Modders Unlock GeForce RTX 20, 30 And 40 GPUs For Up To 2X Compute Gains

Modders have successfully unlocked the artificial compute limits placed upon consumer-oriented NVIDIA GeForce GPUs in order to make them nearly as capable as professional NVIDIA RTX GPUs for relevant AI and compute workloads. Specifically, modders have pulled this off with NVIDIA GeForce RTX 20 Series, RTX 30 Series, and RTX 40 Series GPUs.

Gamers shouldn't get too excited about this one. While more GPU compute is definitely good to have, these mods aren't really going to help boost your in-game framerates. Instead, it's more about AI and GPU compute workloads that people previously needed to buy RTX Pro GPUs for.

Modders have unlocked hidden AI compute performance on NVIDIA’s RTX 20, 30 and 40-series GPUs. The modifications remove software-level limits on certain Tensor Core operations, without changing the hardware or flashing a new VBIOS. - RTX 4090: 165.6 → 331.6 TFLOPS FP16 Tensor Core throughput -RTX 3090 Ti: 87.5 → 157.4 TFLOPS BF16 performance - RTX 2080 Ti: 59.9 → 111 TFLOPS FP16→FP32 performance - RTX 3090: 73.0 → 122.8 TFLOPS in a PyTorch GEMM test The gains are mainly relevant to AI and compute workloads, not gaming.

— Pirat_Nation 🔴 (@Pirat_Nation) October 25, 2026

But for those tasks, we're looking at transformative differences in GPU compute performance, ranging from a roughly 1.8x to 2x boost for no extra cost. While this doesn't fully close the gap with enterprise NVIDIA RTX GPUs due to those also having far more VRAM and having full enterprise software support, it does lower the barrier of entry significantly. Those interested in running local AI, for example, should be intrigued by this news.

NVIDIA probably isnt thrilled with this news, but the cat is put of the bag at this point. If you are worried that the files may get pulled down or just want to test these mods ASAP (thanks for the spot, TechPowerUp,)  here are the links for the NVIDIA GeForce RTX 20 Series, RTX 30 Series, and RTX 40 Series, respectively.