{"name":"turing-unlock","repo":"nlzy/turing-unlock","url":"https://midorreal.com/project/nlzy-turing-unlock","repository":"https://github.com/nlzy/turing-unlock","description":"Unlock full speed FP16 tensor core of NVIDIA Turing GPUs","facts":{"language":{"value":"C","source":"https://github.com/nlzy/turing-unlock","checked":"22 Nov 2026"},"releases":{"value":"0","source":"https://github.com/nlzy/turing-unlock/releases","checked":"20 Nov 2026"},"contributors":{"value":"2","source":"https://github.com/nlzy/turing-unlock/graphs/contributors","checked":"20 Nov 2026"},"archived":{"value":"no","source":"https://github.com/nlzy/turing-unlock","checked":"22 Nov 2026"}},"stars":374.0,"writeup":{"what_it_is":"A Linux kernel module that unlocks the full speed FP16 tensor core of NVIDIA TU102 (Turing) GPUs. It writes two speed selectors and releases an ACR mutex during GPU boot, then returns through the normal Booter finalize path.","audience":"Linux users with NVIDIA Turing TU102 GPUs who want to run the FP16 tensor core at full speed","claims":[{"kind":"specific","claim":"Tested on the Dell RTX 2080 Ti 22GB with NVIDIA driver 575.51.03","excerpt":"Tested on the Dell RTX 2080 Ti 22GB (`10de:1e07`) with NVIDIA driver 575.51.03.","status":"not_checked"},{"kind":"specific","claim":"Both the proprietary driver and the open driver work","excerpt":"Both the proprietary driver and the open driver work.","status":"not_checked"},{"kind":"specific","claim":"Reports a benchmark of cuBALS FP16 matrix multiply with FP32 accumulate on RTX 2080 Ti, with 2048 size rising from 97.9% to 160.6% of peak","excerpt":"TFLOPS of cuBALS FP16 matrix multiply with FP32 Accumulate on RTX 2080 Ti @250W","status":"not_checked"},{"kind":"specific","claim":"Packages build and install via make targets for Arch, Debian/Ubuntu, and Fedora/RHEL","excerpt":"`make install` finds your distribution. It builds and installs the packages.","status":"not_checked"},{"kind":"specific","claim":"Module does nothing by default and must be enabled with a kernel command line option","excerpt":"The module does nothing by default. Add one option to the kernel command line:","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-10-27"},"why_now":null,"owner_supplied":null,"score":{"verdict":"too_early","hype":51,"reality":21,"gap":30,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T08:00:00+00:00","verdict_source":"rule"},"early_read":{"estimate":true,"lean":"too_close","gap_low":-8,"gap_high":33,"rule":"early-2.0"},"changes":[{"date":"2026-10-24T15:18:08+00:00","text":"Repository created","source":"https://github.com/nlzy/turing-unlock"}],"cite":"turing-unlock is a GitHub project at https://github.com/nlzy/turing-unlock written mainly in C, described there as \"Unlock full speed FP16 tensor core of NVIDIA Turing GPUs\".","attribution":"Data from Mid or Real, https://midorreal.com/project/nlzy-turing-unlock, checked 22 November 2026. Attribution with a link is required.","still_real":null}