The Ray Reconstruction engine caches one immutable descriptor set per unique combination of input views (hard limit: 16). Games that rotate their intermediate textures through a ring buffer generate a new combination each frame; with K independent ring buffers of size N the number of distinct combinations grows as N^K, quickly exceeding the cache limit.
Fix: track the number of distinct VkImageView handles seen per input (11 inputs: colour, motion, depth, 8 guides). When an input exceeds 4 distinct views it is flagged as rotating and its data is copied to a canonical plane (a stable internal texture) instead of being borrowed directly. The `direct` lambda now takes an index and checks the per-input flag, so only the inputs that actually rotate incur a blit.
Stable inputs (single view, no ring buffer) continue to be borrowed directly with zero extra cost. The worst-case overhead is one blit per rotating input per frame (~0.05 ms each).
The flag is one-way: once an input is marked as rotating it stays that way for the lifetime of the engine. If a game changes its rendering mode mid-session the flag persists, adding a negligible number of extra blits.
- GameFrameR/RFrameImages: DLSSD.ScreenSpaceSubsurfaceScatteringGuide (arg 264)
and DLSSD.ColorBeforeParticles (output stage, texture 416 with its own history
pair 384/392, zero-initialised on first use); guides array 6 -> 8
- DLSS.Use.HW.Depth: byte 980 and output byte 443 bit 24 (Cyberpunk passes the
nonlinear depth buffer); replay's 'inverted' field becomes a bitmask
- shim: bind the two new resources, read the HW-depth flag, and accept every
preset hint except E's 5 by mapping it to D's diamond_wallaby kernels
(harness captures show 0x80000002/0x80000005 run preset D)
- harness: D4R_HARNESS_RR_EXTRA=particles,sss and D4R_HARNESS_RR_HW_DEPTH=1
- ptx2rr: out-of-range integer fetches return zero, matching ZLUDA's behaviour
(the reference the stages are checked against)
- engine/rr/README: the capture findings behind the offsets
- engine/rr: d4r's own shaders for the denoiser frame (exposure, six encoders,
five decoders, hkpn output stage) at NGX layouts, plus PTX->GLSL/HIP packing
scripts, reference implementations and weight extraction
- runtime.h/runtime_rr.inc, game_rr.{h,cpp}: GameDenoiser wired into the engine
build (d4r-rr replay binary) and the D3D12 shim's feature-13 path
- shim: answer GetFeatureRequirements from d4r's own state only (the old
cross-binary CUDA handoff crashed OptiScaler's scan); drop the core's
CUDA GetFeatureRequirements import
- shim/harness: exact guide capture/replay of a frame's RR inputs
- kernels/rr: expanded enc0/engine prototypes; kernels/tools: ptxsim rewrite
plus rrsim and an engine test; vulkan: ptx2glsl FP8/cooperative-matrix
support, kvk inout buffers and KVK_DISABLE_OPTIMIZATION
- bridge: report the surface array in cuSurfObjectCreate traces
- packaging: RR settings and the ICU launch overrides in D4R_README
Tk-Glitch