Argomenti trattati
Nvidia’s new graphics card emerges in benchmarks
Nvidia has yet to officially announce its Blackwell-based graphics boards designed for professional visualization (ProViz) applications. However, early testing is underway, with results surfacing on Geekbench. An OpenCL benchmark result for Nvidia’s anticipated RTX Pro 6000 Blackwell Workstation Edition graphics card has come to light, confirming several rumored specifications while raising intriguing questions about its performance capabilities.
Specifications of the RTX Pro 6000 Blackwell
The RTX Pro 6000 Blackwell Workstation Edition is built around the GB202 graphics processing unit (GPU), featuring a staggering 24,064 CUDA cores distributed across 188 streaming multiprocessors. This GPU reportedly operates at clock speeds reaching up to 2,617 MHz and is equipped with 96 GB of memory, which includes error-correcting code (ECC). In comparison, Nvidia’s current flagship for gaming, the GeForce RTX 5090, incorporates the same GB202 GPU but with 21,760 CUDA cores, clocking in at a maximum of 2,410 MHz and housing 32 GB of GDDR7 memory.
Benchmark results compared
At first glance, the RTX Pro 6000 Blackwell Workstation Edition seems poised to outperform its gaming counterpart. However, the initial benchmark scores tell a different story. The RTX Pro 6000 achieved a score of 368,219 in the GB6 6.4.0 OpenCL benchmark, while the GeForce RTX 5090 scored approximately 376,858. Although this performance difference of 2.3% may seem minor, it becomes more significant when considering the substantial hardware variations between these two boards.
Detailed performance metrics
- OpenCL Score: RTX Pro 6000 – 368,219 | GeForce RTX 5090 – 376,858
- Background Blur: RTX Pro 6000 – 63,762 | GeForce RTX 5090 – 75,075
- Face Detection: RTX Pro 6000 – 60,254 | GeForce RTX 5090 – 73,968
- Horizon Detection: RTX Pro 6000 – 684,753 | GeForce RTX 5090 – 637,294
- Edge Detection: RTX Pro 6000 – 864,739 | GeForce RTX 5090 – 838,261
- Gaussian Blur: RTX Pro 6000 – 832,815 | GeForce RTX 5090 – 795,994
Performance limitations and considerations
It’s important to note that the RTX Pro 6000 is a pre-release product, and the performance metrics could fluctuate as it approaches its final launch with fully optimized drivers. Furthermore, the total graphics power (TGP) for the RTX Pro 6000 may be capped, limiting its compute performance compared to the GeForce RTX 5090, which can reach up to 575W. Additionally, current drivers restrict the memory available to OpenCL applications to just 24 GB, despite the card having 96 GB of onboard memory. This limitation indicates that the driver is not yet finalized.
Future prospects for Nvidia’s pro graphics line
Based on prior reports, the RTX Pro 6000 Blackwell is not positioning itself as Nvidia’s premier ProViz offering. The top-tier model is expected to be labeled as the RTX Pro 6000 X Blackwell, suggesting that there are even more powerful variants on the horizon. This anticipation adds to the excitement surrounding Nvidia’s graphics technology developments.
Stay tuned to Tom’s Hardware for the latest updates, deep analyses, and comprehensive reviews in the tech sphere. Remember to click the Follow button to keep abreast of all the advancements.