Prime95...

[Blargh]

How stable is a rig that fails primes torture test at 6 and a half hours?

I ask because right now im ocing my rig (p4 1.8a, asus p4t533-c, 512mb samsung pc800) to 2.48ghz pc830 with 138/34 fsb 3x multiplier (id go for pc 1066 but prime crashes in under an hour no matter what vcore i set) and it seems faily stable, tho fails out in the prime test at around 6 hours 30 minutes. This good ebuf to be considered stable? Or should i go back to 133/33 at 3x for 2.4ghz and pc800 :/ ?

 

[Duvie]

I would consider that stable for most of the apps I run...divx encoding etc....Maybe if I was run seti or some other computing program taht cracks away 24/7 I may what this a little better....


I think you are fine...Was it a simple rounding error??? It is likely you could of ran several other torture test and not even had a failure at 12hours....

Too bad you couldn't get the pc1066 to be stable!!!

 

[Actaeon]

In my opinion, stable enough.

You aren't going to be using a CPU that goes with $100% load for 6 hours straight... and plus, Even though you failed it at 6 hours, it didn't mean you crashed, which imo, its when its time to lower the clock speed.

 

[Blargh]

Wow that was a quick responce

Yeah Duvie it was a rounding error, and as far as the pc 1066 thing goes its gotta be the motherboard, as it wouldnt even load at pc1066 with the original mobo bios that came with the board.... so i went and updated to the newest beta bios (at the time 1005.001) and it would run at pc 1066 completely stable ... but it wouldnt detect all of the memory i had (would only detect like 128 or 192mb when i have 512), so i went and updated to the highest non beta bios and now it will detect all 512mb of the rdram... but it wont run stable for more then an hour on prime. So im guessing that maybe the next bios might fix the problem... i hope anyway. Thanx for the replies guys.

 

[Den]

I have the same MB with Kingston PC1066 RDRAM, and once I manually set the multiplier to 4x, it runs PC1066 fine with the original bios...

 

[Blargh]

Originally posted by: Den
I have the same MB with Kingston PC1066 RDRAM, and once I manually set the multiplier to 4x, it runs PC1066 fine with the original bios...

Yeah when i went for pc 1066 with my pc800 samsung i used 4x too....... like i said, wouldnt work with the original bios, would be stable with beta 1005.001 but would show incorrect ammount of memory, and the newest non beta bios 1003 will show the correct ammount of memory but wont be stable for mor ehten an hour in prime 95.

As for kingston pc 1066..... if i HAVE to i might buy it but i really would rather not having to spend $300 for the pc 1066 equivalent of what i have now. Speaking of i have a thread about this in the motherboard forums here

 

[cbuchach]

I don't consider any overclock stable that fails Prime95 after any amount of time. I usually go for at least 24 hours to make sure my overclocks are OK.

I guess my opinion is that if your system crashes or you have strange things going on with it, then you never know if this is due to an unstable processor.

 

[paralazarguer]

I would tend to agree. If you can reproduce errors in prime 95 after any ammount of time that means that with enough time and pressure your system will corrupt data which is not fun. I would call that not stable as it could malfunction with climate changes especially.

 

[gizbug]

Although I don't overclock my system, I like to run it when installing a new cpu/mb.
I have had it running now for 9 hrs on my home computer, with my new Athlon XP 2100+ and my Abit KX7333 mb.
So far nothing has failed, and it's running great. The CPU temp is steady around 50C.

How long should I run this test for? Or is this long enough to determine everything is stable ?

 

[gururu]

any computer can be made to crash...

so my first impression is that you are probably ok as Duvie said. You do have a lot of space between 133 and 138 however,
maybe compromise and go for 135, 136; maybe you'll get your 1066 going too.

do your temperatures stabilize while running prime95? Or do they keep gradually increasing over the 7 hour period?
they should stabilize.


gururu

 

[THUGSROOK]

6.5 hours = passed

 

[SupermanCK]

so you are overclocking your rdram right? you only have pc800 and overclocking to pc1066...sounds like to me your ram might not be holding out...at that high bus

 

[Duvie]

It depends on use....server no....computer will valuable information no....


However I had an athlon cpu that would fail frim at 6 hours...I ran it for 4 months in apps that ran 7-8 hours full blow and never had problems....

This is user preference....

For me I like 12 hours....I ran my 1.8@2.4ghz default vcore for 18 hours but that is cause I was selling it and wanted to make sure...

 

[WarCon]

Wanna hear a horror story. My dead (it runs but gives random errors) 1.6A, would run Prime95 for 1hr or 5hrs or sometimes 8hrs, but would always fail. (Sometimes wasn't even rounding errors) Or my system would randomly reboot from that beast. But playing Mechwarrior4, I could rarely finish a single mission before it would error out. Try some other stuff that will stress the processor and memory and video card to see if its really anywhere near stable.

The moral of this horror story is that anything over 1.75vcore is BAD for your P4.

My new 1.8A runs Mechwarrior like a champ..........

 

[THUGSROOK]

yea Prime95 aint the "end all be all" of stability tests - ive had the same results....
10hours prime stable, but still crashes during certain simple operations.

ive found that looping X-Isle demo for 5-6 hours is a much more thorough system test.
not even 3dmark2k1se can compare to it.

ive really lost faith in testing anything for over 6 hours anymore.