What is the actual performance increase in overclocked cards, specially factory overclock ones?

[GeneralOreo]

Actual meaning in games.

I've been messing around with my 8800GT and the results I'm getting are pretty weird. 2DMark06 of course goes up but Crysis doesn't - at all. Far Cry neither, except when I hit 1645 shader clock and WHOA! I get a 140 FPS score in the hardwareOC benchmark... this is compared to about 85 FPS. o_o

 

[taltamir]

don't use artificial benchmarking

 

[j0j081]

Originally posted by: taltamir
don't use artificial benchmarking

artificial boobs are sometimes okay though. D:

 

[Borealis7]

not exactly the model you were looking at but
TrustedReviews made the comparison between a stock GT and a Factory OvercloCKED one

 

[CrystalBay]

In Crysis the difference between a stock 88Gt and a ssc 88gt may only be 2-5fps average, depending on monitor resolution and in game settings... In UT3 however it could be 10fps+ again depending on certain settings

 

[secretanchitman]

as long as you have an 8800GT from a company that includes overclocking in their warranty, then you're fine.

 

[taltamir]

Originally posted by: CrystalBay
In Crysis the difference between a stock 88Gt and a ssc 88gt may only be 2-5fps average, depending on monitor resolution and in game settings... In UT3 however it could be 10fps+ again depending on certain settings

considering I get 0.7 fps on crysis at max settings I would say 2-5fps can be significant... if it goes from 20 to 25 fps then its a 25% performance increase which is really impressive...
Heck if it goes from 5 to 10 then it is a doubling... if goes from 200 to 205 then its a mere 2.5% and doesn't really matter.

 

[nullpointerus]

Originally posted by: taltamir

Originally posted by: CrystalBay
In Crysis the difference between a stock 88Gt and a ssc 88gt may only be 2-5fps average, depending on monitor resolution and in game settings... In UT3 however it could be 10fps+ again depending on certain settings

considering I get 0.7 fps on crysis at max settings I would say 2-5fps can be significant... if it goes from 20 to 25 fps then its a 25% performance increase which is really impressive...
Heck if it goes from 5 to 10 then it is a doubling... if goes from 200 to 205 then its a mere 2.5% and doesn't really matter.

But the actual performance increase would be on a percentage basis, not an absolute value like 2-5 fps. Moreover, the percentage improvement will not exceed the core clock increase. So if you have a regular card at 600 MHz core clock and a SSC card of the same kind at 720 MHz core clock, the resulting performance increase would be no more than 20% maximum. Average increase would be 5-15% IMO, depending on what in the card was limiting the game in the first place, how powerful your CPU is, etc.

 

[Belles Toaster]

Overclocking has always been a security blanket. Manufacturers always leave a margin
so that they don't get swamped with RMA's. This varies from 7% to as much as 15%.
Look at the new 8800GT cooler fan undoubtedly there was a problem. As you go to
premium cards you wonder if the savy end user would set up his "rig" any cooler than the
"noob". Also would he heed the warning signs that a problem is brewing under the hood,
by high load temps and artifacts. All cards overclock, but to how much and for how long is another story since although we think chip maufacturers yield excellent samples all chips are not 100% equal and some companies will stress test sample cards to re-badge them as extreme editions. (More with an eye to limit RMA's hehe.)

 

[jonks]

http://www.extremetech.com/art.../0,1697,2230250,00.asp

There ya go.
Cliffs: not much difference

 

[taltamir]

Originally posted by: nullpointerus

Originally posted by: taltamir

Originally posted by: CrystalBay
In Crysis the difference between a stock 88Gt and a ssc 88gt may only be 2-5fps average, depending on monitor resolution and in game settings... In UT3 however it could be 10fps+ again depending on certain settings

considering I get 0.7 fps on crysis at max settings I would say 2-5fps can be significant... if it goes from 20 to 25 fps then its a 25% performance increase which is really impressive...
Heck if it goes from 5 to 10 then it is a doubling... if goes from 200 to 205 then its a mere 2.5% and doesn't really matter.

But the actual performance increase would be on a percentage basis, not an absolute value like 2-5 fps. Moreover, the percentage improvement will not exceed the core clock increase. So if you have a regular card at 600 MHz core clock and a SSC card of the same kind at 720 MHz core clock, the resulting performance increase would be no more than 20% maximum. Average increase would be 5-15% IMO, depending on what in the card was limiting the game in the first place, how powerful your CPU is, etc.

That was actually my whole point... he was giving fps increases instead of percentage increases.. He was saying how crysis shows that the GTS is insiginificantly more powerful as it only results in 2-5 fps increase. I countered that I only get 0.7FPS on maxed out crysis... so its not a valid measurement and that we need to know the percentages.

 

[GeneralOreo]

Forgot about this thread, but just wanted to point out that the difference between the shader clock didn't result in the 60 or so frames in the benchmark - the benchmark itself just runs slower the first time and I sometimes restarted my PC or closed the benchmarking application then opened it again while overclocking. I just discovered this recently.

It makes sense now, no way can a 30/40Hz extra in the shader clock make the frame rate jump up ~ 60 FPS, especially when it went down to 85 just when I reduced the clock back, and other clocks (core/memory) didn't make a similar difference either. I feel like such a moron.