{"name":"gpu-unlock-ada","repo":"JustAResearcher/gpu-unlock-ada","url":"https://midorreal.com/project/justaresearcher-gpu-unlock-ada","repository":"https://github.com/JustAResearcher/gpu-unlock-ada","description":"GPU tensor unlock source for RTX 4090 and experimental RTX 4070 Ti SUPER; firmware generated locally","facts":{"license":{"value":"GPL-2.0","source":"https://github.com/JustAResearcher/gpu-unlock-ada","checked":"22 Nov 2026"},"language":{"value":"C","source":"https://github.com/JustAResearcher/gpu-unlock-ada","checked":"22 Nov 2026"},"last_release":{"value":"v1.1.1, 24 Oct 2026","source":"https://github.com/JustAResearcher/gpu-unlock-ada/releases","checked":"22 Nov 2026"},"releases":{"value":"2","source":"https://github.com/JustAResearcher/gpu-unlock-ada/releases","checked":"19 Nov 2026"},"contributors":{"value":"25","source":"https://github.com/JustAResearcher/gpu-unlock-ada/graphs/contributors","checked":"19 Nov 2026"},"archived":{"value":"no","source":"https://github.com/JustAResearcher/gpu-unlock-ada","checked":"22 Nov 2026"}},"stars":4956.0,"writeup":{"what_it_is":"A Linux kernel module that enables an SM speed-select override on certain NVIDIA GPUs before the stock NVIDIA driver starts. It uses a payload integrated from a published AD102 analysis and firmware generated locally during the build.","audience":"Linux users with supported NVIDIA GeForce or Ada GPUs who want to test tensor throughput changes","claims":[{"kind":"specific","claim":"RTX 4090 support shows about 2x FP16 and BF16 tensor throughput with FP32 accumulation","excerpt":"RTX 4090 (`10de:2684`, AD102) support has demonstrated","status":"not_checked"},{"kind":"specific","claim":"Measured FP16 to FP32 accumulate throughput rises from 165.6 to 331.6 TF/s on the original test","excerpt":"| FP16 → FP32 accumulate | 165.6 TF/s | 331.6 TF/s |","status":"not_checked"},{"kind":"specific","claim":"INT8, dp4a and scalar FP32 stayed near their baseline rates","excerpt":"INT8, `dp4a` and scalar FP32 stayed near their","status":"not_checked"},{"kind":"specific","claim":"The module selects the AD102 firmware and payload automatically for RTX 4090 device ID 10de:2684","excerpt":"The module selects the AD102 firmware and payload automatically for RTX 4090","status":"not_checked"},{"kind":"specific","claim":"Build fetches pinned NVIDIA inputs over HTTPS and verifies their SHA-256 hashes","excerpt":"Builds fetch the pinned NVIDIA `610.43.03` source and binary inputs over HTTPS","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-11-10"},"why_now":null,"owner_supplied":null,"score":{"verdict":"overhyped","hype":93,"reality":50,"gap":43,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T07:00:00+00:00","verdict_source":"rule"},"early_read":null,"changes":[{"date":"2026-10-24T00:28:11+00:00","text":"Latest release v1.1.1","source":"https://github.com/JustAResearcher/gpu-unlock-ada/releases"},{"date":"2026-10-23T21:01:29+00:00","text":"First release","source":"https://github.com/JustAResearcher/gpu-unlock-ada/releases"},{"date":"2026-10-23T21:00:53+00:00","text":"Repository created","source":"https://github.com/JustAResearcher/gpu-unlock-ada"}],"cite":"gpu-unlock-ada is a GitHub project at https://github.com/JustAResearcher/gpu-unlock-ada written mainly in C, described there as \"GPU tensor unlock source for RTX 4090 and experimental RTX 4070 Ti SUPER; firmware generated locally\".","attribution":"Data from Mid or Real, https://midorreal.com/project/justaresearcher-gpu-unlock-ada, checked 22 November 2026. Attribution with a link is required.","still_real":null}