Brad rebase and refactor fix for staging buffer memory types #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* Fix raytracing trim capture crash The commit fix raytracing trim capture crash issue for a d3d12 title. In trim handling of BuildRaytracingAccelerationStructure, some alias resource cause GetResourceWrapperForGpuVa return wrong resource and trigger GPU crash later, the commit fix the issue. * Add some changes to address review comments Remove default argument to use both GPUVA and buffer size for searching resource, add copyright info. * Add new functions to address review comments Add new functions for searching resource wrapper through GetAccelerationStructureResourceWrapperForGpuVa which return resource used for raytracing acceleration structure. * Add some changes for review comments Add resource desc struct in the ID3D12ResourceInfo so resource desc info can be used without injecting API call, remove GetAccelerationStructureResourceWrapperForGpuVa and combine its function into GetResourceWrapperForGpuVa. * Add some changes for source code optimization Change function names and optimize some source code to address review comments. * Add changes for GPUVA searching function Add changes for checking resource initial state to find matched acceleration structurer resouce, remove resource desc in resource info and also remove checking resource flags. * Change function pointer name Change function pointer name used for checking acceleration structure resource and add some other changes to address review comments. * Change function pointer type and parameter name Change function pointer tyoe and parameter name to more specific name and add some other code optimization changes. * Add some minor change Change function GetResourceWrapperForGpuVa parameter name to be more specific. * Add some change to address review comments Remove checking resource initial state in TrackBuildRaytracingAccelerationStructure method when searching GPUVA for D3D12_BUILD_RAYTRACING_ACCELERATION_STRUCTURE_INPUTS, remove unnecessary assert. --------- Co-authored-by: Zheng <[email protected]>
The problem The message is out of date. The solution Update the message.
Allocating the backing store for all these containers on every call to the function and then freeing it on every exit can be avoided by having them be member variables that are cleared on each call. At least for the vectors this should change their size but not their capacity, and thus not hitting the heap allocation functions once a steady state of reserved capacity is reached.
It forces to using only QueueFamilyIndex: 0 and queueCount: 1 on capturing to avoid replay error for unavailble VkQueue.
* Annotate GPUVA Problem If a title updated DXR resource data by Shader and dispatch the shader at same commandlist as DXR call, GFXR was not able to read out the DXR data and patch it. Solution GFXR capturer annotates all GPUVA, GPU Descriptor Handle and ShaderID during capture, scans managed CPU memory and marks the data which have been annotated. When replay the trace file, GFXR only need to translate the marked value. Test If there were artifacts during capture, try to change the MASK of GPUVA by ENV. * Meger dev branch
|
Hi, @jacobv-nvidia ! If this PR looks good to you, and you merge it with your branch, then LunarG#1115 is probably ready to merge. |
bradgrantham-lunarg
changed the title
|
Hey, sorry for the delay. Seeing how these changes have already been merged as part of LunarG#1173, I'll close this out. Thanks! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Elevate the logic for choosing the device memory properties into the functions that have that context