When a graphics card driver ships with a bug severe enough to stop cooling fans from spinning, a quick patch is obviously welcome. But Nvidia's latest attempt to clean up its recent driver mess appears to have introduced a new problem for owners of high-end GeForce RTX 40 and 50 series cards, one that is quietly eating into the performance headroom many customers paid a premium to access.
Nvidia's GeForce Game Ready driver 595.71 was released earlier this week as a direct replacement for version 595.59, which had to be pulled from circulation after users discovered it could cause GPU fans to stop functioning entirely. The initial release, 595.59, included a serious bug that affected fan monitoring and even stopped fans from working on some GPUs. That is a significant safety concern for hardware drawing hundreds of watts under gaming loads, and Nvidia acted quickly to withdraw it.
The replacement driver, 595.71 WHQL, was released to coincide with the launch of Resident Evil Requiem and is causing significant performance drops on RTX 40 and 50 series GPUs due to GPU core voltage limitations that reduce boost clock speeds and power draw. The problem is not appearing in Nvidia's official patch notes, and the company has offered no public explanation for the behaviour.
The performance implications are measurable and, for enthusiast users, material. One user with an RTX 5080 reported that their GPU used to hit 3,100 to 3,200 MHz with previous drivers and can now only achieve 2,955 MHz with 595.71. Another RTX 5080 owner compared 3DMark scores between the previous 591.86 driver and 595.71, with a 450 MHz GPU overclock applied, and found the newer driver was running the GPU 300 MHz lower and pulling 43 fewer watts, dropping from 403W to 360W. In benchmark testing reported by Tweaktown, performance in the Heaven Benchmark running on a GeForce RTX 5090 dropped from 183 FPS on driver version 591.74 to 144 FPS on 595.71.
The technical mechanism appears to be a hard ceiling on core voltage. At stock values, the GPU's core frequency and voltage behave as expected, with a 150 MHz overclock also proving unproblematic. Attempting to push further, however, results in the voltage inexplicably dropping from around 1.05V to 0.99V, with no way to achieve 3 GHz. Reports of affected cards span the RTX 4090, RTX 4080 Super, and RTX 4070 Ti Super, in addition to RTX 50 series models.
Critically, the issue is not universal. Not all RTX 50 series GPUs appear to be affected; three users with Gigabyte Aorus Master RTX 5090 cards reported having no restrictions whatsoever. Another RTX 5090 owner with a PNY Epic OC variant reported no issues and a maximum overclock of 3,157 MHz, and two RTX 5070 owners, one with an Asus variant and the other with an MSI Gaming Trio OC, also reported no problems. That inconsistency points strongly toward a software bug rather than a deliberate, uniform policy change.
There is, however, a competing theory circulating in hardware communities. Nvidia may have implemented these voltage limitations because of numerous melted connector reports, which would theoretically limit the maximum voltage going through the 16-pin connectors, thereby preventing burning connectors or damaged GPUs. The 12V-2x6 power connector used by high-end Nvidia cards has attracted scrutiny since isolated reports of connector damage emerged in recent months. If Nvidia is using a driver-level voltage cap as a precautionary measure against connector stress, it is doing so without informing customers, which raises its own accountability concerns.
Nvidia has not officially acknowledged the issue, but the artificial voltage limits appear to be a bug rather than an official change, since the patch notes make no mention of any new voltage limits and certain GPU models are unaffected. For affected users, graphics card owners have the option of rolling back to older drivers in the meantime.
The broader picture here is one of a company struggling with the complexity of shipping stable software for an increasingly power-hungry product line. Nvidia's RTX 50 series cards are among the most power-hungry consumer GPUs ever produced, and managing that power delivery safely across dozens of board partner variants is a genuinely difficult engineering problem. From that angle, erring on the side of caution, even silently, is a defensible instinct.
The reasonable concern, though, is transparency. Consumers who spend thousands of dollars on flagship hardware deserve to know when a software update materially changes the performance envelope of their product. A silent voltage cap, whatever its motivation, bypasses that basic expectation. Whether Nvidia issues a corrective release or hotfix driver in the near future, with any further explanation for the issue, remains to be seen. Until then, affected users face an uncomfortable choice between fan reliability and peak performance, two things that should never be in conflict.