A new open-source UEFI tool boosts tensor performance on RTX 20, 30, and 40-series desktop and laptop GPUs running Windows.reported The tool, named nv-uefi-unlocker, allows users to bypass the manufacturer’s half-speed AI throttle without modifying video BIOS or installing patched drivers.A new free tool lets users get more AI speed from their Nvidia graphics cards without changing the card's permanent settings.The nv-uefi-unlocker project utilizes a V67 payload to bypass write-protected registers, achieving full tensor clock rates on Windows.
Evidence: reported
Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptopsaifleet-tomshardware
The claim
The project claims to bypass Nvidia's tensor limiter by employing a community V67 payload that includes read-back verification.reported This method deceives the signed Booter firmware into unlocking the write protection on two speed-selector registers, allowing the tensor units to operate at their maximum speed. The developer’s own laptop RTX 4060 reportedly achieved a 65% increase in FP16 math performance after using the tool.reported
Evidence: reported
Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptopsaifleet-tomshardware
Evidence: reported
Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptopsaifleet-tomshardware
The response
The distribution provides solely scripts and user-compiled binaries, omitting Nvidia firmware from the download. Instead, scripts grab public 610.43.03 driver packages, verify them, and rebuild firmware files locally. The project acknowledges Zagitta’s gpu-unlock as its porting source and incorporates other community contributions, such as a previous Windows EFI tool.
The practical upshot
Users must download the release, run the script, and copy rebuilt files plus a BOOTX64.EFI to a FAT32 USB stick.reported The unlocker runs from a one-time boot menu within 30 seconds.reported Since the tool functions in a firmware stage prior to the Nvidia driver loading, the unlock is not permanent and must be run again at each startup.
Evidence: reported
Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptopsaifleet-tomshardware
Evidence: reported
Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptopsaifleet-tomshardware
Claims versus evidence
The performance gain is evidenced by a single developer benchmark on an RTX 4060 laptop.reported The mechanism relies on tricking firmware write locks, a method previously experimental on Windows. The addition of read-back verification addresses prior safety concerns, but the tool remains a community project without official vendor support or independent security audits.
Evidence: reported
Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptopsaifleet-tomshardware
Why it matters
- Product teams testing AI models on consumer hardware can potentially access higher throughput without purchasing data-center GPUs.
- The non-persistent nature of the unlock means testing environments must handle repetitive boot-time configuration, impacting workflow stability.
- Reliance on specific driver versions (610.43.03) creates compatibility constraints for teams managing diverse hardware fleets.
- The requirement to disable Secure Boot may conflict with enterprise security policies, limiting deployment in regulated environments.
Topics: Nvidia, GPUs, AI performance, UEFI, open-source
Updated
Then / now / next
Unlocking tensor performance primarily required manual Linux modules with experimental Windows support.
A unified UEFI tool allows Windows users to lift caps via bootable USB without BIOS flashing.
Monitor whether Nvidia patches the V67 payload vulnerability or updates driver checks to block this method.
Story timeline
- Then Unlocking tensor performance primarily required manual Linux modules with experimental Windows support.
- 01 Nov 2026 Open-source USB tool unlocks Nvidia's half-speed GPU AI limiter on Windows — lifts GPU's FP16 math performance up to 65%, UEFI unlocker covers RTX 20, 30, and 40-series desktops and laptops
- Now A unified UEFI tool allows Windows users to lift caps via bootable USB without BIOS flashing.
- Next Monitor whether Nvidia patches the V67 payload vulnerability or updates driver checks to block this method.
The strongest doubt
Counterpoint from the sources
The unlock requires disabling Secure Boot or trusting self-signed certificates, which may violate organizational security policies.
For readers
Developers
- Can test AI models on consumer hardware with higher throughput.
- Must manage repetitive boot-time configuration for non-persistent unlocks.
Consumers
- Can potentially improve AI performance on existing RTX cards.
- Requires technical steps including USB booting and firmware rebuilding.
Researchers
- Access to higher tensor rates may accelerate local model inference experiments.
- Reliance on community tools introduces unverified security and stability risks.
Ask the article
Local keyword search of this article; no generated answers.Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Questions readers ask
Does this tool permanently modify the GPU hardware?
The unlock does not persist through power loss and requires re-execution at every boot.
Which GPU series are supported by this utility?
The utility supports desktop and laptop graphics cards from Nvidia’s RTX 20, 30, and 40 series.
Is Secure Boot required for this tool to function?
Secure Boot must be disabled or configured to trust a self-signed EFI.
Connected stories
Reading receipt
- The nv-uefi-unlocker project lifts FP16 performance by up to 65% on RTX 20, 30, and 40-series cards.
- The utility runs from a bootable USB stick prior to the operating system loading, eliminating the need for permanent firmware modifications.
- Usage requires disabling Secure Boot or trusting a self-signed EFI certificate on UEFI systems.