Ray tracing compaction #6609
Ray tracing compaction #6609
Conversation
…mpaction # Conflicts: # wgpu-core/src/resource.rs
# Conflicts: # .gitattributes # wgpu-core/src/command/compute.rs # wgpu-core/src/command/render.rs # wgpu-core/src/device/life.rs # wgpu-core/src/device/resource.rs # wgpu-core/src/track/mod.rs
# Conflicts: # examples/src/ray_cube_compute/mod.rs # wgpu-core/src/command/ray_tracing.rs # wgpu-core/src/device/ray_tracing.rs # wgpu-core/src/resource.rs # wgpu-hal/src/dx12/command.rs # wgpu-hal/src/dx12/device.rs # wgpu-hal/src/metal/command.rs # wgpu-hal/src/vulkan/command.rs # wgpu/src/context.rs # wgpu/src/ray_tracing.rs
…cing-compaction # Conflicts: # examples/src/ray_cube_compute/mod.rs # player/src/lib.rs # wgpu-core/src/command/ray_tracing.rs # wgpu-core/src/device/ray_tracing.rs # wgpu-core/src/device/resource.rs # wgpu-core/src/device/trace.rs # wgpu-core/src/lock/rank.rs # wgpu-core/src/ray_tracing.rs # wgpu-core/src/resource.rs # wgpu-core/src/scratch.rs # wgpu-hal/src/empty.rs # wgpu-types/src/lib.rs # wgpu/src/api/command_encoder.rs # wgpu/src/backend/webgpu.rs # wgpu/src/backend/wgpu_core.rs # wgpu/src/context.rs
…ERATION_STRUCTURE_QUERY.
|
The synchronization around I don't understand the big picture here, but I think these things need to either use the trackers, or there needs to be a Mutex somewhere. |
|
Now that I think about it, I'm not sure that a mutex would be sufficient. I think that the synchronization may need to be redone. Here are the requirements I have thought up that should be implemented:
I'll get started on this. |
|
The synchronisation between threads should now be improved, and I think this should be synchronised correctly. |
|
Nvidia has this neat diagram showing the state machine of the compaction process https://developer.nvidia.com/blog/tips-acceleration-structure-compaction/#how_does_compaction_work |
…-tracing-compaction
…cing-compaction # Conflicts: # CHANGELOG.md # tests/tests/ray_tracing/as_build.rs # wgpu-core/src/device/queue.rs
Connections
Follow up to #6291
Description
Adds the function
compact_blasas specified in #1040. This needed two more functions in wgpu-hal. These are based on DX12 and the function callEmitRaytracingAccelerationStructurePostbuildInfo(). On Vulkan this call is emulated with query sets. Compaction does not need any additional featues. Depending on the GPU and BLAS contents compaction may not be possible. In that case the BLAS will be duplicated (logic done by GPU driver).This change is shown by ray_cube_compute, which now compacts the BLAS before using it. This shows the requirements of the function.
Testing
Adds two more tests, testing valid and invalid compaction operations.
Checklist
cargo fmt.taplo format. - done by CIcargo clippy. If applicable, add:--target wasm32-unknown-unknown--target wasm32-unknown-emscriptencargo xtask testto run tests.CHANGELOG.md. See simple instructions inside file.