want to run xess (xmx) on radeon? check out x4r!
nightly for windows and linux can be found here
d4r runs NVIDIA's official DLSS Super Resolution (nvngx_dlss.dll) and experimental Ray Reconstruction (nvngx_dlssd.dll) in Windows games on AMD Radeon GPUs under Linux and Proton. The game asks for DLSS as usual; the DLSS network runs on the AMD GPU through ZLUDA (CUDA on ROCm/HIP), with the heaviest kernels replaced by hand-written RDNA2, RDNA3 and RDNA4 code.
Supported DLSS models: DLSS 3 CNN (E), DLSS 4 transformer (K, default), and DLSS 4.5 transformer (M, plus L for Ultra Performance). Ray Reconstruction presets D, E and F are experimental (building). DLSS 5 is PURPOSELY not supported.
Proof of concept: d4r shows that dlss can be ran on radeon gpus, See supported games for the tested games and models. As well check the charts out below.
Not affiliated with NVIDIA or AMD.
To contact me my discord is ._ayo. or join the d4r discord server
DLSS Ultra Performance demo video - more videos
All tests are using DLSS K and FSR 4.1.1 and at 1440p (unless specified otherwise).
Radeon RX 7700 XT, SILENT HILL Townfall.
| Mode | DLSS 3 CNN (E)* | DLSS 4 (K) | DLSS 4.5 (M)* | FSR FPS | DLSS K upscale time | FSR upscale time |
|---|---|---|---|---|---|---|
| Quality | 71.7 fps | 75.7 | 49.3 | 73.3 | 2.05 ms | 2.42 ms |
| Balanced | 80.8 | 82.5 | 58.0 | 80.2 | 2.01 ms | 2.39 ms |
| Ultra Performance | 89.1 | 100.6 | 92.2 | 96.9 | 1.99 ms | 2.26 ms |
DLSS 4 (K) FPS was measured on 2026-10-21, Using no approximations. * Presets E and M are from 2026-09-29, on an older d4r build, so compare them with each other rather than with the newer columns.
Radeon RX 7700 XT, Ghost of Tsushima.
| Mode | DLSS 4 (K) | FSR | DLSS 1% low | FSR 1% low | DLSS upscale time | FSR upscale time |
|---|---|---|---|---|---|---|
| Native AA (DLAA / FSR AA) | 56.5 | 53.9 | 51.3 | 48.8 | 1.72 ms | 2.20 ms |
| Quality | 73.7 | 70.4 | 67.1 | 63.3 | 1.66 ms | 2.19 ms |
| Balanced | 78.0 | 74.1 | 68.3 | 66.7 | 1.64 ms | 2.16 ms |
| Performance | 80.8 | 77.1 | 72.2 | 66.4 | 1.61 ms | 2.14 ms |
| Ultra Performance | 85.7 | 78.9 | 75.8 | 68.0 | 1.57 ms | 2.09 ms |
Measured on 2026-11-12 at the highest possible graphics using no approximations. K is 4.7–8.6% faster than FSR 4.1.1 at every mode, with upscale times about 0.5 ms lower.
Radeon RX 7700 XT, Cyberpunk 2077.
| Mode | DLSS 4 (K) | FSR | DLSS upscale time | FSR upscale time |
|---|---|---|---|---|
| Quality | 87.6 | 86.4 | 2.08 ms | 2.28 ms |
| Balanced | 99.5 | 98.0 | 2.04 ms | 2.20 ms |
| Ultra Performance | 134.0 | 132.5 | 1.98 ms | 2.08 ms |
Measured on 2026-10-21, Using no approximations. DLSS averages 1.1–1.5% higher FPS than FSR across these runs, with upscale times 0.10–0.20 ms lower.
Open OptiScaler's menu (Insert):
- DLSS model: under DLSS Settings, tick Render Presets Override, pick the preset (E, K, L or M) and press Apply Changes. Unticking it returns to
[DLSS] Modelind4r/d4r.ini. - FSR: set Upscaler (DX12) to FSR 3.X/4. RDNA3 and RDNA4 get FSR 4.1 (needs GE-Proton, which supplies AMD's FSR 4 library); other GPUs get FSR 3.1.
Both are on by default; [OptiScaler] FollowPresets and Fsr4 in d4r.ini turn them off.
Change OptiScaler settings and use Save INI normally. Steam launches use the saved INI; d4r's ASI setup also preserves your choices after initial installation.
The vkhip backend runs preset K on RDNA3 and is on by default.
Experimental XeFG/FSR frame generation and XeLL/Anti-Lag latency conversion
are bundled in the OptiScaler/ folder. Enable FG in OptiScaler’s Insert menu or use the
per-game setup command with INI backups. Fresh installs keep FG off. The pinned v10
build is a prerelease; game testing is still required.
- In some games using some models native upscaling can show visible artifacting. For fidelity over speed, set
[Kernels] PreferAccuracy = trueind4r/d4r.ini(native kernel numerics).
!WE ARE LOOKING FOR VIDEOS OF IT WORKING, IF YOU HAVE A MENTIONED HERE AND DOESNT HAVE A VIDEO PLEASE TAKE ONE!
| GPU | Chip | Native kernels | Runtime testing |
|---|---|---|---|
| RX 7700 XT, RX 7800 XT, RX 7700, Radeon PRO W7700 | gfx1101 | K, M, L (WMMA) | RX 7700 XT |
| RX 7900 GRE / XT / XTX, Radeon PRO W7800 / W7900 | gfx1100 | K, M, L (WMMA) | RX 7900 XTX: external video |
| RX 7600 / 7600 XT, RX 7650 GRE, RDNA3 integrated GPUs | gfx1102, gfx1103 | K, M, L (WMMA) | untested |
| RX 9070 XT / 9070 / 9070 GRE, Radeon AI PRO R9700 | gfx1201 | K, M, L (WMMA, native FP8) | RX 9070 XT: external video |
| RX 9060 XT / 9060 | gfx1200 | K, M, L (WMMA, native FP8) | compile only |
| RX 6000 series, Steam Deck, RDNA2 integrated GPUs | gfx1030–gfx1036 | K (software matrix products) | validated on RDNA3 only |
| RDNA1 and older | unsupported |
RDNA2 has no WMMA instructions: its K kernels do the same matrix products with v_dot2 and DPP broadcasts, bit-identical to the WMMA build. Running that RDNA2 path on an RX 7700 XT for testing, K took 3.6 ms per 1080p frame, against 46.8 ms fully translated (what RDNA2 had before); RDNA3 itself keeps WMMA (1.48 ms). E and K are the models to use there; M and L run fully translated. Kernels missing for a GPU fall back to ZLUDA and can be much slower. Details: docs/native-kernels.md.
game (D3D12) ──► OptiScaler ──► d4r nvngx.dll (NGX D3D12 API) ──► NVIDIA's NGX core + nvngx_dlss.dll (CUDA path)
▼
native RDNA kernels ◄── ZLUDA (PTX → AMDGPU) ◄── nvcuda.dll (Wine CUDA bridge)
The shim keeps the game's inputs and output in VRAM through vkd3d-proton's Vulkan interop; ZLUDA compiles NVIDIA's PTX and serves hand-written kernels in place of the heaviest ones. An optional native inference engine ([Engine]) runs K, L and M with d4r's own Vulkan shaders. Details: docs/architecture.md.
- Extract
d4r-<version>.zipinto the folder that holds the game's main.exe(OptiScaler 10.0.0-pre1 asdxgi.dll, itsOptiScaler/dependencies, the d4r-patched vkd3d-proton, AMD's FSR files and thed4rfolder). - Run
bash install.shthere. It downloads NVIDIA's DLSS and NGX files from NVIDIA, checks their SHA-256 hashes and puts them ind4r/. Needs Bash, curl, sha256sum and 7-Zip; no sudo, ROCm not required. - In Steam, select a Proton version for the game (most work; tested: GE-Proton11-3). No launch options are needed:
d4r_boot.dll, which d4r'sdxgi.dllloads first, shows the game an NVIDIA GPU itself (whatPROTON_FORCE_NVAPI=1 DXVK_NVAPI_GPU_ARCH=AD100did before; those options still work and can be removed).
When updating, replace dxgi.dll, d4r_boot.dll, d3d12.dll, d3d12core.dll and d4r/nvngx.dll together. Already using OptiScaler? The -asi zip adds d4r as a plugin (ASI guide). The full guide ships in the zip as D4R_README.txt.
DLSS Frame Generation (NVIDIA's own, 2x-6x, and FG Lite) runs on RDNA3 and RDNA4: install.sh sets it up (it downloads DLSS-G 310.9.16 from NVIDIA's Streamline SDK and translates it on your machine), also for games without DLSS-G of their own through OptiScaler; d4r.ini [FrameGeneration] Model = auto uses FG Lite on RDNA3 and the full model on RDNA4 (docs/frame-generation.md).
Northlight games (Control, Alan Wake 2) keep their original motion vectors by default. Experimental display-resolution motion-vector reconstruction requires [Interop] NativeMotionVectors = true in d4r.ini and a matching patched driver. It can cause motion artifacts; restart the game after changing it.
Linux with an RDNA2, RDNA3 or RDNA4 GPU, ROCm 7.2 with HIP and its clang, GE-Proton 11, a Rust toolchain, meson/ninja, winegcc, x86_64-w64-mingw32-g++, clang-cl and Python 3. The full sequence is in docs/building.md; scripts/package_release.sh builds the release zip.
| Path | Contents |
|---|---|
tools/ |
the NGX shim, the Wine CUDA bridge, a D3D12 DLSS harness, probes |
kernels/ |
native kernels (k/ DLSS 4, m/ DLSS 4.5, tex/ texture tails, common/ matrix layouts) and their tools |
engine/ |
the native K/L/M inference engine |
patches/ |
ZLUDA, vkd3d-proton and OptiScaler patches |
scripts/, config/, packaging/, docs/ |
build/launch scripts, the developer d4r.ini, release files, documentation |
GNU General Public License, version 3 only (GPL-3.0-only; see LICENSE). The patches in patches/ are offered under the licenses of the projects they modify: ZLUDA (Apache-2.0 or MIT) and vkd3d-proton (LGPL-2.1). See NOTICE. The downloaded NVIDIA files and the kernels built from NVIDIA's code are not covered by this license.