{"name":"gpu-unlock-ada: Enable Tensor Speed-Select on Ada GPUs","description":"A Linux kernel module that uses the AD102 payload from TOTORONG/AD103_win_unlocker to enable an SM speed-select override before NVIDIA’s driver loads. It targets RTX 4090 and opt-in experimental Ada GPUs, with locally generated firmware assets.","github":"https://github.com/JustAResearcher/gpu-unlock-ada","url":"https://osrepos.com/repo/justaresearcher-gpu-unlock-ada","source":"osrepos.com","sourceDescription":"This repository profile is provided by osrepos.com, an open source repository discovery platform.","repositoryProfile":"https://osrepos.com/repo/justaresearcher-gpu-unlock-ada","generatedFor":"open source discovery and AI-assisted research","markdown":"https://osrepos.com/repo/justaresearcher-gpu-unlock-ada.md","json":"https://osrepos.com/repo/justaresearcher-gpu-unlock-ada.json","topics":["c","linux","performance","kernel-module","gpu-unlocking","tensor-computing"],"keywords":["c","linux","performance","kernel-module","gpu-unlocking","tensor-computing"],"stars":null,"summary":"A Linux kernel module that uses the AD102 payload from TOTORONG/AD103_win_unlocker to enable an SM speed-select override before NVIDIA’s driver loads. It targets RTX 4090 and opt-in experimental Ada GPUs, with locally generated firmware assets.","content":"## Overview\n\n`gpu-unlock-ada` is a Linux kernel module that integrates the AD102 payload from TOTORONG’s `AD103_win_unlocker` into the Linux module and firmware-generation base from Zagitta’s `gpu-unlock`. It enables an SM speed-select override before the stock NVIDIA driver binds, addressing reduced tensor instruction throughput on supported GPUs without requiring a patched NVIDIA driver or UEFI application.\n\nThe primary target is the RTX 4090; other Ada GPU IDs are available only as explicit experimental opt-ins, with evidence varying by model. This is a hardware-level change, not a general-purpose GPU performance tool: measured instruction throughput gains do not necessarily translate to application gains. The README also describes a separate Windows reference port for one MSI RTX 4070 Ti, but that does not add Windows support to the Linux module.\n\n## Key Features\n\n- Runs as a Linux kernel module early in boot, before the NVIDIA driver binds.\n- Selects the AD102 firmware and payload automatically for RTX 4090 devices.\n- Supports selected experimental Ada IDs through individual opt-in flags and, for some models, exact PCI-address selection.\n- Builds firmware assets locally from pinned NVIDIA inputs and verifies asset hashes before use.\n- Checks the payload completion marker, permissions and speed-select readout, including after PCIe function-level reset.\n- Includes host-side tests and optional CUDA instruction benchmarks, plus published hardware reports and testing guidance.\n- Provides package and initramfs integration for mkinitcpio, initramfs-tools and dracut.\n\n## Use Cases\n\n- RTX 4090 owners who want to test higher FP16 or BF16 tensor instruction throughput on Linux and can validate the result on their specific card.\n- Owners of listed experimental Ada GPUs who are prepared to opt in by exact device and assess the model-specific evidence before deployment.\n- GPU researchers investigating tensor instruction rates and firmware-level speed selection, rather than expecting equivalent gains in every workload.\n- Linux system builders who want an early-boot module and locally generated firmware assets instead of a patched NVIDIA driver or UEFI application.\n\n## Our Take\n\nThis is a narrowly scoped, actively maintained hardware experiment with meaningful validation, but its results and safety depend on the exact GPU and boot state.\n\n**Good fit if:**\n- You have a supported Linux GPU and can follow the cold-boot, driver-ordering and post-boot verification steps.\n- Your goal is to investigate FP16/BF16 or other measured tensor instruction rates, and you understand that application performance may differ.\n- You can build the module and firmware assets locally and accept GPL-2.0-only obligations for the covered project code.\n\n**Look elsewhere if:**\n- You need a broadly supported performance improvement for an unlisted GPU, or a guaranteed gain across inference and other application workloads.\n- You need Windows support from this repository; its Windows reference is a separate, single-profile contribution, not Linux-module target support.\n- You cannot validate GPU health and recover from a boot or firmware-state problem on the machine you plan to modify.\n\n## Project Health\n\n| Signal | Value |\n|---|---|\n| Status | **New**: less than six months old and changing fast |\n| Last commit | 2026-10-26 (1 day ago) |\n| Commits, last 90 days | 100+ |\n| Releases, last 12 months | 2 (latest v1.1.1, 2026-10-24) |\n| Contributors | 20 (top contributor: 43% of commits) |\n| Issues closed, last 90 days | 33 (typically closed in 1 day) |\n| Pull requests merged, last 90 days | 23 (typically merged in 0 days) |\n| Project age | 3 days |\n\nChecked on 2026-10-27 with the GitHub API.\n\n## Project Facts\n\n- Language: C\n- License: GPL-2.0\n- Stars: 4.1k\n- Forks: 538\n- Archived: no\n\n## What You Need\n\nDetected in the repository:\n\n- C (main language on GitHub)\n- A test suite\n\n## Getting Started\n\n```sh\ngit clone https://github.com/JustAResearcher/gpu-unlock-ada.git\ncd gpu-unlock-ada\nmake install\n```\n\nReview the README’s supported-target table and boot instructions before enabling the module. Build inputs and firmware are generated locally; consult the README for distribution-specific targets, kernel command-line options and validation steps.\n\n## License in Plain Words\n\n**GPL-2.0** (strong copyleft).\n\n- Commercial use: yes\n- Modify and redistribute: yes\n- You must keep: the license notice\n- Share your changes: yes, the full source of anything you distribute that includes it, under the GPL\n\nA summary, not legal advice: the LICENSE file is what applies.\n\n## Considerations\n\nThe repository is very new and changing quickly, although it has recent releases, frequent commits and issue and pull-request activity. Contributions are not dominated by one person, but the short project history limits what that says about long-term maintenance or hardware coverage.\n\nAn unlock requires a compatible GPU, Linux kernel headers, a working DKMS toolchain, locally generated firmware assets and correct initramfs ordering. The first attempt requires a cold start, and a warm reboot may not restore the hardware state needed to rerun the firmware sequence. Experimental Ada support is explicitly limited to listed IDs and small numbers of tested cards; laptop results in particular may not carry across models. Reported instruction-rate gains do not ensure faster real workloads, and some measured workloads showed no benefit. The module, packaging and payload integration are GPL-2.0-only; NVIDIA firmware remains subject to its own terms.","metrics":{"detailViews":2,"githubClicks":0},"dates":{"published":null,"modified":"2026-10-27T13:53:24.000Z"}}