The relevant parameter configurations of NVIDIA’s new generation workstation graphics card RTXPRO6000 have finally surfaced.This graphics card uses the Blackwell architecture and is equipped with 96GB of GDDR7 video memory with a memory width of 512bit, which also means that it will be the first desktop graphics card to use a 3GB GDDR7 module.
Its CUDA core count is as high as 24,064. Compared with the previous generation RTX6000Ada graphics card, the number of CUDA cores has increased by 32%, and the video memory capacity has doubled.
In addition, the power consumption of RTXPRO6000 Blackwell has also been significantly increased, with its TGP reaching 600W. This may be the first time that NVIDIA's workstation graphics card consumes more power than game graphics cards.
However, dual fans are still used for cooling, similar to the dual-flow cooling solution of the RTX5090. In addition, the graphics card also uses a PCIe5.0x16 interface and is equipped with four DisplayPort2.1 interfaces.
It is worth noting that the video memory of this graphics card supports ECC (error correction) function, which is crucial for professional application scenarios that require high reliability and stability.
The detailed specifications of RTXPRO6000 Blackwell have not yet been fully confirmed, but it is expected to be officially released at the GTC2025 conference from March 17th to 20th.