The world of extreme overclocking just got a serious jolt. Team OGS has officially shattered previous limits, demonstrating that cutting-edge Blackwell architecture can handle unprecedented speeds. They have successfully pushed a Galax GeForce RTX 5090D HOF OC Lab Edition to an astonishing 4GHz core frequency.
This monumental achievement isn’t just about speed; it’s a testament to meticulous engineering and extreme thermal management. The test, validated through the GPUPI v3.3 32B benchmark, wrapped up in a blistering 35.377 seconds, proving that pushing the limits on these flagship GPUs is achievable.
To achieve this leap, Team OGS employed some serious hardware wizardry. They utilized a GB202-based GPU core paired with memory clocks reaching 1860MHz. Crucially, they managed this extreme performance by leveraging an Elmor External Clock Board set to 28.7MHz, overriding the standard internal reference clock.
Keeping such volatile components from melting required state-of-the-art cooling techniques. The overclocking rig was cooled by liquid nitrogen, utilizing a Bitspower Strata LN2 GPU Pot in tandem with Thermal Grizzly Kryonaut Extreme thermal paste—a pairing designed to handle the intense thermal load generated during maximum stress testing.
The system itself was built on a foundation of top-tier components. Supporting the overclock were an Intel Core i9-14900KF running at 6.0GHz, an Asus ROG Maximus Z790 Apex Encore motherboard, 32GB of Corsair Vengeance DDR5 memory, and a robust Corsair WS3000 power supply.
It is important to note that the team utilized the original Galax RTX 5090D HOF OC Lab Edition variant. This specific model was chosen for its distinctive layout, setting it apart from newer iterations that feature reduced memory configurations and narrower bus architecture.
This success opens up exciting possibilities for future overclockers. The question now shifts to how much further the Blackwell architecture can be pushed under similar extreme conditions, hinting at an exciting new frontier in high-performance computing.
Buy on Amazon
Credit: KitGuru
