RTX 3090 unlock exploit boosts PyTorch FP16 throughput 75→135 TFLOPS at same 360W
lxfater · x · 2026-10-24
A new open-source Linux kernel module, gpu-unlock, exploits a GPU boot-firmware vulnerability to write speed-selector registers before the stock NVIDIA driver loads, unlocking the FP16/BF16 instruction issue rate on RTX 3090 to professional-card levels. Under the same 360W power wall, KBlueleaf measured PyTorch throughput jumping from 75 to 135 TFLOPS (80% gain). The module combines the CMP unlocker's V67/PLM technique with UEFI-style pre-driver timing, avoiding manual repatching on every driver update. A related branch reportedly lifts 4090 instruction throughput from 165 to 331 TFLOPS, nearly doubling it.
More from Infra
- Used NVIDIA DGX Spark reselling at $6,200, well above its ~$4,000 retail price — AIFlow_ML · 2026-10-25
- SGLang Matches DeepInfer on Prefill, Adds 7.4% Decode Throughput on 64×GB300 — BanghuaZ · 2026-10-25
- Poldi: A Fully Offline Local Voice AI Assistant That Runs on 8GB VRAM — rhinodevil · 2026-10-25
- kalomaze says he no longer has meaningful GPU access, back to toy baselines — kalomaze · 2026-10-25
- Is the explosion of LLM inference engines healthy competition or fragmentation? — Anstellos · 2026-10-25
- Jev ecosystem explodes: 5 open-source apps from a 7.1s flight-search agent to SQL-in-English — udmrzn · 2026-10-25