Presler 4.5ghz Cedar Mill 5ghz club

[Ross]

Yes, and vice versa regarding P95 & PC04. What bugs me to no end is people that *insist* 2xP95 for 12+ hours is "the only" stablility test worthwhile....especially when they can't finish a PCMark04 with a score at the same OC that passed their 12 hour Primes. The point here is, PCMark04 will tell you a lot quicker if the OC is even close to stable.

As far as I'm concerned, P95 for days on end is only useful for 1) people who are going to use that rig/OC for folding or 2) people who are going to use that rig/OC for work. Short of that, run a PCMark04 for a quick test, which takes all of 5 min and will let you know if you are even in the ballpark.

Is it possible that an OC would pass a PC04 and fail P95 in a couple of hours? Sure. Is it possible to pass 2xP95 for 12+ hours and fail PCMark04? Sure. When you get to an OC you think you want to run daily, PCMark04->couple loops of 3DMarks->PCMark04 again (hour or 2). Or if you prefer, PCMark04->2x P95 as long as you want->PCMark04 again. If your system can do all those and scores on the final PCMark04, that's one very stable OC.

If you don't want to go through all of that crap like me: PCMark04 stable OC minus 100-200MHz or so = stable daily no problem (work and games)...

 

[pscout]

I like to take the best of all worlds myself

-pcmark04 (now that i have learned its value) for the initial tests
-or when in a rush (usaully?) just a simple sandra bandwith to prove the rig will do something after booting at the current OC
- once both of the above are ok and i have pushed my cpu to the max, then i prime blended at least half way through the second test
- then i run dual folding ... initially with network disabled/disconnected (i like disabled best since i have som many cables i hate to touch em if i dont have to)
- after 4-8 hours i re-enable network so my WU's can turn in and keep going
- after 2-3 weeks of dual or single folding depending on DC or HT, then i call it stable.

But i am just a simple folding farmer ... so that's what I do.

I believe we all have multiple mixed purposes for our OC efforts, and I doubt that 1 method can ever suit all.

So why worry about it ? we can all be right!

/edit ... to Ross's point ... having the luxury of multiple rigs I can just not run my main/work rig on the edge ... the others are just a kvm switch push away ...

 

[proth]

From what I've been noticing, Prime95 pushes the cpu more that PCMark04. IHS temps are +3c more running prime. So it sounds like prime is better for CPU stability and PCMark is better for "system" stability. From what I've also seen about prime is if it runs a loop or 2 after reaching max temp, it will run forever...

 

[pscout]

From what I've been noticing, Prime95 pushes the cpu more that PCMark04. IHS temps are +3c more running prime. So it sounds like prime is better for CPU stability and PCMark is better for "system" stability. From what I've also seen about prime is if it runs a loop or 2 after reaching max temp, it will run forever...

personally, i think folding pushes things even better than prime ... but you have to be careful not to screw up the science by allowing your rig to turn in a lot of failed work units.

This does not diminish the value of any other stablilty tools.

BTW, prime95 has a scientific purpose too if you join GIMPS!

 

[proth]

Scout,
Thanks for the heads up. I'll be checking out protien folding after my tools are put away and the case is finally buttoned up...LOL

PS: That could take a year or so

 

[mjgent]

personally, i think folding pushes things even better than prime ...

Especially folding two QMDs at once. HOT!

 

[pscout]

Scout,
Thanks for the heads up. I'll be checking out protien folding after my tools are put away and the case is finally buttoned up...LOL

PS: That could take a year or so

LOL ... i hear ya ... when you button it up you can put it to work on the farm while you work on its sucessor

 

[ivanovich]

benching 3dmarks 03 at 5127mhz now .nearly 11000 with a 6600gt .not bad i would say still having trouble with pcmark though? fails in grammer check , test 3 no matter what clk i set . http://service.futuremark.com/compare?2k3=4574268

 

[Ross]

Have you tried running PCMark04 at default clocks/settings to see if it passes there? Maybe you're just not lowering the clock enough.

 

[hyperasus]

I see there's alot of talk about pcmark04/prime95. Back in the good ol northwood days it seemed that p95 did the best job at checking stability. My northwoods would pass pcmark04 and fail p95. These dual cores are different. I have no idea why but they are. I've yet to see any of my pentium D's that wouldn't pass overnight dual p95 and then fail pcmark04. I'm not saying it couldn't be the other way around for someone else, I'm just saying that all 9 PentiumD cpu's that I've worked with were that way. I don't usually even run pcmark04 all the way threw when I'm checking an overclock. All you need is the first 3 tests to narrow down an overclock. For that matter, all you need is test 2 & 3. After it completes test three I hit abort then click Detail, then Results. If there are no failed tests then the overclock is pretty stable. Once I narrow down the overclock with pcmark then I use dual p95 for about an hour just to make sure my cooling can keep up when running at full load for long periods of time.