• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

The 8800GTX/GTS are officially CPU Limited

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

dominick32

Senior Solid State Aficionado
Joined
Dec 19, 2005
Location
New York
I will run the 3m06 when I come home from work tonight but these are the 3dmark 05 results:

TONIGHT I WILL BE POSTING THESE ADDITIONAL FREQUENCYS - 1.8 GHz, 2.4 GHz, 2.66 GHz, 2.93 GHz, 4.2 Ghz, possible higher.

Conroe E6700 @ 2.93GHz


cpulimit.JPG



Conroe E6700 @ 4.2GHz



8800gtx1.JPG



So we have an official CPU bottlnecked card. THIS CARD IS A BEAST!!! This confirms that other guys post with the Opteron. Its not your GTS thats holding you back, its your processor bro. Dont let it get to you dude.

JUST A LITTLE MORE FUEL TO ADD TO THE DISCUSSION:
Forget about quads. Lets talk about clock speed here and 65nm architecture. In a single thread app or game, such as 3dm05 or "bogo mark" lol, once I upped the cpu clocks over 3.0 GHz, the 3d performance seemed to explode. Would that not indicate that the GPU has plenty of room left in it, simply waiting for more single threaded processing power? And lets say the GPU ran out of processing power, that would mean even if I increased clockspeed to lets say 4.2 GHz from 3.0 Ghz, there would only be a VERY VERY minute change in 3dmark05 points. At least thats what I found with ATI X1900 XT. Let me see if I can find some old benchmarks.

But if I remember correctly:
ATI X1900XT (STOCK CLOCKS)

E6600 Conroe @ 2.93 GHz = 12,200 3dmarks
E6600 Conroe @ 4.1 GHz on my Vapochill = 13,647


So, that would mean theoretically the X1900XT is maxed out of processing power and even by increasing CPU clock speed dramatically it did relatively nothing to my score.

Now here is my current benchmarks::
EVGA 8800GTX (STOCK CLOCKS)

E6700 Conroe @ 2.93 GHz = 16,438
E6700 Conroe @ 4.2 GHZ = 19,137


By analyzing the difference between both cards, almost duplicate setups, with exact clockspeeds. Doesnt that show that one GPU ran out of processing power, and the 8800GTX still has plenty on tap?
 
Last edited:
Can you post the details of ea. test? I predict there is a huge jump in the return to proxycon test, which is very CPU limited regardless of VGA.

dan
 
Dan0512 said:
Can you post the details of ea. test? I predict there is a huge jump in the return to proxycon test, which is very CPU limited regardless of VGA.

dan

I will post up the details tonight, as well as a lot more results.. Its not only return to proxycon. Its every benchy. And I do not run the CPU benchmark, I escape out of the benchy before CPU score runs.

More results when i get home from work.
 
And someone has been talking about SLI :p
Well.. at least krag is going kentfeild right off. :beer:
 
4.2/2.93 = 43% speed increase.
19137/16438 = 16% speed increase.

Are you sure that shows a CPU limitation? Post all the sub-scores it would be nice to see where it picks up the points, one bad thing about 3DM is it has a purely CPU score component. Just run the full test and take off the CPU score. Maybe do runs at more points and graph it out, the current range might overlap the point where it shows linear increase with CPU speed so we don't have a real idea where that happens.

As for quad, it would only matter for 3Dmark anyway for the time being.
 
MadMan007 said:
4.2/2.93 = 43% speed increase.
19137/16438 = 16% speed increase.

Are you sure that shows a CPU limitation? Post all the sub-scores it would be nice to see where it picks up the points, one bad thing about 3DM is it has a purely CPU score component. Just run the full test and take off the CPU score. Maybe do runs at more points and graph it out, the current range might overlap the point where it shows linear increase with CPU speed so we don't have a real idea where that happens.

As for quad, it would only matter for 3Dmark anyway for the time being.

Please do me a favor and read my previous post, I press escape before CPU score is computed.
Please read where I said:
And I do not run the CPU benchmark, I escape out of the benchy before CPU score runs.

It only computes GPU FPS scores, not CPU score. So please take the 2.93 vs 4.2 out of the equation.
 
Last edited:
wow 2800 points, these things are monsters, looks Like I will have to push the hell out of my e6600 to keep from bottlenecking.
 
I saw that. I don't know if 3DM is consistent when aborting it mid-test. If it was me I'd let the full test run and take off the CPU scores manually but if it doesn't matter then cool.

I don't know what you mean by 'take 2.93 vs 4.2 out of the question.' Isn't that the whole point of the thread? :) It could scale linearly across that whole range just not in a 1:1 ratio but...

You still should run tests at various speeds instead of two extremes only. Think about it, what if all or a majority of the speed increase comes before 4.2? What I mean is, it may not scale linearly all the way from 2.93 to 4.2 and the only way to know that is test speeds in between and see if they all fit on the same line.
 
MadMan007 said:
I saw that. I don't know if 3DM is consistent when aborting it mid-test. If it was me I'd let the full test run and take off the CPU scores manually but if it doesn't matter then cool.

I don't know what you mean by 'take 2.93 vs 4.2 out of the question.' Isn't that the whole point of the thread? :) It could scale linearly across that whole range just not in a 1:1 ratio but...

You still should run tests at various speeds instead of two extremes only. Think about it, what if all or a majority of the speed increase comes before 4.2? What I mean is, it may not scale linearly all the way from 2.93 to 4.2 and the only way to know that is test speeds in between and see if they all fit on the same line.

Tonight, just for you I will run @ 1.80GHz, 2.0GHz, 2.4GHz, 3.0GHz.... than my bigtime 4.4Ghz run!!! :beer: Just for you madman! And only you! lol
 
Fr3@k3r said:
BITE ME!!!

lmfao im not upgrading for another $1000 just to get a better score with my GTS and opteron LOL

altho it would be nice, i dont just buy the cards for scores... overclocking now and testing again :)
[/IMG]

Are you being sarcastic bro? I was just being sympathetic towards your concern in the other thread about your low scores. :bday:
 
yeah no question its his opty holding him back.

i loled myself because i saw "@4.2ghz" and wondered who had it. It was stupid, of course it was dominick ;)
 
dominick32 said:
LOL ^^^^^^
What's so "LOL"? The stock effective speed is 2.5Ghz(or was it 2Ghz effective?), 3Ghz is the max on most mainboards, so I say gun it! It'll help, it even helps a bit on my X1900XT when I have it at my current speeds.

It does not look to be completely CPU bound IMO.
 
jcll2002 said:
yeah no question its his opty holding him back.

i loled myself because i saw "@4.2ghz" and wondered who had it. It was stupid, of course it was dominick ;)

I know. I always forget that putting a number up like 4.2 GHz is useless in a comparison for 99% of you guys not running extreme cooling. I shouldve put something more realistic like 3.6 GHz or something. hehe
 
Fr3@k3r said:
altho im not getting many good 3dmark scores... im getting awesome framerates in gaming ;)

Tell me about man. I was just telling rattle in the other thread that I have never played a completed maxed out quality, high res, and maxed out filtering game of FEAR before that was smoother than 800 X 600 at low quality. These 8800 cards are truly incredible! I never congratulated you on the purchase. Congrats! :beer:
 
Back