GoKawiilFirmware hack unlocks full tensor speed on Nvidia RTX 20/30/40 GPUs
Modders have released firmware exploits that remove Nvidia's artificial half-speed limit on FP32-accumulate matrix math for RTX 20, 30, and 40-series GeForce cards, according to VideoCardz. The unlock lets the RTX 4090 reach double its normal tensor rate and boosts the RTX 3090 Ti's matrix math by up to 80%, though gains vary by card and chatbot token speeds stay roughly the same since those depend on memory bandwidth rather than compute. Three separate projects—gpu-unlock, gpu-unlock-ada, and turing-unlock—provide unofficial Linux kernel modules covering different GPU generations, with experimental Windows builds also in development.
GoKawiil's interpretation of the reporting above, not reported fact.
The throttle exists on consumer GeForce cards but not on Nvidia's equivalent Turing and Ampere professional cards, so the hack effectively narrows a software-enforced gap Nvidia uses to differentiate consumer and professional product tiers. For users running local AI workloads like prompt processing or image generation, this could mean meaningfully faster performance without new hardware, though real-world speedups will likely fall short of the doubled peak figures. Because the fix relies on unofficial firmware modifications, it may carry stability or warranty risks that official drivers would not.
- Modders removed Nvidia's half-speed FP32-accumulate limit on RTX 20, 30, and 40-series consumer GPUs via firmware exploits.
- Gains vary widely by card and workload: the RTX 4090 can double peak tensor rate, the RTX 3090 Ti gains up to 80% in matrix math, while chatbot token speeds are largely unaffected.
- Three community projects (gpu-unlock, gpu-unlock-ada, turing-unlock) offer unofficial Linux support, with experimental Windows builds also underway.
Nvidia GeForce RTX 4090 GPU — If you're diving into AI workloads and matrix-math tinkering, the RTX 4090 is the flagship card at the center of these firmware unlock experiments. It already offers huge tensor throughput out of the box, and enthusiasts are showing it can go even further with these mods. A great pick for anyone wanting the best stock performance before even considering hacks.
See Nvidia GeForce RTX 4090 GPU on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: tomshardware.com — Shane Downing, 2026-10-26
Published there as: “Nvidia RTX 20, 30, and 40-series GPUs hacked to remove a half-speed AI throttle”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.