The RTX 5090 doesn’t feature the largest Blackwell GPU in the range, nor does it boast the largest memory configuration in the Nvidia catalog. No, that title goes to the RTX Pro 6000 Blackwell, a card designed for the prosumer sector. We’re talking about a card with 2,304 more cuda cores than the RTX 5090… And a lot more VRAM!
RTX Pro 6000: a pro card based on the RTX 5090!
In this case, it’s the Leadtek website that has let the cat out of the bag about NVIDIA working on this card. Inspecting the site’s web page, the card’s name stands out clearly. In short, we’re in for a real monster with this card, as it incorporates 2,304 more cuda cores than the RTX 5090. As for memory capacity, we’re talking about 96 GB of VRAM, triple that of the 5090, which pales in comparison.
Despite this, this model should have a TDP of 600W, equivalent to the RTX 5090. VideoCardz points out that it’s rare to see a card in this range with a TDP similar to its gaming equivalent.
Back to the memory: not only is memory capacity tripled, it’s also of the ECC type. In other words, it incorporates error correction mechanisms. Given that it’s intended for a professional environment, it would be a shame for the card to generate calculation errors, wouldn’t it?
Otherwise, with this model, NVIDIA would be abandoning the blower-type cooling system as was the case with the RTX 6000 Ada. Here, we find a heatsink equivalent to the RTX 5090, but in all-black with a slight design change. Well, in the pro world, we don’t really care about looks, what counts is that it cools. And given the efficiency of this cooling system on the 5090, we’re not too worried… But what about its performance in multi-GPU systems?
