Skip to main content

Nvidia Serves Up Low-Profile RTX A2000 GPU for Slim Workstations

Nvidia
(Image credit: Nvidia)

Nvidia has announced a new RTX A2000 desktop GPU entering its A-series workstation lineup, featuring 6GB of memory and a low-profile form factor. This GPU will be Nvidia's new mid-range offering for the workstation market, packing the latest Ampere architecture with 2nd gen RT cores and 3rd gen Tensor cores. This new A2000 is not to be confused with the already released mobile variant.

We don't have a lot of information about this new GPU, so we can only speculate as to its performance. All we know is the name, and that the A2000 will have 6GB of GDDR6 memory. The 6GB frame buffer likely means we're looking at our first workstation GPU based on Nvidia's GA106 die, the same die as the RTX 3060 12GB. If true, we could at a potential configuration of up to 3840 CUDA cores with a theoretical maximum power output of 170W.

That might make the A2000 slightly more powerful than the RTX 3060, since that GPU has a cut-down GA106 core with just 3584 CUDA cores enabled. However, the RTX 3060 has double the VRAM capacity, which is an odd state of affairs. Normally, workstation GPUs have more memory than their desktop equivalents, not less. Power use on Nvidia's workstation cards are normally less than that of its gaming counterparts as well, so don't be surprised if the TDP gets cut to 150W or even 120W.

That's entirely likely, considering Nvidia notes that the A2000 desktop card will be equipped with a low-profile dual-slot GPU for 3D creators, giving us more reason to believe the A2000 will have a low wattage TDP. That also makes the A2000 Nvidia's first ever low-profile Ampere solution, giving the card an incredibly small form factor to fit in low-profile computer chassis. That's something you don't see as much in the desktop world, but it's more common with pre-built entry-level workstations.

Nvidia hasn't disclosed pricing, but the RTX A2000 will be available starting in October 2021.

  • JWNoctis
    FWIW STH quoted 3328 CUDA/104 Tensor/26 RT cores and 70W TDP for this card.

    Probably much faster in FP64 math than absolutely anything in the consumer lineup, still.
    Reply