Absolutely not. We are skeptical of Maxwell on a niche forum. The average person will buy your 980s close to full price.....
I got my R9 290 for $200, and the person went out and bought a GTX 970 new. It's doing worse in the games he is playing than the 290 he sold to me.
People just buy based on what they believe is right, they rarely look at benches, and I've seen people look at benches and not understand them. Which I don't get at all....
I got my R9 290 for $200, and the person went out and bought a GTX 970 new. It's doing worse in the games he is playing than the 290 he sold to me.
Actually Tahiti came in January 2012, two years after GTX 480 (March 2010).
actually the paper launch of the 7970 was december 2011
wow.
He just formatted too due to the NVidia driver thing. Like I felt bad, but well, he wanted to go with Nvidia and who was I to stop him.
But he also switched too because of AMD Raptr or whatever. It wouldn't update his drivers for him so he went to the Geforce Experience.
Hence why I say, Geforce Experience rocks when it works.
I don't recommend ANYONE sell their panel to swap vendors if you invest in Gsync/Freesync. It's a wash then. Does it matter if you're getting 42 FPS and an R9 390 user gets 10 more FPS than you at 1440p? You're still in Gsync range, so it's all gravy baby.
Its not really that surprising with a low level API. Just look at all the Tonga/Fury issues with Mantle and BF4/Thief. And I think its just a DX12 side effect we have to deal with. It needs optimization from the devs for every single uarch.
I wonder if his OS is hosed now with recent driver updates, thanks GE?
Are these guys trying to measure with FCAT or something? Why the heck are they crossing out so much of their reviews.
DX12 + FCAT do not mix.
And what is this talk at the start where people were saying AMD perform poorly in DX12? Because the stock 390 is wrecking a highly OC 980.
Both the 390/X are even close to the 980Ti that's a massively OC model.
It became apparent that both companies are pushing different commercial strategies based on the underlying architectural philosophies in their hardware. One is clearly going for a "subscription" model, giving you good performance as long as your model is the "current generation", because they 100% dedicated into optimizing the current uarch, which tends to differ in a good degree from the previous releases, while the other company gives you performance improvements as long as the underlying architecture (GCN) stays mostly the same, which at this point did for 4+ years.
A sad day for all these GCN cards will be when AMD decides to move foward and flip the table upside down with a totally different architecture. From what I gather Polaris will not be the case, so it is still a good moment to run a GCN card. Kepler and now Maxwell users on the other hand....
It became apparent that both companies are pushing different commercial strategies based on the underlying architectural philosophies in their hardware. One is clearly going for a "subscription" model, giving you good performance as long as your model is the "current generation", because they 100% dedicated into optimizing the current uarch, which tends to differ in a good degree from the previous releases, while the other company gives you performance improvements as long as the underlying architecture (GCN) stays mostly the same, which at this point did for 4+ years.
Sure, that must be the reason why nVidia is 50% faster in GoW. :thumbsup:
Sure, that must be the reason why nVidia is 50% faster in GoW. :thumbsup:
You didn't know AMD released a driver for GOW? Not too bad, since they didn't have access to it and weren't told of the game's release.
60% performance boost!
Even GameWorks can't gimp GCN for long these days, pretty weak they should stop wasting all that $ on a useless program.
update:
The original benchmarks, before updating, had a Fury non-X only a little faster than a 980 at 1080p. That was deemed a failure by two users here.
I have a GTX960 myself,but how can anybody twist this result into a failure for AMD??
It is easy, you need at least 1 INTERNET STRONGMAN. With 2 you are mostly covered tho as you would have 1.7x the FUDING throughput (yeah, they scale as good as the cards in mGPU they are trying to defend), then they will come to the thread quickly to spin the results to whatever the new goalposts are.
The odd thing though is in DX11, Fiji is well optimized.
This does actually suggest DX12 requires game developers to optimize on an architecture-specific basis.
It's clear due to Fiji being frame rate locked to 60 while Hawaii is not, there are major differences at play for sure.