GTX 470 then, I will switch to nvidia's next gen as well, once they solve the global energy crisis in their cards
Guru3d's own review indicates that a 50mhz clock increase yields very little performance boost, likely due to a low ceiling in the current architecture. http://guru3d.com/article/radeon-hd-6850-6870-review/22
^ that bottleneck is due to no voltage control on 6870 imo.. 150 -200mhz seems doable once you start to use voltage tweaks.. you all seem to forget idle issues with 460gtx, that's one of the biggest turnoffs for me, wouldn't touch it even if it was only 100€... These DPC latency spikes suck bug time because they're SW controlled, unlike hw by 470gtx.
No voltage control makes the overclocking section in many reviews meaningless. What you've linked is simply its behavior on stock volts, and is not very telling about OCing as a whole.
Im talking stock volts on the 460 as well. Overvolt could get you higher. If you're talking hardware voltmod for the 68xx, I understand.
No, we're not. Of course overvolting each will get you higher results. The point is, you can't compare stock voltage overclocking between two cards, its meaningless. It's made even more meaningless by the fact that Fermi cards seem to have per-card voltages. My 470 is stock at .962, some are stock at lower, some at higher. The point is, they're incomparable without knowing the highest OCs, with overvoltage, on each card. Stock tells you nothing.
Agreed. However, at least you get a general idea what most of the 460's could do as opposed to the 68xx, a general trend if you will.
No, you can't. If all 460s came stock volted at 1.087, then they would OC really well on "stock volts" but when overvolted, they wouldn't get much further than that. Similarly, if a 460 came stock volted at .9v, then it wouldn't overclock for **** until you turned up the voltage. It's about how much margin for error the company allows. How much more voltage they give the card as opposed to what it needs to run at the desired frequencies. Sometimes, it can just barely be enough, as seems to be the case with the 6870, and sometimes, it can be way too much, as seems to be the case with a lot of 460s.
lol eclap, i like the countdown in the specs. Be sure to rememeber and post your results Nothing wrong with .9v on the GTX460 IPL, keeps things nice and cool, and 900mhz is still possible, though not much chance past of going past that though.
Yes, lets everyone bow down to nVidia every time they start whining and crying. They want reviewers to use software they provide when reviewing their cards....they want their OC'd cards compared to AMD stock clocked cards....well hell, what's next? Benchmark the cards and multiply the results by 2? If they're willing to cater to nVidia when reviewing competing products, their reviews are biased in favor of nVidia and are completely useless. I don't care to read biased reviews...regardless of who it's in favor of. Trusting a biased review, is no different than trusting a criminal to handle your finances....
You will find it was everyone who bowed down to AMD when they started whining, and telling reviewers what they should, or shouldn't test with:3eyes: Are you forgetting 90% of 1gb GTX460's are factory overclocked, nearly everyone who owns one is using a overclocked one, and the factory overclocked cards are CHEAPER than the stock 6870. It quite biased to compare the stock GTX460 to the stock 6870 when the RRP of the 6870 is 25% more than the one for the GTX460. And the whole criminal finance stuff at the end is just making you look a little crazy and angry.
Reference cards are reference cards. They should be compared to other reference cards in the same price range and market segment. Comparing a reference card to an overclocked card is the same as comparing a Ford Pinto to a Chevrolet Corvette on the basis of performance. If nVidia wanted the GTX460 to be clocked higher....they should have either set it higher to start with, or release another card with the intended higher clocks.
From a scientific point of view, maybe. But this ain't science. As I mentioned before, reviews are to get an idea of what's best in a price range. They're to tell you what's out there and help you and others decide what to buy.
Comparing overclocked cards to stock clocked reference cards doesn't convince me to buy an overclocked card or the same card at stock clocks....makes me think it lacks performance or, as in the past, someone is posting completely false benchmark results. Why would I want a card that has to be overclocked to compete...and why would I trust false benchmark scores? The only reason I own an OC card now...is because Newegg mislabeled it. The description said "MSI GTX460 Cyclone 1GD5 1gb memory". After I ordered, they changed it to "MSI GTX460 Cyclone 1GD5/OC 1gb memory" and changed the listed clocks to reflect the OC.
Who cares what it convinces you to do, your the only one with this deluded opinion. Showing just the stock GTX460 is misleading consumers who will assume all the GTX460's as that much slower than the 6870. There are plenty of reviews that use the stock GTX460's, most of them actually, why is it so bad that some of them are basing the review the questions that consumers would be asking rather than catering to what AMD/Nvidia marketing team would want. Your card is barely overclocked anyway, if i ran mine as low as yours it would probably idle in the teens lol.
850/2000/1700 is low?? My card clocks down to 50/67/100 at idle and clocks up to 850/2000/1700 under load. I'm even still using the stock 1.012v that the card came with. Unlike some, I don't need my OC noted in my specs to inflate my epeen size. Now, on to my issue with comparing OC'd models to reference. There are no overclocked HD6850 or HD6870's currently available....so, again, the results are bull**** as once non-reference HD6850 and HD6870 models come out, the results become misleading and in fact don't help people pick a card based on true performance.
