SP2004 question

[shinzwei]

when i run stress CPU it fails after 1 minute.....but when i run the blend test i ran it for an hour then i stopped it cuz i wanna use my computer. why?

 

[Matt2]

are u overclocked?

 

[shinzwei]

yes Oc'ed

Opty 144 @ 2.7ghz 1.58V

 

[GML3G0]

Could be a bunch of different things. CPU can't handle it, memory controller crapping out, memory timings need tweaking. You have to find the source of the problem.

First of all, to use SP2004 correctly, use the custom test. Set the minimum FFT size to 8 and max to 4096. Uncheck "Run FFTs in-place" and set the memory to use to about 700-750MB so it doesn't use the page file.

If it can't pass 16+ (24+ recommended) hours without error, then your computer is not stable.

You can use Memtest86+ off a floppy, and if errors, chances are it's the RAM / memory controller, but I've seen MemTest86+ go on for hours without error, even while unstable.

1.58v. That's a lot of voltage. What are your load temps? Considering backing down a little bit. You're putting that CPU at a lot of risk with that kind of voltage.

Good luck. :beer:

 

[Megatomic]

Originally posted by: GML3G0
First of all, to use SP2004 correctly, use the custom test. Set the minimum FFT size to 8 and max to 4096. Uncheck "Run FFTs in-place" and set the memory to use to about 700-750MB so it doesn't use the page file.

Thanks for the tip. I'm going to give that a try. :beer: for YOU!

 

[Shimmishim]

errr...

most people run sp2004 by doing the small fft test.

i find that small fft is best for testing cpu stability and large fft is for mem.

i've had a chip do 3.0 and run large fft for 6+ hours but fail small fft in 1 minute.

the FFT length for small is 8 to 30-something i believe

 

[Megatomic]

My system will run small FFT prime95 (two instances with affinity set manaully) all day long on small FFT with CAS set to 2.5. But it fails within minutes on long FFT with CAS set to 2. Bad memory or bad CPU?

 

[RichUK]

Originally posted by: Megatomic
My system will run small FFT prime95 (two instances with affinity set manaully) all day long on small FFT with CAS set to 2.5. But it fails within minutes on long FFT with CAS set to 2. Bad memory or bad CPU?

Run and pass memtest for preliminary memory testing. Then run Large FFT's within Windows to test the RAM when operating under the OS.

As said before:

Small FFT?s = CPU

Large FFT?s = Memory

 

[Megatomic]

RichUK, I've done both and at CAS2.5 I have no problems. It seems I missed the fine print on these memory sticks of mine, CAS2 is only for Intel systems. AMD gets CAS2.5. Some guys in another thread clued me into that late last night. Sucks, but my system performs really well even at CAS2.5.

 

[shinzwei]

which is better for OC? Blend or CPU Test.

 

[crazycarl]

i think there is a bug in sp2004. it will pop up the windows 'sorry this program has to close...' message after exactly two minutes, but then not close. in fact exiting the program, the process continues to run and eat cpu cycles and ram.
very bizarre. i guess i'm just going back to good old prime95.