Now that NVIDIA’s second GTC occasion of the 12 months has wrapped up, we’ve lastly gotten an opportunity to comply with up with NVIDIA on final week’s announcement of their RTX A6000 video card, and what which means for the Quadro model. In quick, NVIDIA has confirmed that the Quadro model goes away for positive, and as we suspected, it’s largely because of the overlap between graphics and compute.
As a fast refresher, final week NVIDIA launched their new skilled visualization-focused video card, the RTX A6000. Based on the brand new GA102 GPU, the cardboard ticks all of the bins for a high-end, pro-grade video card; and beneath regular circumstances, it might be a part of NVIDIA’s Quadro household of merchandise. However the cardboard was notably excluded from the Quadro household in one thing of a last-minute change. At the time it wasn’t clear simply what this meant for the Quadro model as a complete, however now that GTC has wrapped we’ve been given some higher insights into what’s occurring.
First and foremost, NVIDIA has confirmed that the Quadro model is being retired, or “streamlined” as the corporate calls it. Similar to the Tesla model a few years again, the model is ready to be slowly retired from the market, as new skilled visualization playing cards are launched with out the Quadro branding. Going ahead, all of those playing cards can be given brand-less names, such because the “NVIDIA RTX A6000” and “NVIDIA A40”.
The extra fascinating facet to this alteration is why: why would NVIDIA retire one among its oldest video card manufacturers after so lengthy? After all, the marketplace for professional playing cards isn’t going away, and it stays a tidy, worthwhile enterprise for NVIDIA. At the time we suspected that this has to do with the growing overlap in NVIDIA’s product strains between skilled visualization playing cards and compute playing cards, and the corporate has since confirmed that our hunch was right.
As NVIDIA has continued to develop into the compute market, their skilled visualization (ProViz) and compute merchandise have more and more overlapped when it comes to options and pricing. As NVIDIA already prices “full” worth for each their compute and ProViz playing cards, there are little-if-any characteristic variations between the 2: desktop ProViz playing cards have the identical entry to compute options as compute playing cards. And compute playing cards, although virtually solely server-mounted, might be provisioned as a digital ProViz card as properly.
One of the results of which has been that NVIDIA’s personal messaging on what playing cards can do what duties has change into unfocused, by no means thoughts probably complicated prospects. If you want an actively-cooled desktop card for working neural community prototyping, for instance, what card do you purchase? Previously it was the Quadro card, even if it was a ProViz half. Similarly, the ex-Tesla V100 makes an excellent half for provisioning a digital Quadro occasion, despite the fact that it’s not a Quadro half.
As a consequence, NVIDIA has opted to go the route of basically merging their compute and ProViz {hardware} lineups in an effort to simplify their choices. NVIDIA needs for there to be a single model – NVIDIA – which covers each markets, reflecting the pliability of their playing cards and (largely) eliminating questions over which playing cards can be utilized for graphics or compute. At the identical time, this additionally permits NVIDIA to cut back the variety of {hardware} SKUs supplied, as they now not want overlapping merchandise on the fringes of those markets.
Ultimately, the marketplace for ProViz and for computing has shortly change into one and the identical. Though the 2 differ of their particular wants, they nonetheless use the identical NVIDIA {hardware} and pay the identical NVIDIA “premium”. So each are set to change into a single product line to cowl the wants of all of NVIDIA’s skilled and business prospects,…