Nvidia CEO warns of “extraordinary, unusually turbulent, disappointing” Q4
37 replies, posted
It sounds like data centers shopped around and got better contract deals with AMD 7nm cards
They're going to die when Intel gets into the market
Two 88Us lasted me until I bought a 580, the trick was no antialiasing.
and so was I, I also deliberately avoided using overclocked examples because a comparison between the HIGHEST OC'd SKU which is conveniently posted in your video so I don't even have to point it out and a base-build 2080 with a custom shroud also isn't any kind of objective. The other cool thing you're failing to mention off of that video is the FTW also costs 400 bucks more than a vanilla 1080ti MINIMUM, and the average cost is exactly 100 bucks less than a 2080ti FTW.
The TU silicon is simply more expensive than a competing GP die because of the space dedicated to RTX and Tensor transistors. The price for a full card is also more expensive due to GDDR6.
It's just a matter of fact that Nvidia wants to (and CAN) maintain their Pascal margins.
AMD has failed to address critical design flaws in GCN (super fat-compute core that hinders the overall clockspeed, and is super power hungry, ineffective ROP count, esp compounded by the clock inefficiency, inefficient front-end and schedueler) and it's biting them in the ass, I can no longer defend AMD's design practices when they've made this mistake 3 generations in a row.
AMD also needs to dedicate resources to get VCE/UVD/VCN up to par with NVENC and NVDEC.
AMD had made some good design decisions, I still think HBM is good technology, but has executed horribly since the R(X) 2XX series.
On the other side of the lake:
https://www.tomshardware.com/news/amd-earnings-4q-2018-ryzen-epyc,38523.html
Vega got 1/3 of the R&D allowance whilst Navi got 2/3 and it's evidently show.
https://www.forbes.com/sites/jasonevangelho/2018/06/12/sources-amd-created-navi-for-sonys-playstation-5-vega-suffered/#36aeb9b424fd
Sorry, you need to Log In to post a reply to this thread.