Xeon phi was a multi core atom chip.
This is a GPU. Totally different beast.
Turing is still better than rdna on an older node. Rdna 2 will close the gap, but too late cause ampere will widen it again. TBH, AMD is pretty lame with gpus
You do see it here so it does exist. This is a workstation gpu, not a consumer gaming gpu. If one of the chiplets has 512eus, than one chiplet is close to rtx2080ti (10+ tflops)
"AMD is pretty lame with GPUs"
Why is it that people expect AMD with a far, far, far smaller budget to compete with Nvidia? Not only does Nvidia have a bigger budget, but we know it's also pulling stuff like the GeForce partner program and other unhanded forms of "marketing" (just like Intel who was convicted of Bribing OEMs to such an extant that even if AMD were to give them free chips, they still claimed theyd lose money". I'm not saying that definitely something nefarious is going on, but it'd be pretty naive to think it's impossible, just based on Nvidia's past precedent of doing this exact stuff.
It's just like the double standard with pricing.... Nvidia gouged with its RTX line, but as soon as Navi was priced similarly due to similar performance, everyone cried murder. Somehow people think telheyre entitled to cheap as dirt GPUs from AMD, but Nvidia can price however they want
And then of course, I can't stand hearing the people wishing that AMD was more competitive and blaming AMD for not being more competitive. Well, the undeniable fact is that in the past, like in the late 2000s, even when AMD had a far better GPU at a better price, you computer enthusiasts STILL bought more Nvidia, and that has basically happened every time AMD has a better product and if you think that's crazy, just look at Ryzen 4000 for mobile, clearly a better product, but will it outsell the competing Intel chip (even if it wins an equal amount of OEM designs) ? Most likely not. Basically what I'm saying is, no one really has a right to blame AMD, because even when they gave you exactly what you claimed you wanted... You STILL didn't buy it. As long as there are consumers out there that DON'T take the literal 5 minutes to Google some reviews and just trust what idiots on the internet say or the uninformed Best Buy salesmen, Nvidia will always outsell AMD, and Intel too for that matter. it's really frustrating to me because I won't even buy a $40 piece of electronics without doing as much research as possible to be absolutely sure I'm making the best possible purchase, so I don't understand how any consumer can buy computer hardware that costs multiple hundreds of dollars, without doing the same... And yet the vast majority do. As long as Nvidia and Intel can depend on the willful ignorance of consumers, it doesn't matter out much AMD's products outperform them... They just never win
*I'm not an AMD fan, my motivation for these statements is a desire to ensure that people know the true history behind this subject, because I'm sure most of them think that Nvidia has the market share because they always had better products, which is not true at all. Marketing is why they have the most market share, that's the only way to make sense out of the fact that even when AMD had better products for less, people still didn't buy them.
And of course Nvidia is ahead in architectural efficiency, what gave you the impression that I would say otherwise? AMD is now mostly matching them in absolute perf/W with a significant node advantage, which tells us that RDNA (1 at least, no idea about 2 obviously) is still quite a bit behind Turing even if it makes major strides compared to GCN (even when accounting for the node change).
