Help me find out wtf is going on here...

Ok this is what I have:

1800XP @ 2GHz
Abit NF7-S rev2.0 (17 bios - came default)
2x Hyper X 3200 BH5 Dual Channel
9500PRO @ 360/329
Enermax 431w PSU

Here are my voltages and temps:

Vcore = 1.725
Vmem = 2.8
Vchip = 1.7
Vagp = 1.5

CPU = 46c
Case = 31c(PSU diode stuck between ram)

The side of the case is open as well. Ambient temps are 21c.


Now here is something really messed up. When I clock my rig at 9x222(1998 eff) I get the following results:


Gun Metal Benchmark 1 -- 23.48fps average.
Gun Metal Benchmark 2 -- 30.30fps average.

AquaMark 2003 -- 36,609 pts OR 36.6fps avg.

3dMark2001SE 330 -- 15,076 marks


Now when I set it to 10x200 this is what I get:


Gun Metal Benchmark 1 -- 23.86fps average.
Gun Metal Benchmark 2 -- 30.82fps average.

AquaMark 2003 -- 37,288 pts OR 37.29fps avg.

3dMark2001SE 330 -- 15,087 marks


Ok the 222 fsb should kill the 200 fsb scores. But why aren't they?!?!?! How can something slower, outperform something thats suppose to be faster?? Any tips??

omg how is this possible, according to cpu-z my fsb is fluctuating between 221.7~216 fsb. WTF!!!!!!!!!!!!!

The cpu is running at the same speed in both cases, 9*222 ~ 10*200 ~ 2000Mhz. The memory bandwidth will be different but performance that uses the cpu will be dominated by the cpu speed.

So are you saying that the 2MHz diff is making the difference in the benches I use? I always thought that the higher fsb will always be faster regardless. Why is it that ppl suggest you run at high fsb with low multi as opposed to higher multi and low fsb.. Im confused now.

No, my point was that I see no difference in your results. Graphics are dominated by the cpu speed and I see "little to no difference" in your test results.

Have you tested to see what variation is from one test to the next? I've not used those particular benchmarks but my guess is that the variation would easily be accounted for by what you would see from one test to the next. Do 5 tests at one setting to see the variation and its range are.


Edit: to clearify

Are you running your memory sync at both FSBs?

Rezin777 said:

Are you running your memory sync at both FSBs?

Click to expand...

Yes using the 6:6 ratio