- Add BeginRendering/EndRendering commands with imported/created attachment
views, view plumbing, and aspect-gated depth/stencil slots
- Derive and validate render-pass spans at compile time; contract spans into
single DAG nodes before hazard analysis; hoist cross-boundary barriers to
pass begins with in-pass policy and created-view resolution
- Replace the four-pass ordering pipeline with a single command walk
(build_command_orderings): span derivation, validation, lifetime-only
injection, dependency edges, and Kahn batching in one pass; batches are
ranges into execution_order and RenderPassSpan keeps only begin/end
- Perf: node-collapsed frontier analysis, dense command/resource indexing
(no SipHash on hot paths), inline barrier storage, exact preallocation,
lazy Kahn ready heap, scoped barrier fast paths
- Scope buffer and image alias barriers to the incoming resource per the
Vulkan resource-memory-aliasing rules (availability unscoped, visibility
scoped to the new owner)
- Add criterion compile benchmarks (two_pass/{no_pass,pass}/per_{20,50,100})
- Replace gpu-allocator with vk-mem 0.5.0 (VMA 3.3.0) in renderer_device
- Add public GpuAllocation wrapper, MemoryLocation, AllocationDesc, and
allocate_for_buffer/image + bind_buffer/image_memory helpers
- Restrict CPU-mapped allocations to HOST_VISIBLE|HOST_COHERENT memory types
- Handle guards: raw buffer/image handles owned before allocation so failure
paths never leak
- Use renderer_device::device::MemoryLocation across renderer, renderer_resources,
and renderer_rendergraph; remove all gpu-allocator dependencies
- Error on dedicated allocation below Vulkan 1.1 instead of silent downgrade
- Add Send/Sync asserts, coherent-mapping and dedicated-allocation validation
tests; update AGENTS.md docs
janis