Q9300 Benching Results

I installed Windows XP on my Q9300 rig last night and did some more testing with it. The highest that I have ever run the system is 3600 MHz (480 FSB). It is Prime95 stable for at least 1 hr with the settings that I have, but it is not Folding stable at that speed. This is the speed that I originally benchmarked the CPU at last week.

Anyway, I decided to see how far I could push it using SetFSB, which actually works with the P5Q Pro mobo. I upped the FSB to 490, CPU at 3675, and I was able to run SuperPI, PIFast, wPrime, and UCBench without any problems. I then upped the FSB to 500, CPU at 3750, and again I was able to run everything without any problems. When I upped the FSB to 510, CPU at 3825, the computer instantly locked up. RAM is good for 533, although I might need to up the voltage some. Currently at 2.0v, and it’s rated for 533 at 2.1v. The CPU voltage probably needs to be increased as well. I ran out of time last night, so I will see if I can get to 4 GHz tonight. Keeping my fingers crossed.

Here are my previous results (Win 7 @ 3600) vs my current results (Win XP @ 3750). Rank is in ().

SuperPI – 14.258 (35) – 13.640 (25)
PIFast – 27.13 (28) – 25.81 (18)
wPrime – 11.091 (33) – 10.344 (15)
UCBench – 711 (5) - 764 (3)

I tried increasing the FSB over 500 MHz, but nothing I do seems to be working. At 500 FSB, I can run as many benchmark tests as I want. But at 505 FSB, the computer locks up even without any load on it, usually within 10-30 sec.

I tried increasing the CPU, PLL, FSB, NB, and DRAM volts. I tried relaxing the memory timings. But nothing seems to help. Does anyone have any experience with the Q9300 or P5Q Pro?

Probably needs moar cold, but I have no experience with either....

more fsb voltage may help but i assume this is air so like super says more cold will help.

i had to really up my fsb/nb voltages on my p5q-pro to get it into the 550+ ranges.

I made to 510 FSB (3825 MHz) last night. I really had to crank up the FSB and NB volts to get there. At 500 FSB, I can use 1.38v FSB and 1.42v NB. At 510 FSB, I had to use 1.50v FSB and 1.52v NB. I put a large box fan blowing at the mobo to help cool everything down.

It was only semi-stable at the settings I had. It never locked up or rebooted, but I was not able to run SuperPI without getting calculation errors. I was able to run wPrime 32M/1024M and UCBench without any problems.

I don't know if wPrime and UCBench are easier to run than SuperPI, or if they simply do not have error checking. Does it mean anything that I was able to run one but not the other? CPU voltage or FSB/NB voltage?

Edward2 said:

I don't know if wPrime and UCBench are easier to run than SuperPI, or if they simply do not have error checking. Does it mean anything that I was able to run one but not the other? CPU voltage or FSB/NB voltage?

Click to expand...

The issues you are seeing there are consistent with what I have found in the past when I run a CPU at its absolute max speed and it's right on the edge of stability.

Usually SuperPi 1m errors out well before UCBench 2011 and wPrime32m.

My experience has been with socket 939, but I'm assuming it wouldn't be any different with another socket type.

I'm not sure exactly why though, might just be that UCBench 2011 and wPrime are running a different type of benchmark that is not as susceptible to instabilities or it could be that they are performing access to the CPU in a way different from SuperPi 1m.