Intel doesn't want to or have much reason to compete with NVidia, it's AMD who's stealing all their high-margin marketshare with their server racks full of Vega's & Epycs. This is a GPU that will use exactly the same principles, and compete in the same sphere, as Fury & Vega(The latter being massively successful in its target arena and posing a continued threat to Intel heading into 2019 as 7nm Vega looks to further revolutionise AMD's Epyc server rack options, but a "flop" to gamers, which is why I expect gamers to percieve Intel's cards to be the same regardless of how well it compete in its primary markets) . It will undeniably be an expensive card given their targets & projected technology, HBM is absolutely critical for their goals, so it might get excellent perf/watt when gaming, but expect absolute performance to be in line with other single-slot 75W cards of the era while having a price tag closer to a 300W beast. Since this is primarily targeting use in racks of servers with tens of them, efficiency will be key, which means a fairly modest clock speed/performance target for an individual card, with the benefits of the architecture truly coming to play when they're stacked together in large number, much like Vega(A good chunk of Vega's die is committed to managing large asymmetric memory hierarchies for cases like this), while gaming famously doesn't lend itself well to such levels of paralellisation with current software approaches.
Rada Koduri's legacy at AMD was not relatively run of the mill consumer-focussed cards like Polaris using mature technologies, it was the use of progressive & expensive technologies ahead of their large scale debut like multi-die interposers to create high bandwidth compute monsters for what a few years ago were niche markets for GPUs but have now became the primary markets in terms of revenue. Polaris kept Radeon alive in the short term, but Fury & Vega sowed the seeds necessary for their long term survival & showed true skill at predicting, anticipating & preparing for future trends even in the face of consumer adversity(Not that consumers really ever know what's good for a company who primarily makes products for markets a million miles from the average consumers use case).
Rada Koduri's legacy at AMD was not relatively run of the mill consumer-focussed cards like Polaris using mature technologies, it was the use of progressive & expensive technologies ahead of their large scale debut like multi-die interposers to create high bandwidth compute monsters for what a few years ago were niche markets for GPUs but have now became the primary markets in terms of revenue. Polaris kept Radeon alive in the short term, but Fury & Vega sowed the seeds necessary for their long term survival & showed true skill at predicting, anticipating & preparing for future trends even in the face of consumer adversity(Not that consumers really ever know what's good for a company who primarily makes products for markets a million miles from the average consumers use case).
Last edited:
