Side bar question; what are your needs for the compute heavy on GPU?Damn. Expected this, but not that bad.
I'll get me V56 for no money and skip this one. Was hoping next gen would be on par with Fury, but maybe not quite...
Side bar question; what are your needs for the compute heavy on GPU?Damn. Expected this, but not that bad.
I'll get me V56 for no money and skip this one. Was hoping next gen would be on par with Fury, but maybe not quite...
If you give the little monkey really sugar, then it is even close to the 2.1 GHz mark and in places even more! Of course, then, what flows through the veins, no more blood stream, but high-energy electricity needed life juice. But it works. Easier and more stable than the Radeon VII has ever done. At about 2.1 GHz, there seems to be a kind of physical barrier, but what the hell. Exactly there will be a follow-up as article and video.
A question for those who have actually had the time to read more than the last page of a couple of random reviews, if I may:
What's the TL,DR?
Navi 10 looks solid on price/perf/efficiency for gaming workloads, reference cards and drivers leave something to be desired?
The realtime GI system i'm working on. Somehow i can't imagine Navi or Turing beating GCN performance for me. So i want Vega, and the current offers are too good, and Stadia uses it...Side bar question; what are your needs for the compute heavy on GPU?
PSA: V-Ray GPU is EOL. It has been replaced by V-Ray Next GPU (benchmark here) which is unsurprisingly CUDA only because $$$ (VRay GPU's OpenCL path was 6 years too late and totally useless with literally no development effort put into it after the initial release)The realtime GI system i'm working on. Somehow i can't imagine Navi or Turing beating GCN performance for me. So i want Vega, and the current offers are too good, and Stadia uses it...
It's hard to say how those newer architectures really perform by looking at benchmarks. For instance in VRay Navi is faster than any Vega (https://www.guru3d.com/articles-pages/amd-radeon-rx-5700-and-5700-xt-review,27.html):
Still unsure, but all my GPUs died in one week recently - only 5950 left which sometimes turns off. I should get another GCN while they are still available. Can't imagine life without GCN, hahaha
Yeah, that one has been bugging me. Unfortunately I don't have a good explanation right now. I consistently get those results. It may be another driver oddity.AT has synthetics that look strange, especially Vega doing much better on pixel fillrate,
https://www.anandtech.com/show/14618/the-amd-radeon-rx-5700-xt-rx-5700-review/14
Basically at the given price points unless you need or want RT, Navi offers arguably better value than NV cards currently.
If price doesn't matter then obviously the top of the NV product stack has no competition.
Regards,
SB
They still need another 25% in perf/watt to nulify the node difference.While these cards aren't for me, I have to say I'm both surprised and encouraged that AMD has basically matched Turing in perf/watt. And exceeded them in perf/mm^2. However, it's not all rosy as they are still at a feature deficit (RT while questionable in value right now is a glaring omission, for example).
They still aren't competing at the highest enthusiast level, but at least they offer viable competition now, enough that NV were forced to offer better value propositions on the lower end of their product stack.
Regards,
SB
No. 7nm vs 12nm / 16nm and lacking some important features (RT, mesh shaders, virtualink...)While these cards aren't for me, I have to say I'm both surprised and encouraged that AMD has basically matched Turing in perf/watt. And exceeded them in perf/mm^2.
Trick question: Which card needs another 25% in perf/watt to nullify the node difference?They still need another 25% in perf/watt to nulify the node difference.