Magical P4 performance?

[Leo V]

(updated: ) uploaded 3dmark2001se scores, placed 1st for ddr p4 (comparable GHz) systems using ti4200. (see below)

For some reason, my P4 seems to have 17% better clock-for-clock Sandra integer/MIPS performance than any other P4 results I've seen so far. Have a look at the MIPS... all other P4's perform similar to each other MIPS-wise, but significantly lower than mine. And I just saw SupermanCK's results using very similar setup to mine, and he still got fewer integer points (in his 150fsb test, I am still 10% ahead clock-for-clock).

I'm using 2.26B @ 2.55ghz, 1.50V (150x17.0) on Asus P4B533, Corsair XMS3200C2 @ cas 2-3-2-7. The high MIPS performance, however, was observed at various speeds (all the way to 2.62ghz max stable frequency @1.50V).

Any ideas? Did I forget something, or am I just very lucky? (These tests were run after booting up into XP Pro sp1).
--Leo

 

[SupermanCK]

at one point of my previous set up have had that result...but after i played with it some more it returned to normal results...
i wouldn't trust sandra 100%...

 

[TheHunter]

What about 3DMark2001SE? are the results thare still ahead 10% from similar systems?

 

[Leo V]

Well here are 3dmark2001se scores (using gf4 ti4200 128mb)...

various configurations w/3x drivers (lots of my test runs posted, they get better towards the end)

overclock w/40.41 drivers -- 1024x768x32, 13392 3dmarks

 

[Mikki]

Did you guys see the scores our friend SupermanCK is getting in his thread? :Q

 

[Leo V]

Yep... but he has a Radeon 9700! Hence he's Superman, and I am just a common man

 

[Mikki]

Me too!

 

[SupermanCK]

 

[TheHunter]

Check MadOnion O.R.B for similar system's... if the score is almost the same then something happend to sandra...

 

[Leo V]

Originally posted by: TheHunterCheck MadOnion O.R.B for similar system's... if the score is almost the same then something happend to sandra...

Done. (13296 3dmarks). I placed 5th out of 145 people using P4 (at 2.62ghz +- 1%) with Ti4200. The four people who score above me:

#1) Used PC1224 RDRAM, unknown ti4200 overclock (failed to provide 50% of results)
#2) Used RDRAM, much higher ti4200 overclock
#3) Another RDRAM user, unknown overclock
#4) PC1240 RAMBUS... gimme a break!

...which means I am The undisputed champion of DDR systems clock-for-clock performance in my gigahertz class, WHOOOO Maybe my P4 *is* magical!

UPDATE: UT2003 benchmark @1024x768x32: 153.3 / 67.1 FPS (using 40.41 drivers; I don't have the willpower to go back to 30.82 drivers...)

 

[LastRide]

I just noticed your settings

 

[Leo V]

320 core, 595 memory... the highest I can go 100% stable without memory artifacts. For everyday usage, I have moved back 3% (2.55ghz, 310/575) to have a safety margin and a saner mind.

 

[LastRide]

My visiontek I did 310/546 but for some reason I can't get 546 memeory anymore without getting tearing.Can only go to about 310-535 memory now.I tried taking off the HSF one time.I only tried to pry a little cause its has thermal epoxy on it.I wonder if that did something to it.I want to take it off and put AS3 on there.You know what I can use to take off the epoxy once I get the HSF off?.Also,I think I might have to put my card in the freezer in a zip lock bag for 15-20 minutes to get the HSF off.It makes me cringe at the thought of doing that.

 

[Leo V]

Sorry, the freezer trick is all I know... and it's pretty dangerous with thermal epoxy.

From my experience, you can also damage memory via overclocking. I once got a cheap GF2MX w/143mhz sdram, and tried to put the memory at 166. It failed badly, but later that day it started giving artifacts even at its native speed! (Even on the desktop.) I haven't a clue whether this was electromigration or plain cheap memory, but I'm 90% sure the overclocking made that card unusable.