So with the Vega launch being a success, is it the beginning

>>61907993
I'll admit I'm an AMD fanboy and even I don't think so. AMD still can't touch the 1080 Ti and unless Raja has something up his sleeve with Navi the same will probably remain true for Volta.

>>61907993
t. redteam+ member

>>61907993
t. constipated redteam+ member

>>61908101
>>61908089
t. leather BLACKED man

>>61908045
The fact that Vega FE forced Nvidia to give the Titan XP Quadro drivers means they are worried.

>>61907993
They finally have a competitor to nvidias 70 card on release but anyone who wants it for games probably wont be able to buy them

>>61908182
And releasing the drivers made them stop worrying. Thanks for getting them to release the drivers though. That was nice.

>>61908298
They still can't do nothing against SSG.

>>61908312
>They still can't do nothing against SSG.
Neither will anyone who buys it.

>>61908312
is that train wreck released?

File: unseasoned.png (379KB, 621x672px) Image search: [Google]

379KB, 621x672px

>>61908208
>taking over a year to barely hobble together a mid tier gpu
>competitor

>>61908358
>mid tier gpu

Mid-tier cards are the rx 580 and gtx 1060, 1070 and Vega 56 are high-tier, vega 64 and gtx 1080 are high end

>>61907993
I consider the Vega architecture to be a success, Vega FE is great for its intended purposes, and vega allows them to finally break into datacenters and AI. As for the vega gaming cards, the 56 is pretty good, the 64 isn't worth it for most people imo. I'm also curious to see how fp16 will impact performance in games like far cry and wolfenstein.

Does anyone have any speculation on how Vega 20 will turn out? Will porting vega to 7nm just mean better power efficiency and much higher clock rates? sounds like architecturally it'll be the same.

>>61908407
Vega20 is no longer 7nm. It's 14+.

>>61908407
>and AI
Did they release any new API with it? Because CUDA has got a lot going for it and the AI community isn't happy with openCL/openGL

Success?

>>61908379
in reality, amd's high end has been competing with nvidia's midrange since the Kepler days aka gk104. HD 7970 was so weak that nvidia decided to use their gxx04 chips to compete with amd's high end

>>61909664
Look.
That's some shitty bait.

>>61909676
nvidia has been using gxxx0 for their high end since fermi and below, the less number, the higher in performance rank. There are several articles about this.

File: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9SLzcvNjkzMDQzL29yaWdpbmFsLzAyLVNvbGlkd29ya3MtUGVyY2VudGFnZS5wbmc=.png (36KB, 711x533px) Image search: [Google]

36KB, 711x533px

>>61908407
>>61908420
This is how Vega 10 performs in geometry when it isn't memory bandwidth bottlenecked. You can deduce for yourself the implications for a gaming version of Vega 20.