How could you defend a just released 14nm gpu which is slower and more power hungry then an 2 yrs old gtx970 on 28nm??
The hype train derailed guys, its over...no need to get angry...there is no gtx980 performance at 100w as some of you speculated
Sorry for only posting results that I had seen? How many false leaks did you post about?
Computerbase was able to undervolt their card and get it to be faster, cooler and no throttling.
https://www.computerbase.de/2016-06...el_potenzial_fuer_undervoltage_bei_der_rx_480
Also, unlike the $100 "premium" 1080 FE cooler (https://www.computerbase.de/2016-05/geforce-gtx-1080-test/6/) the 480 didn't throttle after 20 minutes: https://www.computerbase.de/2016-06/radeon-rx-480-test/5/
The 480 is only ever throttling due to power draw being set too high, not thermals like the 1080 FE.
So now APUs are faster than i3s? Guess people should recommend them over Intel for budget builds then.
I thought we were trying to look at CPU scaling here, but it seems like you've already made your mind up since you attack me every time I post information.
Since the very first benchmark posted is ROTR DX12, it might be interesting to look into what Scot Wasson posted. Scott is now with AMD, so naturally his post must be judged in that context, but he still has the TR pedigree to back him up and raises an interesting point.
If this is how you experience a somewhat higher average framerate in a AAA DX12 title, choosing between these two GPUs might not be an easy choice to make, even with higher DX11 CPU overhead for the AMD product.
What is your view regarding this situation?
That graph is likely not describing the full DX12 spectrum for these cards, but I reckon it is enough to start a conversation, and more importantly bring back frame times in the discussion, since CPU bottleneck might affect the experience in worse ways than average FPS (and not necessarily in favor of AMD).If that graph is representative across the dx12 spectrum you cant argue with any of what he said can you? Would like to see how the 4GB version fares though.
Since the very first benchmark posted is ROTR DX12, it might be interesting to look into what Scot Wasson posted. Scott is now with AMD, so naturally his post must be judged in that context, but he still has the TR pedigree to back him up and raises an interesting point.
If this is how you experience a somewhat higher average framerate in a AAA DX12 title, choosing between these two GPUs might not be an easy choice to make, even with higher DX11 CPU overhead for the AMD product.
What is your view regarding this situation?
And isn't this thread about the 970? The implications are quite on topic.it's specific to the 970, not maxwell or 4GB cards
And isn't this thread about the 970? The implications are quite on topic.
Which is most likely what that bench intended to exploit.sure, but that's (probably) a vram bug specific to the 970 (32bit 512MB slowing down everything), not Nvidia Maxwell or 4GB (or even 3GB, look at the 780) cards being horrible at DX12
Which is most likely what that bench intended to exploit.
Clearly both graphs show very high difference between 980 and 970 which is not generally seen between the two unless VRAM limited.needs to be tested in a card with more vram, 970 is broken
look at the 980... it's specific to the 970, not maxwell or 4GB cards
No, it's just Very High Textures in RotTR actually NEEDS *real* 4GB vram else it will stutter-fest. This is well known since early this year, lots of youtube videos covered it. Even NV's official performance tuning guide recommends not to use those texture settings unless the GPU has >4GB vram.
Some folks here were talking about how 4GB vram isn't enough for 1080p (I disagree) for the next 2 years... let's not get into a debate whether it's worth it to buy a 970 with 3.5GB...
Just recommend the 1060 coming up.
Which exploits the gtx970s memory configuration. You can see the 980 does far better with a conventional memory setup.
Why should I recommend the 1060 for anything? Ive seen no benchmarks.
No, it's just Very High Textures in RotTR actually NEEDS *real* 4GB vram else it will stutter-fest. This is well known since early this year, lots of youtube videos covered it. Even NV's official performance tuning guide recommends not to use those texture settings unless the GPU has >4GB vram.
Some folks here were talking about how 4GB vram isn't enough for 1080p (I disagree) for the next 2 years... let's not get into a debate whether it's worth it to buy a 970 with 3.5GB...
Just recommend the 1060 coming up.
All I see is people whining about a benchmark not going in their favor? I am not locking this due to this being info you don't "agree with". Ignore it and move to a different thread if you don't feel it is appropriate.
-Rvenger
Bacon1, yes or no, you are aware of the higher CPU overhead AMD GPUs require?
Pretty sure this has been well documented and tested over the last few gens.
So, yes or no?
More and more DX-12 titles are launching, those who buy new GPUs may have to be more concerned about the GPU DX-12 performance than CPU in older DX-11 games.
And yes DX-11 games will still be available but GPU DX-12 performance may not be enough for the price you will pay in 2016-2017 with some GPUs
I think that every gamer's needs are different. If somebody plays titles that tend to favor AMD's architecture more, then AMD could be the better choice. Similarly, if one plays a lot of titles that favor NVIDIA's architecture more, then NVIDIA could be the better choice.
Trying to predict how future titles will perform, IMO, is an exercise in futility. Based on the Beta, everyone thought Doom would favor AMD's architecture but when the title launched, it became clear that it favored NVIDIA's.
Buy GPUs based on the games you're most interested in playing today, IMO.
I recently came across a review that compared gpu perf with different CPUs. They had pair of GPUs with relatively close performance from both vendors and a lot of CPUs. It showed perfectly that there is no "amd driver overhead killing OCed haswells" effect.
Here it is:
https://www.computerbase.de/2016-05...mm-the-witcher-3-mainstream-gpus-im-vergleich
You can see that those GPUs stay the same relative to each other across a super low IPC old bulldozer and latest and greatest i7.
Yep, I posted their 480 CPU scaling as the 2nd post in this thread, but it is getting ignored.