Modders unlock hidden AI power on Nvidia RTX 20, 30, and 40 series GPUs, nearly doubling some benchmarks
Summary
Modders have unlocked hidden AI processing capabilities on Nvidia RTX 20, 30, and 40 series GPUs, nearly doubling performance in some benchmarks by accessing disabled hardware resources.
Similar Articles
DLSS 5 leaked and modders are putting Nvidia’s AI effects on everything
An unofficial version of Nvidia's DLSS 5 AI upscaling technology has leaked and is being modded into various games, demonstrating enhanced visual effects but also causing performance drops.
RTX Pro 4500 Blackwell Performance Numbers
A user shares performance benchmarks comparing the Nvidia RTX Pro 4500 Blackwell 32GB GPU against the RTX 5060 Ti 16GB for AI inference, showing 1.6-6x speed improvements depending on model size and quantization.
Modify and load custom VBIOS on Windows and Linux, on Blackwell 2.0 (RTX 50 Series)/Ada(RTX 40 Series)/Ampere (RTX 30 Series) without flasing, for example to lower power limits (and which cards were tested)
Three new tools let users modify and load custom VBIOS images on NVIDIA RTX 30/40/50-series GPUs on Windows and Linux without flashing, enabling higher power limits, unlocked voltage, and larger overclocking offsets while keeping the card's original firmware intact.
Project Blackwell: It Will Work, Eventually — Making an RTX Pro 6000 Run in a Dell R730 at 650K Context
A developer documents the extensive hardware and firmware hacking required to run an NVIDIA RTX Pro 6000 Blackwell GPU in a legacy Dell PowerEdge R730 server, achieving 650K context length for local AI inference.
@no_stp_on_snek: ohlala, what do we have here?
A Linux kernel module hack reportedly unlocks 1.9x faster BF16 tensor math on NVIDIA RTX 3090 cards, boosting BF16 TFLOPS from 82 to 155 and cuBLAS GEMM performance, since NVIDIA ships the cards with fp32-accumulate tensor math at half rate.