Do you give a rats ass about Prime95?

[gistech1978]

i use p95.
but i dont run it for hours and hours.

 

[batmang]

you guys do know, i made this thread because i am jealous.

i want 3.2ghz, not 3.0ghz, i figured if i got enough people to say prime95 doesnt matter, i could run at 3.2ghz without guilt, knowing that its really unstable, cause prime95 said so.

 

[ErikS]

Use Toast, it works the CPU harder than anything else I've ever found.

 

[beatle]

P95 isn't the end-all-be-all of testing programs. It's nice because it gives you a report on how long it ran and it loads the CPU AND memory heavily (take a look at the memory usage!) I ran my tbird 1.1 @ 1333 for over 2 years without passing P95 overnight. Everything else was stable, however.

 

[valky]

P95 isn't the end-all-be-all of testing programs. It's nice because it gives you a report on how long it ran and it loads the CPU AND memory heavily (take a look at the memory usage!)

Memtest is more comprehensive............. :beer:

 

[wicktron]

If you have an HT P4, run 2 instances of P95 with affinity set to each logical CPU and run 3DMark in loops on top. If it passes that for 12+hrs, it's stable to me.

 

[aka1nas]

Memtest doesn't catch some errors that prime does. I start testing with memtest86 immediately following the overclock, if it gets errors in that, then I know right off that it needs to be adjusted. Once I get it stable in memtest, then I go into windows and have it run prime95. If it can do prime for an hour or so without error, then I usually leave it running prime overnight with 3dmark2k1 looping as well. I also leave mbm5 on so I can log temps and see if it gets to a temperature that i'm uncomfortable with after long periods of stress.

 

[Gerbil333]

Yea. Each time I build a computer or make a major change (RAM, CPU, motherboard, overclocks), I run my own set of stability tests.

I use the following simultaneously:

-Prime95
-Winamp 5 set to play random songs in a 1000 song playlist
-3DMark 2001 SE set to loop the benchmark infinitely

I do that for about 12 hours. Then I close 3DMark and use the computer for a bit, which involves surfing the net, e-mail, Paint Shop Pro 8, and games (GTA3, NFSU, Tribes 2, FarCry). Next, I launch Tribes 2 and host a server with 6+ bots and observe the game for several hours (minimum of 4 hours)...T2 is coded pretty bad and tends to crash on seemingly stable machines, so it's a fairly good indicator of stabiltiy. By this time, it's usually been at least 24 hours of straight Prime95 testing with Winamp and many other programs, along with 16+ hours of 3D testing. If it passes all of that, I declare it stable.

I think that works pretty good.

 

[NesuD]

Yes i do.

 

[lookin4dlz]

NO!

My setup would probably "fail" after 3-4 hours (at least that's what I recall), but I give it daily hard use crunching stock price info & playing civ3 with the comp running for month after month after month w/ no problems.

 

[jjyiz28]

Originally posted by: aka1nas
Memtest doesn't catch some errors that prime does. I start testing with memtest86 immediately following the overclock, if it gets errors in that, then I know right off that it needs to be adjusted. Once I get it stable in memtest, then I go into windows and have it run prime95. If it can do prime for an hour or so without error, then I usually leave it running prime overnight with 3dmark2k1 looping as well. I also leave mbm5 on so I can log temps and see if it gets to a temperature that i'm uncomfortable with after long periods of stress.

when i o/ced from 11x133, to 11x166, ram speeds and timmings the same, i fail memtest #5 only, but i pass prime95 for hours on end. 3dmark loop stable too. eh

 

[FPSguy]

I don't see why Prime matters if your computer is stable running the programs you run. Using the elevator analogy above, I don't care if the elevator only holds 1,500 pounds because I only let 2 or 3 people on the elevator at a time, and I don't know anyone heavy enough to break the elevator under those circumstances. If I meet someone that heavy, then maybe I will pay attention to Prime95. Not before.

 

[DarkKnight]

i always use prime 95 cause u never know when ur comp will mess up cause of ur hardware.

 

[uncleX]

I only torture test the computer because I don't want it to lock up when I do some things. Like I don't want it to crash when I'm filling out my taxes on April 15th. I don't want it to crash before I save the game if I have just spent four hours getting through a tough spot. Now, if it didn't matter if it crashed, or I knew it wouldn't crash when it mattered, then I wouldn't care if prime95 crashed.

I figure prime95 is mainly a way of speeding it up if it is going to crash at all. If my computer is going to crash a couple of times a week, I'd just as soon have it crash in a half hour with prime95.

I have run prime95 for several months though to test for primes. If it crashed after 5 minutes, it would be pretty pointless.

If you are a gamer and you don't have problems with your game crashing, or it doesn't concern you if it crashes occasionally, I really don't see why you should care about prime95.

 

[SuctionCat]

For me, Prime95 is a necessary but not sufficient condition for stability. I'm happy at the 12 hour mark or so, but I also use Memtest as well. Apart from that, I've got some simulations I use for my work that thrash the beejesus out of a CPU, so they're good as well.

 

[maluckey]

I use Prime95 to tst stability, but I've had apps crash XP (mainly games like UT2003, or IL-2), after passing Prime95 for 24 hours. Prime95 is the first stability test I use, but it doesn't do diddly for the AGP interface, so it is only good for the CPU and RAM.

It's not as good as the new and improved Memtest86 fo the RAM, so it's only real benefit to me is to troubleshoot CPU instabilities. If it's stable in Prime95, I can look somewhere other than the CPU for the cause.

I've found that looping UT2003 botmatch with 16 bots in DM Inferno, while running Prime95, separates the men from the boys waaaaaay quicker than days of Prime95 alone.