ayo
34e6b0fe71
Add an experimental FSR Ray Regeneration backend for D3D12 RR requests: - Convert depth, motion, normals and albedos into AMD denoiser inputs. - Split and recompose diffuse/specular signals, then run Super Sampling. - Add stability/detail controls, demodulation guards and halo limiting. - Handle AMD provider discovery under NVIDIA GPU spoofing. - Add runtime probes, synthetic reference scenes and per-dispatch profiling. Add exact gfx11 DLSS-RR inference using f16 cooperative matrices: - Reproduce FP8 conversions, accumulation and camera reprojection arithmetic. - Widen model weights and avoid requiring VK_EXT_shader_float8. - Extend model building, replay tools and descriptor metadata. Expand motion preprocessing and experimental AW2 native-motion support: - Add tiled LDS dilation, padded allocation handling and thin-only sky fill. - Order dilation with input waits and NGX on the selected stream. - Add jitter-aligned native composition, hybrid routing and wire history repair. - Add prepared masks, sparse repair and independent motion/repair controls. Add optional fused R11G11B10 input/output conversion, sampled stage timings and profile summarization. Extend configuration, packaging, documentation, FidelityFX headers and CPU/GPU regression coverage.
47 lines
2.9 KiB
Plaintext
47 lines
2.9 KiB
Plaintext
d4r
|
|
Copyright 2026 countervolts and the d4r contributors
|
|
Modified in this fork for CUDA Ray Reconstruction support and validation (2026).
|
|
|
|
This product is licensed under the GNU General Public License, version 3 only
|
|
(GPL-3.0-only; see LICENSE), except for the files
|
|
listed below.
|
|
|
|
patches/zluda/*.patch
|
|
Changes to ZLUDA (https://github.com/vosen/ZLUDA). ZLUDA is dual-licensed under the Apache
|
|
License, Version 2.0 and the MIT license; the ZLUDA portions of these patches are offered
|
|
under the same terms. Changes to LLVM within these patches retain the Apache License,
|
|
Version 2.0 with LLVM exceptions.
|
|
|
|
patches/vkd3d-proton/*.patch
|
|
Changes to vkd3d-proton (https://github.com/HansKristian-Work/vkd3d-proton), which is licensed
|
|
under the GNU Lesser General Public License, version 2.1. These patches are offered under the
|
|
same license.
|
|
|
|
third_party/fidelityfx/
|
|
AMD FidelityFX SDK 2.3.0 API headers (ffx_api.h, ffx_api_types.h, ffx_denoiser.h), copied
|
|
unmodified; the SDK license lists them under the MIT license (third_party/fidelityfx/LICENSE.txt).
|
|
AMD's runtime DLLs are not included and remain under AMD's binary terms.
|
|
|
|
kernels/support/fsr_rr_prepare.comp
|
|
Its diffuse/specular split follows the DLSS-RR to FSR-RR adapter in Zach Hembree's OptiScaler
|
|
branch ffx-denoise-experimental and burak113's continuation (GPL-3.0).
|
|
|
|
Release zips (scripts/package_release.sh) also bundle binaries of ZLUDA (Apache-2.0 or MIT, with
|
|
LLVM under Apache-2.0 with LLVM exceptions), the patched vkd3d-proton (LGPL-2.1) and OptiScaler
|
|
(https://github.com/optiscaler/OptiScaler, GPL-3.0, unmodified). Each zip carries their license
|
|
texts in d4r/licenses and says where every file comes from in d4r/source/SOURCES.txt.
|
|
Release zips include install.sh to download NVIDIA's nvngx_dlss.dll and _nvngx.dll separately.
|
|
Optional builds may bundle those DLLs and nvngx_dlssd.dll. Release zips also include DLSS kernels compiled from NVIDIA's
|
|
PTX. Those are NVIDIA's property, are not covered by this repository's license, and are included
|
|
by whoever publishes the zip. Downloaded DLLs remain governed by NVIDIA's license terms.
|
|
|
|
This repository contains no NVIDIA software. The NVIDIA DLSS feature libraries (nvngx_dlss.dll and
|
|
nvngx_dlssd.dll) and the NGX core (_nvngx.dll) are NVIDIA's; you supply your own copies, and their use is governed by
|
|
NVIDIA's license terms. At build time, kernels/tools/extract_dlss_ptx.py reads PTX from your copy
|
|
of nvngx_dlss.dll into kernels/extracted/, which is not part of the repository and must not be
|
|
redistributed. The same holds for the model files the native engine's tools write (engine/README.md):
|
|
weights.bin, offsets.bin, lut.bin, hipnet.bin and model.json hold values read from NVIDIA's DLSS.
|
|
|
|
NVIDIA and DLSS are trademarks of NVIDIA Corporation. AMD, Radeon and FSR are trademarks of
|
|
Advanced Micro Devices, Inc. This project is not affiliated with or endorsed by either company.
|