In the past couple of years, GPU crypto mining became unprofitable and essentially got replaced with the AI craze. To help prevent more gaming GPU shortages, Nvidia started providing a wider variety of workstation GPUs for both desktop and mobile use, so now AIB partners sell all sort of cards with some confusing name schemes. Unlike the current gaming lineup that features cards from the RTX 4000 series, the workstation lineup includes cards ranging from RTX 2000 all the way up to RTX 6000. Granted, Nvidia appends the ADA denominator at the end, but this does not help too much. Even more confusing, some cards like the RTX 4000 ADA and now the RTX 2000 ADA get separate variants with lower TGP.
The newest member of Nvidia’s desktop workstation GPU family is the RTX 2000E ADA, which technically is the same as the existing entry-level RTX 2000 ADA when it comes to CUDA cores, but it comes in a more compact form-factor with slightly lower TGP. While the RTX 2000 ADA comes with a dual-slot low-profile blower fan design and 70 W TGP, the new RTX 2000E ADA features a single-slot design with 50 W TGP and no power connector.
Otherwise, both cards are based on the same AD107 cut-down variant of the AD102, with 2816 CUDA cores, 16 GB GDDR6 VRAM and 128-bit bus width. Due to the lower TGP, the RTX 2000E is slightly slower with 8.9 instead of 12 TFLOPS, yet the Tensor performance remains unchanged at 71 TFLOPS. The new cards also offer 4x miniDP 1.4a video outputs.
Thanks to the slimmer design that only requires properly powered PCIe 4.0 slots as well as the 16 GB VRAM that are a necessity for LLM and generative AI, more RTX 2000E cards can be connected per motherboard, leading to lower deployment prices for AI training purposes.
For now, this new RTX 2000E ADA variant does not appear on the official Nvidia site, and the $849 MSRP reported by shi.com for the cards produced by PNY is $220 higher than that of the RTX 2000 ADA.
The above is the detailed content of Nvidia RTX 2000E ADA single-slot desktop workstation card quietly introduced as energy efficient AI processor. For more information, please follow other related articles on the PHP Chinese website!