{
  "added_at": "2026-10-18",
  "additional_evidence_urls": [
    "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/docs/ARCHITECTURE.md",
    "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/docs/ngx-verification/310.9.16/NGX-VERIFICATION.md",
    "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/docs/ngx-verification/310.8.0/NGX-VERIFICATION.md",
    "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/THIRD_PARTY.md",
    "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/CLAUDE.md",
    "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/LICENSE"
  ],
  "category": "reconstruction-cases",
  "description_en": "Reimplements the neural-rendering network of NVIDIA’s DLSS 5 library as Vulkan compute shaders for AMD RX 9000 (RDNA4) cards and supplies it to games through an NGX-compatible core and ReShade or Vulkan-layer routes. The weights are not distributed: installers extract them, hash-checked, from the user’s own nvngx_dlssnr.dll 310.9.16 or 310.8.0. The project compares its output with NVIDIA’s DLL on an RTX 5090 and reports close, not bit-identical, pictures.",
  "description_zh": "将 NVIDIA DLSS 5 库中的神经渲染网络重新实现为面向 AMD RX 9000（RDNA4）显卡的 Vulkan 计算着色器，并通过兼容 NGX 的核心模块以及 ReShade 或 Vulkan 层等途径提供给游戏。项目不分发权重：安装程序从使用者自己的 nvngx_dlssnr.dll 310.9.16 或 310.8.0 中提取权重并核对哈希。项目将输出与 NVIDIA DLL 在 RTX 5090 上的结果对比，报告画面接近但并非逐位一致。",
  "entry_url": "https://awesomerea.org/resources/dlssnr-amd-neural-rendering-case/",
  "entry_url_zh": "https://awesomerea.org/zh/resources/dlssnr-amd-neural-rendering-case/",
  "evidence_url": "https://github.com/mochizuki0323/DLSSNR-AMD/blob/ab00d2b23584de0f42ac91eca4d7d2d2fad11465/README.md",
  "group": "cases",
  "id": "dlssnr-amd-neural-rendering-case",
  "json_url": "https://awesomerea.org/resources/dlssnr-amd-neural-rendering-case/record.json",
  "kind": "case-study",
  "markdown_url": "https://awesomerea.org/resources/dlssnr-amd-neural-rendering-case/index.md",
  "markdown_url_zh": "https://awesomerea.org/zh/resources/dlssnr-amd-neural-rendering-case/index.md",
  "name": "DLSSNR-AMD: reimplementing DLSS 5 neural rendering for RDNA4",
  "name_zh": "DLSSNR-AMD：为 RDNA4 重新实现 DLSS 5 神经渲染",
  "reconstruction_method": "other",
  "record_type": "resource",
  "relationship": "independent-reconstruction",
  "requirement_facets": [
    "prerequisites",
    "prerequisites",
    "prerequisites",
    "testing",
    "license"
  ],
  "requirements": [
    "The reviewed 0.0.x main targets RX 9000 cards because the network needs RDNA4 FP8 matrix instructions, reached through VK_KHR_cooperative_matrix; README says it was tested only on an RX 9070 XT. Linux under Steam/Proton with Mesa 26.2+ is the main version; Windows is an experimental preview tested with AMD Software 26.9.2, slower by default and closer to Linux with precompiled ACO machine code. OpenGL games are unsupported, and README warns against online games with anti-cheat.",
    "Requires the user’s own nvngx_dlssnr.dll, version 310.9.16 (both models) or 310.8.0 (the old model only); other versions are refused, and a model file is written only when every extracted entry matches a known hash. README says packages containing the model must not be shared. The optional int4-mixed network derives its weights at install from the user’s model and project calibration tables, which THIRD_PARTY.md says hold no weight values.",
    "The documented build runs on Ubuntu 24.04 with git, curl, Python 3, CMake, Ninja, MinGW-w64 and multilib g++; fetch_deps.sh pins third-party components, including glslang 16.5.0 for reproducible SPIR-V, and the Windows DLLs are cross-compiled. Games reach the network through an NGX core loaded by OptiScaler-NR, for DirectX 11/12 titles with an upscaler option, or through ReShade-based routes that estimate motion from the picture.",
    "For the 310.9.16 model, the verification notes report single-frame PSNR of 49.15–50.59 dB and SSIM of about 0.997 against NVIDIA’s unmodified DLL on an RTX 5090, plus short moving sequences, from one Tomb Raider (2013) frame and its downscales; the 310.8.0 comparison reports 45.56–49.06 dB. These are project measurements on one AMD card.",
    "MIT covers the repository’s code; third-party components keep their own licenses, including GPL-3.0 OptiScaler-NR shipped as released. NVIDIA’s DLL and weights remain NVIDIA’s, and the project states it is not affiliated with NVIDIA or AMD. The repository does not describe how the network structure was recovered; its CLAUDE.md addresses coding assistants."
  ],
  "requirements_zh": [
    "所审阅的 0.0.x 版 main 分支面向 RX 9000 显卡，因为网络需要通过 VK_KHR_cooperative_matrix 调用 RDNA4 的 FP8 矩阵指令；README 称仅在 RX 9070 XT 上测试。主版本运行于 Steam／Proton 下的 Linux，需要 Mesa 26.2+；Windows 为实验性预览版，测试驱动为 AMD Software 26.9.2，默认较慢，使用预编译的 ACO 机器码后接近 Linux。不支持 OpenGL 游戏，README 还警告不要在带反作弊的在线游戏中使用。",
    "需要使用者自己的 nvngx_dlssnr.dll：310.9.16 版（含两个模型）或 310.8.0 版（仅旧模型）；其他版本会被拒绝，只有当每个提取条目都与已知哈希一致时才会写出模型文件。README 说明不得分享包含模型的软件包。可选的 int4 混合网络在安装时由使用者的模型和项目校准表生成权重，THIRD_PARTY.md 称这些表不含权重数值。",
    "文档说明的构建环境为 Ubuntu 24.04，需要 git、curl、Python 3、CMake、Ninja、MinGW-w64 和 multilib g++；fetch_deps.sh 固定第三方组件版本，其中 glslang 16.5.0 用于生成可复现的 SPIR-V，Windows DLL 通过交叉编译生成。游戏可经由 OptiScaler-NR 加载的 NGX 核心模块接入网络（适用于带超分辨率选项的 DirectX 11／12 游戏），也可经由从画面估计运动的 ReShade 途径接入。",
    "对于 310.9.16 模型，验证说明报告：以一帧 Tomb Raider (2013) 画面及其缩小版本为输入，与在 RTX 5090 上运行的未修改 NVIDIA DLL 相比，单帧 PSNR 为 49.15–50.59 dB，SSIM 约 0.997，并比较了短运动序列；310.8.0 的对比结果为 45.56–49.06 dB。这些是项目在单张 AMD 显卡上的测量。",
    "仓库代码采用 MIT 许可；第三方组件保留各自许可，其中按原样附带的 OptiScaler-NR 为 GPL-3.0。NVIDIA 的 DLL 和权重仍归 NVIDIA 所有，项目声明与 NVIDIA 或 AMD 无关。仓库未说明网络结构如何恢复；其 CLAUDE.md 面向编码助手。"
  ],
  "targets": [
    "binary",
    "game"
  ],
  "url": "https://github.com/mochizuki0323/DLSSNR-AMD",
  "verification": "source-reviewed",
  "verified_at": "2026-10-18"
}