GPU Overclocking not affecting 3dMark06 score

Tropicocity

Member
Mar 12, 2008
28
0
0
Hi all,

Ran 3dmark on my 8800gts 512 sli setup at stock 675/1674/1944 clocks on an e6600 @ 3.6ghz.

After a few 3dmark runs at stock, average score was about 15.25k-15.3k, so i thought I'd try my hand at clocking the cards up.

So sure enough, I set the fan speed at 100% to be safe, and ran the sli setup at 800/2000/2200. To my surprise, and shock, I got pretty much the exact same 15.25k-15.3k score, possibly a couple of points lower and higher here and there.

Now, I thought maybe it just didn't like that overclock - though no artifacts were shown, i decided to lower the Oc a bit to around 750/1875/2100 - and was still presented with similar scores!

All in all, having disabled Every program running in the background so that 3dmark was basically the only thing running, I think i topped out at about 15.7 ish.

My question is, do you think that my e6600 @ 3.6ghz is bottlenecking the cards?

I realise 3dmark _always_ scores better with a quad even at lower clocks, and way more at the same clocks, I'm just wondering If my dualie should be switched out for a quad to make my Sli setup worthwhile.

Also ran Call of Juarez DX10 bench @ 1680x1050, 46.9 average on stock gpu, 51.6 average with Gpu at 800/2000/2200 - a mere 6 FPS increase. Crysis seems to deliver only 2-3 extra fps with higher clocks too.

thanks :)
 

rgallant

Golden Member
Apr 14, 2007
1,361
11
81
Originally posted by: Tropicocity
did i post in the wrong section :X?

Everyone could be busy talking about the new cards , bad timing for you , I posted abench of the only E8600 test I have seen, in the cpu section and only had 3hits
- but I have the same set up with a E6700 @ 3.6 the same speed as yours and same cards
-@790-1880-1090 I get apox, 16400-16500 3d-06 16950 @ 3.8
-so your out a 1000 points somewhere
-bench testing crysis maybe 5fps (for the avg.) from a low clock to the one posted using xp sp3 .Max fps =70 ,min ,16 , avg. 45-49 fps Island bench test
on 175.16's @1680-1050 all high no AA

-you might have a driver problem or a patch problem I stopped trying 1.21 ,stayed at 1.2
--I get about 14.3 sec. in super Pi 1m
(as a mother board speed reference)
-you have to use the default nvidia control panel settings for reference
-hope some of that helps


http://www.xfastest.com/redire...id=10995&goto=lastpost

-, check this out for your new cpu
 

Tropicocity

Member
Mar 12, 2008
28
0
0
Originally posted by: rgallant
Originally posted by: Tropicocity
did i post in the wrong section :X?

Everyone could be busy talking about the new cards , bad timing for you , I posted abench of the only E8600 test I have seen, in the cpu section and only had 3hits
- but I have the same set up with a E6700 @ 3.6 the same speed as yours and same cards
-@790-1880-1090 I get apox, 16400-16500 3d-06 16950 @ 3.8
-so your out a 1000 points somewhere
-bench testing crysis maybe 5fps (for the avg.) from a low clock to the one posted using xp sp3 .Max fps =70 ,min ,16 , avg. 45-49 fps Island bench test
on 175.16's @1680-1050 all high no AA

-you might have a driver problem or a patch problem I stopped trying 1.21 ,stayed at 1.2
--I get about 14.3 sec. in super Pi 1m
(as a mother board speed reference)
-you have to use the default nvidia control panel settings for reference
-hope some of that helps


http://www.xfastest.com/redire...id=10995&goto=lastpost

-, check this out for your new cpu

Hmm well I am running vista 64 if that helps :\ could attribute to the 3dmark score.
what bugs me is, when i was on my asus p5b, my cpu scor was around 3300/3400 i believe, having used it in my sli setup on a evga 780i, its down to like 2900 :eek:

Also my crysis testing was only on the demo as I dont have the game anymore :
running modded drivers and they actually Upped my score from like 14.9k or something xD

Maybe i'll just go quad :

 

rgallant

Golden Member
Apr 14, 2007
1,361
11
81
Originally posted by: Tropicocity
Originally posted by: rgallant
Originally posted by: Tropicocity
did i post in the wrong section :X?

Everyone could be busy talking about the new cards , bad timing for you , I posted abench of the only E8600 test I have seen, in the cpu section and only had 3hits
- but I have the same set up with a E6700 @ 3.6 the same speed as yours and same cards
-@790-1880-1090 I get apox, 16400-16500 3d-06 16950 @ 3.8
-so your out a 1000 points somewhere
-bench testing crysis maybe 5fps (for the avg.) from a low clock to the one posted using xp sp3 .Max fps =70 ,min ,16 , avg. 45-49 fps Island bench test
on 175.16's @1680-1050 all high no AA

-you might have a driver problem or a patch problem I stopped trying 1.21 ,stayed at 1.2
--I get about 14.3 sec. in super Pi 1m
(as a mother board speed reference)
-you have to use the default nvidia control panel settings for reference
-hope some of that helps


http://www.xfastest.com/redire...id=10995&goto=lastpost

-, check this out for your new cpu

Hmm well I am running vista 64 if that helps :\ could attribute to the 3dmark score.
what bugs me is, when i was on my asus p5b, my cpu scor was around 3300/3400 i believe, having used it in my sli setup on a evga 780i, its down to like 2900 :eek:

Also my crysis testing was only on the demo as I dont have the game anymore :
running modded drivers and they actually Upped my score from like 14.9k or something xD

Maybe i'll just go quad :

try the super Pi ,could be your 780i is lacking in it's settings , there are lots of voltage settings. or a memory conflict/bottleneck
 

Cuular

Senior member
Aug 2, 2001
804
18
81
If you are looking just to boost your 3dmark score, go ahead nad go quad.

So far for gaming going from dual to quad, has little to no boost to performance. A select few games like 3 or 4 actually know how to use more cores, and you'll see a boost. But a few games also actually take a performance hit with quads.

As to getting little to no boost from OCing the graphics card.

For the most part once again, OCing the graphics usually has just a small 2-5 fps boost. I'd theorize that the graphics cards are probably well optimized by the company's that made them. And until you know all the different parts and exactly what each clock does that you are OCing, that you shouldn't expect to see huge increases in fps from OCing.

Again there are a couple exceptions to that theory. Certain games scale almost to the percent, but by and large it has little effect on max fps.

 

Piuc2020

Golden Member
Nov 4, 2005
1,716
0
0
Why is 3DMark so important? It's way too reliant on your CPU.

Call of Juarez and Crysis are not very friendly to OCs, they don't benefit a great deal from them especially at higher settings.

If your CPU was limiting your cards (which I think it's not) you'd be able to crank up the settings a bit more without getting a frame drop but if you as much turn a shadow to high and get a drop in performance then yeah, your CPU is just fine.