OC'ing CPU- no difference?

squiffy · 23 Aug 2006 at 09:58

I've overclocked my 4400+ X2 from 2.2ghz to 2.7ghz and I don't notice any difference in 3D mark final results. Surely if the CPU is faster I should get an increase? ATI is at 662/774. Regardless whatever I do to CPU it stays around 18K in 3dmark 03. I would expect to be 19K when the CPU is overclocked? I did notice a increase when the GFX card was overclocked (17.x before, 18.x after)

oh is 1.350+ 200mV safe for 4400+ X2 (BIOS setting)? BIOS Vcore in power managment and CPU-Z reports that it's around 1.5-1.536.

Or would a better comparison speed CPU test be purely maths (prime 95)

Socaddict · 23 Aug 2006 at 10:15

have you got cool and quiet enabled?

check that its off.

iraiguana · 23 Aug 2006 at 10:18

3d mark does not always show cpu increases, i think it did in 3dmark 2001 but others less so, better would be to use super Pi or other maths programs yes.

squiffy · 23 Aug 2006 at 10:22

Socaddict said:
have you got cool and quiet enabled?

check that its off.

CAQ is disabled.

MeatLoaf · 23 Aug 2006 at 10:28

3DM01 and 06 are good for CPU testing, with 06 it gives an indevidual CPU score

squiffy · 23 Aug 2006 at 11:24

3d mark 05 cpu tests only

2.2ghz / mem 200 cas 2 - 5802
2.6ghz / mem 217 cas 2.5 - 6378
2.7ghz / mem 224 cas 2.5 - 6245 :confused:

--------------------
3d mark 06 cpu tests only

2.2ghz / mem 200 cas 2 - 1696
2.6ghz / mem 217 cas 2.5 - 2000
2.7ghz / mem 224 cas 2.5 - 2068
--------------------

ATI X1900XT at 662 and 774 for all tests. Ran 06 cpu tests as well that's supposed to support dual core?
------------------

3dmark 05 full test
2.2ghz / mem 200 cas 2 - 10959
2.6ghz / mem 217 cas 2.5 - 11630

manveruppd · 23 Aug 2006 at 13:55

Yeah, 05 is very graphics-card dependent. Not that having a faster CPU doesn't make a difference in games, but this is more noticeable at higher resolutions: you won't see a difference in 05 cause the unregistered version doesn't let you change the test settings so you can't set it at a resolution which would stress your CPU. I think the default is 800x600 or something like that, right?

TaKeN · 23 Aug 2006 at 14:22

No, its 1024 x 768

squiffy · 23 Aug 2006 at 17:32

manveruppd said:
Yeah, 05 is very graphics-card dependent. Not that having a faster CPU doesn't make a difference in games, but this is more noticeable at higher resolutions: you won't see a difference in 05 cause the unregistered version doesn't let you change the test settings so you can't set it at a resolution which would stress your CPU. I think the default is 800x600 or something like that, right?

So you think I should run the complete test set to maximum supported rez of the monitor for both 2.2ghz and 2.6/2.7ghz speeds?

fish99 · 23 Aug 2006 at 17:59

TBH apart from the odd 3DMark05 cpu test at 2.7 GHz, those numbers look pretty much as expected. You generally don't overclock the FSB 20% and see a 20% increase in benchmarks, it's usually less than that. Also the 3DMark full test just shows how much those tests rely on video card power.

Also looking at your numbers it looks like you're running a memory divider and higher CAS to get your overclock, which is going to absorb some of the expected performance increase.

manveruppd · 24 Aug 2006 at 00:00

squiffy said:
So you think I should run the complete test set to maximum supported rez of the monitor for both 2.2ghz and 2.6/2.7ghz speeds?

You can't, unless you have the registered version. But I'm sure if you ran an in-game benchmark you would see a difference in FPS at higher resolutions. As fish said though, not 20%!