PDA

View Full Version : Temps for Socket 939 4200+?


CaptNKILL
11-27-07, 01:36 PM
I'm selling my old system to a friend and I'm running Orthos on it to make sure its stable. Its been so long since I've used a stock cooler that I'm not sure if these temps are right.

Using small FFTs on a Socket 939 4200+ (Manchester core I think) with some fresh Arctic silver 5 applied and the original stock heatsink the "core" temp in speedfan is hovering around 64-65C with the side of the case on. Its been running for 10 minutes and the temps aren't going up. In fact it just dropped to 63.

It was a bit hotter with plain old thermal grease on it.

It seems stable, and I think these temps are alright given the crap cooler and the small FFTs test, but I just want to make sure. The other temps listed in speedfan are all in the mid to high 40s. Its possible it isn't reading the core temp right though, because it just doesn't feel that hot in there.

I was getting some blue screens last night with random files and I think it was related to the newest nvidia beta drivers (its running a 7900GT), but I want to narrow things down.

stevemedes
11-27-07, 01:40 PM
If those temps are while it's stressed in Orthos you'll be fine.

Tr1cK
11-27-07, 01:44 PM
I'd make sure I had an intake fan on the front of the case.

I gave a buddy my old AXP 2400+ and 6800GT. He was having lockups due to the 6800 overheating. He had an 80mm fan in the back for exhaust, but none in the front for intake. I moved it to the front and it works without a hitch now and runs much cooler.

Proper cool air intake is just as important as exhausting IMO.

CaptNKILL
11-27-07, 01:51 PM
Yeah I have 1 80mm intake and 1 80mm exhaust. The side window has vent holes in it too which probably helps the CPU fan a bit.

Its still running fine. 65C under orthos (dual core) small FFT load for 25 minutes, no errors. No crashes or blue screens either.

Which is actually a bit anoying... I haven't changed anything since I had problems last night but NOW its working fine. I hate problems like that. I'd rather that I just had an out right bad part than have to worry about it screwing up later after someone buys the system.

Tr1cK
11-27-07, 02:16 PM
Memtest?

Also, I wouldn't sell it overclocked any if I were you. (If it even is.)

CaptNKILL
11-27-07, 02:25 PM
Memtest?

Also, I wouldn't sell it overclocked any if I were you. (If it even is.)
It isn't overclocked.

I ran memtest on it for 2 hours last night and it passed without any problems. It did the full range of tests.

Strangely, I just ran Orthos Blend test and it errored after about 10 minutes. I checked the BIOS and the ram was set to 2.6v for some reason (apparently thats the BIOS default even though most ram is 2.5v). I'm hoping that it just overheated and errored. I set it to 2.5 and I'm running the test again. Its been going for 10 minutes now with no errors, but I'll let it go for a while...

F***...

It just errored. :|

Is there anything else that would cause Orthos to error in the blend test and not the small FFT? :(

I checked CPU-Z and the memory seems to be running at the proper timings. I'll try loosening them up though.

Specs:

DFI Lanparty UT Nforce 4 Ultra
Athlon 64 X2 4200+ (stock cooler)
MSI 7900GT 256Mb PCI-E
1Gb Corsair Value PC-3200 (2x 512Mb)
Antec Smart Power 2.0 450W PSU (brand new)
Western Digital 120GB SATA

DRen72
11-27-07, 02:30 PM
I recall the temps of my old 4800X2 on the stock cooler and that is just about the exact same. I used a Thermaltake Blue Orb on it after a while.

CaptNKILL
11-27-07, 02:56 PM
Well, I lowered the timings a bit (from 2.5-3-3-8 to 3-4-4-8) and the blend test has been running for 19 minutes without a hiccup.

I've got my fingers crossed. I think that might have done it. The memory is rated at CAS 2.5, I don't know why it'd need to run at 3... but whatever. If its stable like this for a couple hours I'll be happy. Latency does squat for performance anyway.

I can't wait to just sell this machine. I'm waiting on another gig of memory then I can get my $475. Not a bad machine for the money really. :)

EDIT: 46 minutes... still error free.

I wonder why the memory didn't error when I ran memtest86+ from a boot CD? Oh well... it seems to be good now.

w00t

stevemedes
11-27-07, 08:21 PM
Well, I lowered the timings a bit (from 2.5-3-3-8 to 3-4-4-8) and the blend test has been running for 19 minutes without a hiccup.

I've got my fingers crossed. I think that might have done it. The memory is rated at CAS 2.5, I don't know why it'd need to run at 3... but whatever. If its stable like this for a couple hours I'll be happy. Latency does squat for performance anyway.

I can't wait to just sell this machine. I'm waiting on another gig of memory then I can get my $475. Not a bad machine for the money really. :)

EDIT: 46 minutes... still error free.

I wonder why the memory didn't error when I ran memtest86+ from a boot CD? Oh well... it seems to be good now.

w00t

CAS latency makes a big difference in performance for DDR1 systems. More so if you go from 2 to 3, not 2.5 to 3.