Caporegime
as above using driver improvements, or like I imagine the GTX470 will always remain significantly ahead as it receives driver improvements too.
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
as above using driver improvements, or like I imagine the GTX470 will always remain significantly ahead as it receives driver improvements too.
It's kinda funny that people assume 460s overclock "better". It's an argument being used against any card out there yet only some 460s barely manage over 900MHz on core (being less than 35% overclock and we're not talking the cheapest models now). Most reference 5850s could achieve about 40% overclock with voltage increase, that is what Hawk edition of MSI 460 is capable of. There are many 470s overclocking to 850MHz (40% increase over stock) on core. I also haven't seen any unbiased review claiming that GF104 overclock with better gains than GF100, neither I've seen ones claiming that original Fermi cards scale better with overclocks than Radeons. It's all claims by hardly unbiased users (eg justifying their purchases).
Also, a 5850 is 10% faster than a 460 and a 470 is 10-15% faster than a 5850. That leaves a pretty significant difference of at least 15% with stock cards. The gap increases with higher resolutions too. SLI scaling is debatable, 470s are always faster in this case scenarios but if anyone is seriously thinking about SLI, 460s are more appealing because of their power usage/heat output.
I can report that 10% increase on clock and 10% increase in memory gives 10% increase in performance on gtx 470. I tested it up to 30% on both memory and clock and gains were 30% or something like 29.9%. Couldn't take it any further as memory doesn't go that far. Overclocking just the clock will see diminishing returns on the increase in performance.
I had 5850 at 1ghz core and 1.3 memory and that had 44fps(stock 33 if irk) in crysis. So do the maths how well that scales.
But the overclock increases obviously are dependent on the tasks? You've not even stated how you'd taken these measures.
What do you mean how I taken this measures? If its my unbiasedness you're worried, I am telling you I owned 2x5770, 5870, 5850, 470.
measure stock performance in vantage, overclock both memory and clock by 10%, record the new result in vantage and calculate the % increase.
% increase in memory and core = exactly same % increase in performance in vantage.
I am not saying it will performs like that in every application, all I am saying in vantage it does.
I personally hate diminishing returns. I preffer to buy best bang for buck card and overclock it to the flagship, I did it with 5850 at 1k ghz I did it with 470 at 850 clock. If only 460 was out when I got 470, I would have gotten 460 and overclocked the socks off that one.
Damn what kind of cooling did that card have?
No, average overclock max for 470 is 30% (800core)
For reference due to insane temps 850 core is quite unreachable for most people on air unless you're lucky to get a low stock voltage 470, which are, as you understand, rare.
Damn what kind of cooling did that card have?
470-460 = 10
I win at math. Bigger number = better.