Originally Posted by ZealotKi11er
You where comparing GTX780 to GTX970. You are right. It's not entirely about how well GTX780 scales but instead GTX780 had more OC room. Most if not all GTX970 where non reference coming with 13XXMHz Boost clocks out of the box. 15XXMHz OC would be like 200MHz which is less then 15% while got GTX780 going to 1200MHz you get much larger % OC. People just associate 15XXMHz are better overclocker in general.
Yes I was only comparing the overclocked 780 to an overclocked 970, to previously show that clockspeeds of the 780 or any Big Kepler card are inherently important. It is relative information, when trying to portray the idea that benchmarks showing a 280X right by a 780 should not be taken seriously. When even TPU has a 280X 13% behind a 780 @ Stock Clocks:
You are right that Kepler had more overclocking headroom, because even a 970 that boosts to 1350MHz out of the box that can overclock to 1600MHz is only a 18.5% increase. So it is just the illusion that Maxwell overclocks better since it hits much higher clocks than Kepler could ever dream of.
Originally Posted by gamervivek
Overclock vs. overclock is after the fact. And you're wrong about clockspeeds that the 780 hits.
And the benchmark you linked here are with catalyst 14.8, more than a year before. Right now in TPU's latest review, the 280X whoops the 770 by around 20%-25% putting it close to the 780. Without crimson, let's see what happens next.
It'd be ridiculous because the difference between 280X and 970 is three times and the 970 launched at 300$ less than 780.
Proceeds to link to linus.
When PClab have AMD behind despite AMD being on par or even better in other reviews for just about every review, it's a red flag. And you were relying on PCLab despite being shown that the other sites in your OP itself don't show the same. Oh and here come the techspot results
I did that but as I said before it wasn't on their test setup/methodology page. Secondly, unlike gamegpu they didn't seem to have becnhmarked the beta, so I was inclined to believe that they used the multiplayer.. At least before putting so much confidence in your assertions.
No it's not after the fact, because most people bought non-reference cards that boosted into the 1100's anyway (with those same non-reference cards overclocking close to the 1300's):
I'm not wrong about the clockspeeds a 780 hits, I own one. While you with your Radeon 4850 from 7 years ago, has no personal experience with what a GTX 780 can boost to at stock settings.
Yes, if you actually read what I wrote you would see that I stated many of these benchmarks are before AMD's late to the party driver optimizations. Let's not exaggerate 20-25% it is closer to 16% as you can see here with TPUs latest review:
and also a 280X is still 13% behind a stock clocked reference 780 (must be NVIDIA biased).
It would be ridiculous because an overclocked 780 performs on par with a 970, therefore any conclusions formed around the basis that a 280X is as fast as a 780 is preposterous. Also, while the 780 launched @ $650. It soon dropped to $500 and then again to $400. Which is the price I payed for my 780. Four months ago you could have gotten an EVGA GTX 780 Classified for $210. A card almost guaranteed to do 1250MHz.
Yes Linus is so NVIDIA biased, even after he was one of the few reviewers who praised the Nano when almost every other review site was like, "but why?"
PCLab is the only reviewer currently to my knowledge that benchmarks a non-reference 780. Which is why I trust them more than the rest of the reviews on a 780. If you read earlier on in the thread I stated that NVIDIA doesn't want non-reference 780's and 780 Ti's pitted up against Maxwell, because it would quickly show people that Maxwell is not nearly as good as reviewers might make it seem. Because as stated again, clockspeeds are very important when trying to compare Big Kepler cards. When they are at stock clocks, their performance is hindered. When you overclock them to a modest 1200MHz, it's like an entirely different card (even non-reference cards @ 1100MHz is a big difference). You also must realize with different reviewers, that they are all using different test setups, with different base components, with different Windows installs, in different rooms, with different ambient temperatures, in different test benches or case setups. It's not a red flag, just a variety of variables come into play when benchmarking. Perfect example is Tech Spot has their 780 Ti @ 71fps with the 970 @ 84fps @ 1080p, while Guru3D has the 780 Ti @ 78 fps with the 970 @ 76 fps @ 1080p. Quite a big difference here.
Point being overall, about your multiplayer benchmarking mistake, is you shouldn't be so quick to reject what someone is saying without doing your own research. You are so quick to call out, "NVIDIA Biased" or "Excuses" yet you barely even look at the information presented to you beforehand. Seems you are hellbent on proving that a 280X is close in performance to a 780 and the 780 is not a good value. When I'm attempting you show you that, this is not the case, for both non-reference 780's as well as overclocked 780's. They are night and day from reference cards.Edited by BiG StroOnZ - 11/25/15 at 2:41am