Nvidia Amplifies Performance of Its Tiniest Workstation Graphics Processing Units with Blackwell Upgrade
Nvidia Unveils RTX PRO 4000 SFF: A Significant Leap in AI and Ray-Tracing Performance
At the Siggraph conference in Vancouver, British Columbia, Nvidia has introduced two new GPUs: the RTX PRO 4000 SFF and the RTX PRO 2000. This article will focus on the impressive performance improvements offered by the RTX PRO 4000 SFF.
The RTX PRO 4000 SFF is a significant upgrade from its predecessor, the RTX PRO 2000. With 280 tensor cores, it delivers an impressive 770 teraFLOPS of FP4 performance, a 2.5× increase in AI performance, and a 1.7× improvement in ray-tracing performance compared to the RTX PRO 2000.
The RTX PRO 4000 SFF is based on Nvidia’s latest Blackwell architecture, featuring fourth-generation RT cores and fifth-generation Tensor cores. This results in a more efficient CUDA core count and improved core architecture, although the exact number of CUDA cores is not publicly disclosed.
In terms of relative performance, the RTX PRO 4000 SFF offers up to 2.5× higher AI performance, 1.7× higher ray-tracing performance, and 1.5× more memory bandwidth compared to the RTX PRO 2000. Both GPUs maintain the same 70-watt power envelope but in a compact form factor.
| Feature | RTX PRO 4000 SFF | RTX PRO 2000 | |--------------------------|----------------------------|------------------------------| | AI Performance | Up to 2.5× higher | Moderate boost over previous gen (up to 1.4–2.3×) | | Ray-Tracing Performance | 1.7× higher | Lower relative to 4000 SFF | | Memory Bandwidth | 1.5× higher | Lower | | Power Consumption | 70 Watts max | 70 Watts max | | GPU Cores (Tensor & RT) | 4th-gen RT cores, 5th-gen Tensor cores | Same generation cores but fewer or less efficient |
The RTX PRO 2000 focuses on mainstream design and AI workflows, offering a 1.6× improvement in 3D modeling, 1.4× higher performance in CAD, and 1.6× quicker rendering over its previous generation. However, these gains are smaller than those of the RTX PRO 4000 SFF.
The RTX PRO 4000 SFF is equipped with 24GB of GDDR7 memory, offering a bandwidth of 432GB/s. This allows it to process tokens in large language models like OpenAI's gpt-oss-20b roughly 54 percent faster than Nvidia's last offering.
In summary, the RTX PRO 4000 SFF delivers considerably stronger raw AI and ray-tracing throughput, better bandwidth, and overall superior performance compared to the RTX PRO 2000, while both maintain a compact design and 70W power usage. The RTX PRO 4000 SFF is set to make a significant impact in the realm of AI and ray-tracing performance.
