + } + + // Create data for all the frames. + for (uint32_t i = 0; i < frames.size(); i++) { + frames[i].index = 0; + + // Create command pool, command buffers, semaphores and fences. + frames[i].command_pool = driver->command_pool_create(main_queue_family, RDD::COMMAND_BUFFER_TYPE_PRIMARY); + ERR_FAIL_COND_V(!frames[i].command_pool, FAILED); + frames[i].setup_command_buffer = driver->command_buffer_create(frames[i].command_pool); + ERR_FAIL_COND_V(!frames[i].setup_command_buffer, FAILED); + frames[i].draw_command_buffer = driver->command_buffer_create(frames[i].command_pool); + ERR_FAIL_COND_V(!frames[i].draw_command_buffer, FAILED); + frames[i].setup_semaphore = driver->semaphore_create(); + ERR_FAIL_COND_V(!frames[i].setup_semaphore, FAILED); + frames[i].draw_semaphore = driver->semaphore_create(); + ERR_FAIL_COND_V(!frames[i].draw_semaphore, FAILED); + frames[i].draw_fence = driver->fence_create(); + ERR_FAIL_COND_V(!frames[i].draw_fence, FAILED); + frames[i].draw_fence_signaled = false; + + // Create query pool. + frames[i].timestamp_pool = driver->timestamp_query_pool_create(max_timestamp_query_elements); + frames[i].timestamp_names.resize(max_timestamp_query_elements); + frames[i].timestamp_cpu_values.resize(max_timestamp_query_elements); + frames[i].timestamp_count = 0; + frames[i].timestamp_result_names.resize(max_timestamp_query_elements); + frames[i].timestamp_cpu_result_values.resize(max_timestamp_query_elements); + frames[i].timestamp_result_values.resize(max_timestamp_query_elements); + frames[i].timestamp_result_count = 0; + } + + // Start from frame count, so everything else is immediately old. + frames_drawn = frames.size(); + + // Initialize recording on the first frame. + driver->begin_segment(frame, frames_drawn++); + driver->command_buffer_begin(frames[0].setup_command_buffer); + driver->command_buffer_begin(frames[0].draw_command_buffer); + + // Create draw graph and start it initialized as well. + draw_graph.initialize(driver, frames.size(), main_queue_family, SECONDARY_COMMAND_BUFFERS_PER_FRAME); + draw_graph.begin(); + + for (uint32_t i = 0; i < frames.size(); i++) { + // Reset all queries in a query pool before doing any operations with them.. driver->command_timestamp_query_pool_reset(frames[0].setup_command_buffer, frames[i].timestamp_pool, max_timestamp_query_elements); } + // Convert block size from KB. staging_buffer_block_size = GLOBAL_GET("rendering/rendering_device/staging_buffer/block_size_kb"); staging_buffer_block_size = MAX(4u, staging_buffer_block_size); - staging_buffer_block_size *= 1024; // Kb -> bytes. + staging_buffer_block_size *= 1024; + + // Convert staging buffer size from MB. staging_buffer_max_size = GLOBAL_GET("rendering/rendering_device/staging_buffer/max_size_mb"); staging_buffer_max_size = MAX(1u, staging_buffer_max_size); staging_buffer_max_size *= 1024 * 1024; @@ -4884,49 +5080,50 @@ void RenderingDevice::initialize(ApiContextRD *p_context, bool p_local_device) { // Validate enough blocks. staging_buffer_max_size = staging_buffer_block_size * 4; } + texture_upload_region_size_px = GLOBAL_GET("rendering/rendering_device/staging_buffer/texture_upload_region_size_px"); texture_upload_region_size_px = nearest_power_of_2_templated(texture_upload_region_size_px); - frames_drawn = frame_count; // Start from frame count, so everything else is immediately old. - // Ensure current staging block is valid and at least one per frame exists. staging_buffer_current = 0; staging_buffer_used = false; - for (int i = 0; i < frame_count; i++) { + for (uint32_t i = 0; i < frames.size(); i++) { // Staging was never used, create a block. - Error err = _insert_staging_block(); + err = _insert_staging_block(); ERR_CONTINUE(err != OK); } draw_list = nullptr; - draw_list_count = 0; - compute_list = nullptr; - pipelines_cache_file_path = "user://vulkan/pipelines"; - pipelines_cache_file_path += "." + context->get_device_name().validate_filename().replace(" ", "_").to_lower(); - if (Engine::get_singleton()->is_editor_hint()) { - pipelines_cache_file_path += ".editor"; - } - pipelines_cache_file_path += ".cache"; + if (main_instance) { + // Only the instance that is not a local device and is also the singleton is allowed to manage a pipeline cache. + pipeline_cache_file_path = "user://vulkan/pipelines"; + pipeline_cache_file_path += "." + device.name.validate_filename().replace(" ", "_").to_lower(); + if (Engine::get_singleton()->is_editor_hint()) { + pipeline_cache_file_path += ".editor"; + } + + pipeline_cache_file_path += ".cache"; - Vector<uint8_t> cache_data = _load_pipeline_cache(); - pipelines_cache_enabled = driver->pipeline_cache_create(cache_data); - if (pipelines_cache_enabled) { - pipelines_cache_size = driver->pipeline_cache_query_size(); - print_verbose(vformat("Startup PSO cache (%.1f MiB)", pipelines_cache_size / (1024.0f * 1024.0f))); + Vector<uint8_t> cache_data = _load_pipeline_cache(); + pipeline_cache_enabled = driver->pipeline_cache_create(cache_data); + if (pipeline_cache_enabled) { + pipeline_cache_size = driver->pipeline_cache_query_size(); + print_verbose(vformat("Startup PSO cache (%.1f MiB)", pipeline_cache_size / (1024.0f * 1024.0f))); + } } - draw_graph.initialize(driver, frame_count, SECONDARY_COMMAND_BUFFERS_PER_FRAME); + return OK; } Vector<uint8_t> RenderingDevice::_load_pipeline_cache() { - DirAccess::make_dir_recursive_absolute(pipelines_cache_file_path.get_base_dir()); + DirAccess::make_dir_recursive_absolute(pipeline_cache_file_path.get_base_dir()); - if (FileAccess::exists(pipelines_cache_file_path)) { + if (FileAccess::exists(pipeline_cache_file_path)) { Error file_error; - Vector<uint8_t> file_data = FileAccess::get_file_as_bytes(pipelines_cache_file_path, &file_error); + Vector<uint8_t> file_data = FileAccess::get_file_as_bytes(pipeline_cache_file_path, &file_error); return file_data; } else { return Vector<uint8_t>(); @@ -4935,11 +5132,11 @@ Vector<uint8_t> RenderingDevice::_load_pipeline_cache() { void RenderingDevice::_update_pipeline_cache(bool p_closing) { { - bool still_saving = pipelines_cache_save_task != WorkerThreadPool::INVALID_TASK_ID && !WorkerThreadPool::get_singleton()->is_task_completed(pipelines_cache_save_task); + bool still_saving = pipeline_cache_save_task != WorkerThreadPool::INVALID_TASK_ID && !WorkerThreadPool::get_singleton()->is_task_completed(pipeline_cache_save_task); if (still_saving) { if (p_closing) { - WorkerThreadPool::get_singleton()->wait_for_task_completion(pipelines_cache_save_task); - pipelines_cache_save_task = WorkerThreadPool::INVALID_TASK_ID; + WorkerThreadPool::get_singleton()->wait_for_task_completion(pipeline_cache_save_task); + pipeline_cache_save_task = WorkerThreadPool::INVALID_TASK_ID; } else { // We can't save until the currently running save is done. We'll retry next time; worst case, we'll save when exiting. return; @@ -4950,7 +5147,7 @@ void RenderingDevice::_update_pipeline_cache(bool p_closing) { { size_t new_pipelines_cache_size = driver->pipeline_cache_query_size(); ERR_FAIL_COND(!new_pipelines_cache_size); - size_t difference = new_pipelines_cache_size - pipelines_cache_size; + size_t difference = new_pipelines_cache_size - pipeline_cache_size; bool must_save = false; @@ -4962,7 +5159,7 @@ void RenderingDevice::_update_pipeline_cache(bool p_closing) { } if (must_save) { - pipelines_cache_size = new_pipelines_cache_size; + pipeline_cache_size = new_pipelines_cache_size; } else { return; } @@ -4971,7 +5168,7 @@ void RenderingDevice::_update_pipeline_cache(bool p_closing) { if (p_closing) { _save_pipeline_cache(this); } else { - pipelines_cache_save_task = WorkerThreadPool::get_singleton()->add_native_task(&_save_pipeline_cache, this, false, "PipelineCacheSave"); + pipeline_cache_save_task = WorkerThreadPool::get_singleton()->add_native_task(&_save_pipeline_cache, this, false, "PipelineCacheSave"); } } @@ -4987,7 +5184,7 @@ void RenderingDevice::_save_pipeline_cache(void *p_data) { } print_verbose(vformat("Updated PSO cache (%.1f MiB)", cache_blob.size() / (1024.0f * 1024.0f))); - Ref<FileAccess> f = FileAccess::open(self->pipelines_cache_file_path, FileAccess::WRITE, nullptr); + Ref<FileAccess> f = FileAccess::open(self->pipeline_cache_file_path, FileAccess::WRITE, nullptr); if (f.is_valid()) { f->store_buffer(cache_blob); } @@ -5122,10 +5319,15 @@ uint64_t RenderingDevice::limit_get(Limit p_limit) const { } void RenderingDevice::finalize() { - // Free all resources. + if (!frames.is_empty()) { + // Wait for all frames to have finished rendering. + _flush_and_stall_for_all_frames(); + } - _flush(false); + // Delete everything the graph has created. + draw_graph.finalize(); + // Free all resources. _free_rids(render_pipeline_owner, "Pipeline"); _free_rids(compute_pipeline_owner, "Compute"); _free_rids(uniform_set_owner, "UniformSet"); @@ -5181,9 +5383,12 @@ void RenderingDevice::finalize() { _free_pending_resources(f); driver->command_pool_free(frames[i].command_pool); driver->timestamp_query_pool_free(frames[i].timestamp_pool); + driver->semaphore_free(frames[i].setup_semaphore); + driver->semaphore_free(frames[i].draw_semaphore); + driver->fence_free(frames[i].draw_fence); } - if (pipelines_cache_enabled) { + if (pipeline_cache_enabled) { _update_pipeline_cache(true); driver->pipeline_cache_free(); } @@ -5205,6 +5410,34 @@ void RenderingDevice::finalize() { } framebuffer_formats.clear(); + // Delete the swap chains created for the screens. + for (const KeyValue<DisplayServer::WindowID, RDD::SwapChainID> &it : screen_swap_chains) { + driver->swap_chain_free(it.value); + } + + screen_swap_chains.clear(); + + // Delete the command queues. + if (present_queue) { + if (main_queue != present_queue) { + // Only delete the present queue if it's unique. + driver->command_queue_free(present_queue); + } + + present_queue = RDD::CommandQueueID(); + } + + if (main_queue) { + driver->command_queue_free(main_queue); + main_queue = RDD::CommandQueueID(); + } + + // Delete the driver once everything else has been deleted. + if (driver != nullptr) { + context->driver_free(driver); + driver = nullptr; + } + // All these should be clear at this point. ERR_FAIL_COND(dependency_map.size()); ERR_FAIL_COND(reverse_dependency_map.size()); @@ -5212,7 +5445,7 @@ void RenderingDevice::finalize() { RenderingDevice *RenderingDevice::create_local_device() { RenderingDevice *rd = memnew(RenderingDevice); - rd->initialize(context, true); + rd->initialize(context); return rd; } @@ -5291,7 +5524,7 @@ void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("screen_get_width", "screen"), &RenderingDevice::screen_get_width, DEFVAL(DisplayServer::MAIN_WINDOW_ID)); ClassDB::bind_method(D_METHOD("screen_get_height", "screen"), &RenderingDevice::screen_get_height, DEFVAL(DisplayServer::MAIN_WINDOW_ID)); - ClassDB::bind_method(D_METHOD("screen_get_framebuffer_format"), &RenderingDevice::screen_get_framebuffer_format); + ClassDB::bind_method(D_METHOD("screen_get_framebuffer_format", "screen"), &RenderingDevice::screen_get_framebuffer_format, DEFVAL(DisplayServer::MAIN_WINDOW_ID)); ClassDB::bind_method(D_METHOD("draw_list_begin_for_screen", "screen", "clear_color"), &RenderingDevice::draw_list_begin_for_screen, DEFVAL(DisplayServer::MAIN_WINDOW_ID), DEFVAL(Color())); @@ -5889,17 +6122,15 @@ void RenderingDevice::_bind_methods() { } RenderingDevice::~RenderingDevice() { - if (local_device.is_valid()) { - finalize(); - context->local_device_free(local_device); - } + finalize(); + if (singleton == this) { singleton = nullptr; } } RenderingDevice::RenderingDevice() { - if (singleton == nullptr) { // there may be more rendering devices later + if (singleton == nullptr) { singleton = this; } } |