NV UEFI Unlocker Opens RTX Tensor Core Throughput in Early Tests
Wccftech reports that the open-source NV UEFI Unlocker can expose previously restricted Tensor Core compute performance on NVIDIA GeForce RTX 20-, 30-, and 40-Series GPUs across UEFI-bootable systems, including Windows. Developed by GitHub user physicsdolphin, it applies temporary changes before the graphics driver loads without permanently flashing VBIOS or patching drivers. Early, hardware-specific tests reported an RTX 2070 rising from 36 to approximately 60–63 TFLOPS in FP32-accumulation matrix multiplication and an RTX 4060 Laptop GPU improving from 29.89 to 49.33 TFLOPS in FP16 compute.
The project offers a potentially simpler, temporary route to greater AI compute throughput on selected older NVIDIA GPUs, but Wccftech says the results are preliminary and equivalent real-world AI or gaming gains have not been demonstrated.
New NVIDIA UEFI GPU Unlocker Brings Up To 75% Higher AI Compute Performance To RTX 20, 30 & 40 Series GPUs On Windows
Sebastian Castellanos•Oct 31, 2026 at 11:00am EDT
Add to Preferred Sources
Some source content is omitted. Continue reading on the original site.
#NV UEFI Unlocker#NVIDIA#GeForce RTX 20 Series#GeForce RTX 30 Series#GeForce RTX 40 Series#RTX 2070#RTX 4060 Laptop GPU