Skip to content
Live
Trending storyPast story

Open source project GPU Unlock releases hidden computing power of RTX 30 series

1 reports1 outlets25 days agoUpdate

The story so far

AI overview

GitHub user Zagitta has open-sourced the GPU Unlock project, which can remove the instruction emission rate limit for Ampere architecture consumer GPUs in FP16/BF16 tensor core operations without modifying drivers or flashing VBIOS. The project labels support for GA102, GA103, GA104, GA106 models, covering most RTX 30 series models, but not all GPUs are compatible, and performance improvements are not guaranteed. According to benchmark tests in the project's GitHub repository, the RTX 3090 Ti with a power limit of 450W shows significant performance improvement in FP32 dense matrix multiplication tests. A Reddit user with an RTX 3090 Ti measured a 58.6% throughput increase for MOSS-Image SFT BF16 in ComfyUI, a 57.2% increase for Qwen-Image 2.1 Turbo BF16, and a 56.1% increase for Z-Image Turbo BF16. This solution does not increase memory bandwidth, does not unlock more CUDA cores, and cannot automatically speed up all loads; the project is still in an experimental phase.

AI-written from the reports · Updated 25 days ago

Coverage timeline

Follow the coverage to see every side of the story.

25 Oct
  1. ITHome · AIFeatured
    The open-source project GPU Unlock can release the hidden AI computing power of RTX 30 series graphics cards, with testing showing an increase of up to 60%

    The open-source project GPU Unlock, developed by GitHub user Zagitta, can remove the instruction emission rate limit of Ampere architecture consumer-grade GPUs in FP16/BF16 tensor core operations without modifying the driver or flashing VBIOS.