From NVIDIA, we learn that the brand has launched its RTX Pro 6000 (tested here), a model aimed at the professional sector. This card offers enhanced features compared to the RTX 5090, but its results in ray tracing tests are not very convincing. Differences are only a few percent.
RTX Pro 6000: unconvincing test results?
On paper, however, the RTX Pro 6000 offers a reference card with enhanced features compared to the RTX 5090. Firstly, there’s three times as much memory: 96 GB instead of 32 GB for the 5090. Secondly, the GPU is more complete, with a total of 188 SM (24,064 cuda cores), compared to 170 SM (21,760 cuda cores) for the 5090. In terms of performance, the RTX Pro 6000 boasts a 19% increase in FP32, RT and AI performance over the RTX 5090.
Otherwise, the card itself exploits a similar design to the RTX 5090, with a dual-slot heatsink and through-airflow. What’s more, it boasts a TDP of 600W while being powered by a single 12V-2×6… Let’s just hope that on this version NVIDIA has integrated the necessary measures to prevent the connector from catching fire.
In actual fact, this pro version outperforms the 5090, and that’s not a marginal point. Nevertheless, in actual cases, on ray tracing benchmarks, the card shows no convincing difference in performance. In fact, on GameTechBench, in the 2160p ray tracing test, the new Red card shows a 5% improvement in performance, while it lags slightly behind in 1440p. On the path tracing test, the gap is even more tenuous: 2% in favor of the RTX Pro 6000.
Despite theoretical performance gains of 19%, the benchmark gap with the 5090 is very narrow. Wccftech also points out that these small differences may be due to under-utilization of the hardware or to graphics drivers that are not yet fully mature.
