Implement pipeline caches

[rael-g]

Checklist

• cargo clippy reports no issues
• cargo doc reports no issues
• cargo deny issues have been fixed or added to deny.toml
• Human-readable change descriptions added to CHANGELOG.md under the "Unreleased" heading.
  • If the change does not affect the user (or is a process change), preface the change with "Internal:"
  • Add credit to yourself for each change: Added new functionality. @githubname

Description

Exposes pipeline caches to the C API, letting the result of shader compilation be reused between executions of a program.

wgpu-core already supports this, and WGPUNativeFeature_PipelineCache was already exposed, but there was no way to actually use it: both pipeline descriptors hardcoded cache: None behind a // TODO(wgpu.h).

Adds:

• The WGPUPipelineCache object, with wgpuDeviceCreatePipelineCache, wgpuPipelineCacheGetData, wgpuPipelineCacheAddRef and wgpuPipelineCacheRelease.
• WGPUPipelineDescriptorExtras, chained onto WGPURenderPipelineDescriptor or WGPUComputePipelineDescriptor to compile a pipeline through a cache.
• A pipeline_cache example, wired into CI. It creates a cache, compiles a pipeline through it, reads the resulting data back, and recreates the pipeline from that data. Adapters that do not support pipeline caches skip it rather than fail, so it is safe to run everywhere.

wgpuPipelineCacheGetData follows the same two-call convention as wgpuInstanceEnumerateAdapters: call it once with data set to NULL to obtain the size, then again with a buffer.

Related Issues