Why don't I get better results in 3dmark after overclocking?

[ThirddOracle]

I recently ran 3DMark06 and i got 3415p with my X1800 XT 512MB. I wanted more, and ofcourse the only way to get more was to overclock. I'd heard a lot about how good ATITool was, so i downloaded the 0.25 beta 14 version. The core and memory speed on my X1800 XT 512MB was 590/690MHz. I overclocked it to 640/751MHz. After that i ran 3DMark06 to see if it went to the better. But it didn't. It's no difference. I still got the same results, and the same fps. Someone told me that i was only changing the 2D clock and not the 3D clock. Is that the problem? If it is, how do i set it to the 3D clock?

 

[spy2520]

i dont get too much of an increase at 675/820 on my x1900xtx either.

 

[FLY3R]

That at some cards just run faster at the right clock><mem speed. Somthing about the right number just working better.

 

[markkleb]

I recently ran 3DMark06 and i got 3415p with my X1800 XT 512MB. I wanted more, and ofcourse the only way to get more was to overclock. I'd heard a lot about how good ATITool was, so i downloaded the 0.25 beta 14 version. The core and memory speed on my X1800 XT 512MB was 590/690MHz. I overclocked it to 640/751MHz. After that i ran 3DMark06 to see if it went to the better. But it didn't. It's no difference. I still got the same results, and the same fps. Someone told me that i was only changing the 2D clock and not the 3D clock. Is that the problem? If it is, how do i set it to the 3D clock?

Did you save your settings? You may have turned off the ATI tools and taken the 3Dmark test with the factory settings.
On my X800 I always see an improvement when I OC.

 

[crunchie]

I have almost the same problem. Have an X800 Pro with stock @ 475/450 I get 17,400 on 3DMark01 and when I OC to 550/500 I get an extra 300 points. Doesn't seem like a great return .

 

[trog100]

dont use 2001 crunchie.. its system limited not grfx card.. tooo old and tooo higher frame rates.. use 2005...

2001 is good as a cpu/memory benchmark but not for grfx cards.. for example i get nearly the same 2001 score with my x850 card as i do with x1900xtx..

my x850 scores about 29000 my x1900xtx scores about 31000..

the high scores are down to my faster system.. when u get to 500 odd frames per second as in 2001 the grfx card aint the bottleneck any more its the rest of the system..


trog

ps.. and orical u are getting the same score cos your card is going the same speed.. u are just changing the 2d speed.. when a 3d apps fires up the ccc thing just reverts back to the default 3d speed.. reading the atitool stuff should tell u how to do it correctly..

 

[Steevo]

06 is highly CPU dependant also.


Just look at system setups that scored the highest.

 

[trog100]

"06 is highly CPU dependant also"

06 is bit different cos it does a seperate dual/single core cpu test and adds that to the overall score.. my system only manages about 5000 in 2006.. a slower but dual core cpu would give me over 6000.. not because any of the "game" test run any faster but the cpu test would just double my score.. the same dual core cpu would drop the score thow in 2005 or 2001 then u have the big penalty for not having sm3..

i spose u have to know how to read benchmarks.. where low frame rates of say 50 40 or 20 are concernd its the grfx card thats the bottleneck when the frame rates start to go up the rest of the system starts to play a bigger part..

if a game runs highs of say 150.. lows of say 30 with an average of say 70 its the lows that bugger the playability up..

if u speeds your system up with faster memory or a faster cpu your average game benchmark score might go up but it will be the highs that just get higher the all important lows wont alter much at all..

if u do it the other way around even use a slower cpu and memory but fit a faster grfx card.. your average score could be the same but it will be made up in a different way.. the lows of 30 could go up to 40 or 50 the highs of 150 stay the same or even get lower the average could still be 70 but the game would be perfectly playable..

benchmarks dont mean a lot unless u know how to read em.. the bottom line being its the grfx card that governs those all important lows and the rest of the system the dont matter a sh-t highs..

reviewers tend to use average benchmark scores which is why too much emphasis is placed on so called "gaming" cpus.. they do appear to win with any average fps score but not in a way thats matters.. its always those unimportant highs that get the boost.. not the lows which are the only ones that matter in real world gaming..

trog

 

[Steevo]

Our 3.6Ghz Intel server at work has a X600, it scores 2K or around in 3Dmark60, however I score around High 4 low 5's with a 2.95Ghz single core 4000+ X1800XTPE OC'd.


This proves to me that the benchmark is unfairly leaning to dual core CPU's, and therefore has a heavy CPU weigh in.

 

[crunchie]

Thanks for the reply trog. I will d/l it and give it a whirl. btw, 2001 doesn't recognise that I have directx 9.0c, but shows that I have only 8 installed.
Also, my 9800Pro was getting in the mid 17's too.

EDIT. OK. I installed 2005 and ran it @ stock and got 4998 points. Ran it @ 550/500 and got 5551, so at least I am showing a decent increase in score now . Have had this card to 595/525 but only have the stock cooler so don't want to test it at that clock. Am going to be getting an arctic cooling silencer 5 for it soon. Will also check if I can unlock the extra 4 pipes whilst I'm at it. Hopefully it hasn't been laser cut.

Sorry for hijacking the thread.

Edit. Got the AC silencer 4 and am now running @ 550/540 with a score of 5814 on 05, so am happy with that. Cannot unlock the extra 4 pipes, but that's life .