Safe Temperature range for XP 4200+? (Overclocked)

[lodog00]

Forgive me if this is covered somewhere, I did a search already but didn't quite find what I needed.

What's a good "safe" temperature for the AM2 4200+? Mine is overclocked to around 2.7GHz (247 fsb/1.475v). I am using the Asus M2N32-SLI Deluxe Wireless mobo.

Speedfan, if I am reading it correctly, lists idle temps at about 27-30C, which I'm OK with. When I run two prime95's, it goes about as high as 54C.

I am using a Lian Li 201B Server case (the one reviewed here), and a Zalman CNPS 9500AM2 Heatsink. The heatsink orientation is set to match the airflow of the Lian Li intake.


Also, how do I go about deciphering Speedfan's temperature listings? I see a bunch of temp1, temp2, temp3, etc...and other than Core and HD, I'm not sure what these temps are. I ask because one temperature listed under Temp3 is spiking up as high as 93C!!!! Everything else seems OK.

The thing is, my machine is completely stable when ONE Prime is running (I've run it for over 24 hours with no problems) but when TWO are running, one Prime will fail because of a "hardware error". This unknown temperature going to 90+C..... whatever it is, it sounds as if it's overheating! The only thing I can think of is that it's reading temps for my graphics cards (quad-sli/dual 7950gx2s). But even then, 90+C seems rather....ridiculous.

Thanks again.

 

[Hochi]

Those temps are fine

 

[myocardia]

That's the temp of your mosfets. And 93C is a good temp for them.

 

[lodog00]

OK, thanks. I couldn't figure out what the heck that 90+C temp was.

Any idea why one instance of Prime would fail while the other one continues to run just fine? Again, I can run ONE instance all day, but when I run two, one will fail within 5-10 minutes (not always at the same place).

Thanks again.

 

[videopho]

Back down your clock speed to stock until both instances start passing simultaneously. Also make sure you have all the latest AMD driver and optimizer in place. The system is not considered stable until both instances of Prime95 (Torture test) running at least 12-24hrs w/o any failure. I'd run Memtest86 first, then Super-Pi(32M) second prior to Prime95 to shave off sometime.

EDITED: I copied this from the sticky (courtesy from John) for you to follow (if not has done so)

# Prime95
# create two shortcuts
# right click one shortcut, and at the end of the target line put a blank space and add -A1
# open both instances
# go to advanced > affinity > uncheck let program run on any cpu and leave the 0 in the box and put a 1 for the second instance
# option > torture test > custom > memory to use in (MB) should be half of your system memory

And keep an eye on which instance or which core that may fail Prime. It's early to evict one of your cores may have been bad.