Nvidia's RTX Pro 6000 Blackwell Outperforms Stock RTX 5090 by 10-15% in Benchmarks

Nvidia's latest workstation GPU, the RTX Pro 6000 Blackwell, has been benchmarked by two independent sources, shedding light on its performance relative to the company’s current top-tier consumer GPU, the RTX 5090. Results from Privaterbok on Reddit and Level1Techs offer a comprehensive look at how the RTX Pro 6000 stacks up in various scenarios. Privaterbok conducted a series of tests using popular benchmarking tools like 3DMark and Geekbench 6. The GPU was evaluated in several sub-tests, including TimeSpy, TimeSpy Extreme, Steel Nomad, and Port Royal. Additionally, both the stock and overclocked performance of the RTX Pro 6000 were compared to the best publicly available scores for the RTX 5090, which often include highly overclocked configurations. 3DMark Benchmarks TimeSpy: Stock RTX Pro 6000: 51,776 Stock RTX 5090: 57,591 Overclocked RTX 5090: 54,300 TimeSpy Extreme: Stock RTX Pro 6000: 28,009 Stock RTX 5090: 29,494 Overclocked RTX 5090: 30,019 Steel Nomad (DX12) with Overclock: RTX Pro 6000: 16,804 Overclocked RTX 5090: 18,325 Port Royal with Overclock: RTX Pro 6000: 42,374 Overclocked RTX 5090: 46,715 On average, the stock RTX Pro 6000 is about 10-15% slower than the overclocked RTX 5090. However, when the RTX Pro 6000 is itself overclocked, the performance gap narrows significantly, with the two GPUs becoming more evenly matched. Notably, the RTX Pro 6000 edges out the overclocked RTX 5090 by 1% in TimeSpy Extreme but trails by 5% in TimeSpy. Geekbench 6 Benchmarks OpenCL: RTX Pro 6000: 434,166 RTX 5090: 375,423 Vulkan: RTX Pro 6000: 431,723 RTX 5090: 395,146 In Geekbench 6, the RTX Pro 6000 demonstrates a significant advantage over the RTX 5090, achieving 9-15% higher scores in both OpenCL and Vulkan tests. These benchmarks provide a clearer, more standardized comparison as they are officially validated and less prone to variability from different setups and overclocks. Gaming Performance Level1Techs tested the RTX Pro 6000 in "Cyberpunk 2077," a demanding game known for pushing GPU limits. The workstation GPU was tested under ultra settings at 4K resolution, with DLSS set to performance mode (auto) and path tracing enabled or disabled. Without Path Tracing: Frames per Second (FPS): 127.02 With Path Tracing: FPS: 92.54 While the RTX Pro 6000 delivered impressive frame rates, especially considering its primary purpose is for professional tasks, these scores offer insights into its potential for high-end gaming. Comparison Summary These benchmarks highlight the RTX Pro 6000's strong performance, particularly in workloads where its specialized hardware and larger memory capacity come into play. The stock version of the Pro 6000 generally performs 10-15% faster than the stock RTX 5090, with this gap narrowing when the consumer GPU is overclocked. The Geekbench 6 results, which are more consistent due to their official validation, indicate that the RTX Pro 6000 can achieve a more substantial lead in compute-intensive tasks. Technical Specifications The RTX Pro 6000, like its predecessor, is built on the Blackwell architecture. It features 24,064 CUDA cores, a 10% increase over the RTX 5090. The GPU is equipped with 96GB of GDDR7 memory, providing ample resources for the most demanding professional applications. The price tag for the RTX Pro 6000 is estimated around $8,000, reflecting its high-end positioning. Conclusion Overall, the RTX Pro 6000 Blackwell stands as a formidable workstation GPU, offering significant advantages in compute tasks and professional applications. While it may not completely dominate the best overclocked consumer GPUs, its performance and specialized features make it a compelling choice for professionals who require top-tier hardware for their work.