mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Hardware (https://www.mersenneforum.org/forumdisplay.php?f=9)
-   -   Hardware Failures at different FFT lengths? (https://www.mersenneforum.org/showthread.php?t=5052)

CADavis 2005-11-27 08:10

Hardware Failures at different FFT lengths?
 
What do failures at various FFTs generally mean when overclocking?

I was overclocking with my CPU, RAM, and CPU HT way under stock speeds, trying to max out the HTT on my motherboard. I was running one instance of Prime95 on an small FFT test, and one instance of Prime95 on a large FFT test. The large FFT test failed as I got to a certain HTT, but the 8k test did not fail. Given the fact that both RAM and CPU were way below stock speeds, does this failure, and the fact that my system froze soon after, mean that I exceeded the maximum HTT that my board will run at?

I was running an AMD 64 3500+ at 1600Mhz, DDR400 at about DDR266, and the HT frequency was set at 600 Mhz as the base. The HTT was at about 245 when the failure happened and then the system freeze.

Any thoughts?
Thanks in advance.

Cruelty 2005-11-27 22:38

What Mobo do you have? What HTT multiplier did you use at 245 (should be max. 4)? Did you monitor your chipset temp. when this error occured?
Overall, if everything else was below specs I would check the chipset cooling + apply a little more voltage to LDT (1.3V) and if that doesn't help eventually chipset itself (1.6V) - you have to assure proper chipset cooling as the stock cooling usually means CRAP.
One more thing: use blend test to verify system stability with A64 (memory controller is embedded in the CPU).
DO NOT run two stress-tests at the same time unless you have a dual-core CPU :no:

Prime95 2005-11-28 03:23

The 8K FFT can run out of the L2 cache. The larger FFT pounds main memory. I suspect a memory subsystem problem.

CADavis 2005-11-28 04:00

I was using the eVGA SLI motherboard. The HTT multi was 3 (actually set to 600 MHz but same thing). I dont know what the exact system temp was, but I think it was under 50. The CPU temp was < 25 definitely.

Since I posted, I tried setting the chipset voltage to 1.6, and running the CPU at 1.55, with the same settings. The same thing happened, hang then crash.

I moved my CPU to my "old" motherboard, an MSI Neo2 Platinum, and tried similar things to test the HTT. It seems to max out at 230, which is even worse. Even with RAM set at DDR200.

Basically I am thinking that it is the memory controller inside the CPU that is giving me the troubles, and hoping that when I get my Opteron 148 this week it will work much better due to the better memory controller. Does that sound right?

Thanks for the feedback in the mean time!

Cruelty 2005-11-28 12:40

Do you have the most recent BIOS for your mainboard? What is the codename of your CPU (Clawhammer, Newcastle, Winchester, Venice)? Especially Venice requires the newest BIOS that actually supports it - without it, it is not prime stable even at stock speed not to mention OC'ing.

Prime95 2005-11-28 21:41

[QUOTE=CADavis]Basically I am thinking that it is the memory controller inside the CPU that is giving me the troubles. Does that sound right?[/QUOTE]

From what I've read, HTT is often the limiting factor in running prime95 on AMD boxes.

CADavis 2005-11-28 23:12

Yes I have the most recent BIOS, but for an eVGA motherboard, that's not saying much...

It is a Newcastle core.

But now I am not using it anymore. I'm beginning my venture into the wonderful world of the socket 939 Opteron. Thanks for the info you guys had!

PS: Haha Prime95 your post count is at 1234, that made me laugh.

Peter Nelson 2005-11-29 06:36

[QUOTE=CADavis]Yes I have the most recent BIOS, but for an eVGA motherboard, that's not saying much...

It is a Newcastle core.

But now I am not using it anymore. I'm beginning my venture into the wonderful world of the socket 939 Opteron. Thanks for the info you guys had!

PS: Haha Prime95 your post count is at 1234, that made me laugh.[/QUOTE]

Laugh? 1234 is not even PRIME!

:-)

Greenbank 2005-11-29 10:29

Only 3 to go then.


All times are UTC. The time now is 23:09.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.