RADV 24.0 Branchpoint Tracker
@daniel-schuermann
-
!26901 (merged) aco: refactor and speed-up dead code analysis -
!25676 (merged) aco: add new post-RA scheduler for ILP -
!18249 (merged) nir/opt_move_discards_to_top: don't schedule discard/demote across subgroup operations
@hakzsam
I already landed everything.
@bnieuwenhuizen
@pixelcluster
-
!26729 (merged) radv/rt: Acceleration structure updates
@pendingchaos
@DadSchoorse
@KonstantinSeurer
-
!26282 lavapipe: Depth bias control, tooling info and document more extensions -
!26821 (merged) vtn: Remove transpose(m0)*m1 fast path -
!26876 (merged) spirv: Handle DepthReplacing correctly -
!26812 (merged) ac/nir: Export clip distances according to clip_cull_mask -
!25548 (merged) radv/rra: Ray history traces -
!26697 (merged) radv: Improve acceleration structure SQTT tracing -
!25224 (merged) radv/rt: Skip compiling unused stages -
!26431 (merged) radv/rt: Lower ray payloads to registers -
!26895 (merged) radv: Implement VK_KHR_ray_tracing_position_fetch
@ishitatsuyuki
-
!26942 (merged) radv: Add a workaround to skip vkUnmapMemory calls, enable for Quantic Dream Engine
@Venemo
-
!26913 (merged) radv: Implement SDMA image copies and introduce perf test flag for transfer queues -
!25594 radv: Add a dedicated transfer-only queue using SDMA (including gang submissions)
Edited by Bas Nieuwenhuizen