Originally Posted by Mookster
Once we found out the die size, all those claims became irrelevant. Just as many people were saying AMD would be incapable of beating a 980 Ti because they have absolutely no knowledge of what die shrinks do. It's been 5 years, after all, so we have a lot of people who haven't even been interested in the enthusiast scene for long enough to see one.
I know I've personally made rebuttals explaining why the new generation of AMD cards could easily surpass any of the current offerings.
But again, die size matters. And fab process matters. It's difficult to compare based on die size now, because we have NV on TSMC's node and AMD on GloFlo's node. We more or less need to compare on price, as opposed to the traditional way (die size) to find out who's touting the superior architecture.
And there are more questions. Everyone is hitting a hard limit when overclocking, reporting extreme artifacting and crashes once they get to the tipping point. This is symptomatic of a GPU that isn't getting enough power. This makes it even more difficult to know what the chip is capable of, but we would hope that AMD didn't do this to save money and keep P10 from interfering with future midrange segments.
It's difficult to make accurate predictions about this stuff. Some people might see that AMD opted for 14nm FF and assume the node is superior to TSMC's 16nm FF, but they'd likely be wrong because GloFlo nodes have been disastrous for years. Some people will see the clocks or temps and assume the chip doesn't have more potential, but they could easily be wrong as well because AMD has junk reference coolers and we might be hitting a power draw wall when overclocking.
Saying you "called it" just means that you made a severely uneducated guess and you happened to be right. That, or you know more about the TSMC, GloFlo, and the testing of this chip that occurred behind closed doors with different coolers, drivers, and power delivery. If it's the latter, I'll excuse myself. But, I know you wouldn't be on here saying you called it if that were the case.
I'll humor you and reply even if your attitude is clearly antagonistic and vengeful.
No, I didn't "know more about the TSMC, GloFlo, and the testing of this chip that occurred behind closed doors with different coolers, drivers, and power delivery."
I simply made an educated guess, that you call an uneducated one, because 1) It's AMD 2) the Fury X underperformed and 3) There's been tremendous unwarranted hype and rumors since the card was announced. Surely, it doesn't take much reason (or a rocket scientist) to figure that the card would probably not live up to the rumors. It's really as simple as that.
Of course, it probably helps that I'm a fairly high ranked enthusiast benchmarker on HWBOT that has benched nothing but AMD cards since 2011. At least 10 of them. The majority are GCN. So I have a fairly deep understanding of how these GPUs work, how they overclock and so forth. I'll assume you're capable of finding my profile there, if you're so inclined. Traditionally, AMD cards overclock less than Nvidia, so I figured the rumors about the RX 480 overclocking to crazy levels were probably false. Though, that may change with AIB cards and better power delivery. However, I strongly doubt this will be the case based on my experiences.
Even IF the RX 480 can
overclock to the 1500mhz range, and by doing so it pulls ahead of the 390/290, from the benchmarks of it I've seen it will still be less powerful than the 390 at the same clocks. E.g. it may take a RX 480 at 1500mhz to equal a 390/290 at 1200mhz. In which case, the card is basically a failure imo. It needed to be faster than the 390/290, but it isn't, probably because it has less ROPs and TMUs.