fabi/Xenia-Canary
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[D3D12/Vulkan] Simplify host GPU fence management

Browse Source 
Replace the `SubmissionTracker`s with new `GPUCompletionTimeline`s with a
more unified interface (using a base class), and without the internal logic
for queue ownership transfers since that idea was scrapped during the
development of the `Presenter`.

Also use this fence management logic for GPU emulation, though without
architectural reworks for now, just on the bottom level.

Still very messy, but can be cleaned up in further GPU command processor
and presenter reworks.
This commit is contained in:
Triang3l
2025-12-14 21:10:06 +03:00
parent 01ae24e46e
commit fe1fd36137
18 changed files with 894 additions and 974 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
src/xenia/gpu/vulkan/vulkan_command_processor.h
View File

@@ -38,6 +38,7 @@
#include "xenia/gpu/xenos.h"
#include "xenia/kernel/kernel_state.h"
#include "xenia/ui/vulkan/linked_type_descriptor_set_allocator.h"
#include "xenia/ui/vulkan/vulkan_gpu_completion_timeline.h"
#include "xenia/ui/vulkan/vulkan_presenter.h"
#include "xenia/ui/vulkan/vulkan_provider.h"
#include "xenia/ui/vulkan/vulkan_upload_buffer_pool.h"
@@ -157,10 +158,11 @@ class VulkanCommandProcessor : public CommandProcessor {
bool submission_open() const { return submission_open_; }
uint64_t GetCurrentSubmission() const {
return submission_completed_ +
uint64_t(submissions_in_flight_fences_.size()) + 1;
return completion_timeline_.GetUpcomingSubmission();
}
uint64_t GetCompletedSubmission() const {
return completion_timeline_.GetCompletedSubmissionFromLastUpdate();
}
uint64_t GetCompletedSubmission() const { return submission_completed_; }
// Sparse binds are:
// - In a single submission, all submitted in one vkQueueBindSparse.
@@ -403,7 +405,7 @@ class VulkanCommandProcessor : public CommandProcessor {
// Rechecks submission number and reclaims per-submission resources. Pass 0 as
// the submission to await to simply check status, or pass
// GetCurrentSubmission() to wait for all queue operations to be completed.
void CheckSubmissionFenceAndDeviceLoss(uint64_t await_submission);
void CheckSubmissionCompletionAndDeviceLoss(uint64_t await_submission);
// If is_guest_command is true, a new full frame - with full cleanup of
// resources and, if needed, starting capturing - is opened if pending (as
// opposed to simply resuming after mid-frame synchronization). Returns
@@ -414,8 +416,9 @@ class VulkanCommandProcessor : public CommandProcessor {
// successfully, if it has failed, leaves it open.
bool EndSubmission(bool is_swap);
bool AwaitAllQueueOperationsCompletion() {
CheckSubmissionFenceAndDeviceLoss(GetCurrentSubmission());
return !submission_open_ && submissions_in_flight_fences_.empty();
CheckSubmissionCompletionAndDeviceLoss(GetCurrentSubmission());
return !submission_open_ &&
GetCompletedSubmission() + 1u >= GetCurrentSubmission();
}
void ClearTransientDescriptorPools();
@@ -460,16 +463,14 @@ class VulkanCommandProcessor : public CommandProcessor {
VkPipelineStageFlags guest_shader_pipeline_stages_ = 0;
VkShaderStageFlags guest_shader_vertex_stages_ = 0;
std::vector<VkFence> fences_free_;
std::vector<VkSemaphore> semaphores_free_;
ui::vulkan::VulkanGPUCompletionTimeline completion_timeline_;
bool submission_open_ = false;
uint64_t submission_completed_ = 0;
// In case vkQueueSubmit fails after something like a successful
// vkQueueBindSparse, to wait correctly on the next attempt.
std::vector<VkSemaphore> current_submission_wait_semaphores_;
std::vector<VkPipelineStageFlags> current_submission_wait_stage_masks_;
std::vector<VkFence> submissions_in_flight_fences_;
std::deque<std::pair<uint64_t, VkSemaphore>>
submissions_in_flight_semaphores_;