Page 1 of 1

Long term stability - lesson learned again

Posted: Wed Sep 19, 2012 12:57 am
by Napoleon
Looks like I'm relearning basic stuff the hard way. Quoting myself:
UPDATE (2012-02-05):
Everything but the CPU has been at stock for the year 2011. Since I've had two suspicious uniprocessor failures on December 2011 - call them PEBKAC(?) and FlakyPower(?) if you like - I've decided to reduce the CPU OC from 2100MHz to 2055MHz (stock clock is 1600MHz) and bump the CPU, memory and chipset voltages a bit. That should fix an implied uniprocessor folding instability, just in case the past 24/7 folding year has caused some wear & tear on the CPU already. Improved case airflow should also allow me to OC both GPUs a little bit.

UPDATE (2012-02-16):
Minor bump to the GT430 OC.

UPDATE (2012-09-06):
Cutting back on the CPU OC again...

UPDATE (2012-09-19):
Cutting back on the GT430 OC and improving case airflow as well. Apparently some 1½ years of 24/7 folding at relatively high temps did strain the card. It was still FAH-stable, but when I decided to see how the card is doing and heated it up beyond FAH temps with Furmark...
Well, it was no longer Furmark-stable, the driver kept shutting down Furmark and reseting the card. :|
Well, it's been something like 1.5+ years of overclocked 24/7 folding at relatively high temperatures with this particular setup, so some wear and tear shouldn't be a surprise. But to me it was, once again. Oh well, past stability isn't a guarantee of future stability and all that. Doesn't hurt to double-check every now and then. :twisted:

Re: Long term stability - lesson learned again

Posted: Wed Sep 19, 2012 1:00 am
by Jesse_V
Thanks for the reminder. We'll be sure to keep it in mind. Sorry to hear about your card.

Re: Long term stability - lesson learned again

Posted: Wed Sep 19, 2012 1:16 am
by Napoleon
No worries about the card. :ewink:
It's Furmark stable once again when I cut back on the OC a bit and reduced temps by improving airflow (read: higher fan speed), so it still has some folding left in it. I've gotten 2.2+ million points out of it thus far, not too bad for a low end part I suppose. If it goes boom in the not so distant future... it's Kepler time. I haven't seen beta stuff for my setup for quite some time, so one might say a minor upgrade is long overdue. :lol:

Re: Long term stability - lesson learned again

Posted: Wed Sep 26, 2012 6:01 am
by user123
The failures could have been caused by overheating.
Dust can build up on the heatsinks and fans and hinder the cooling process.

Re: Long term stability - lesson learned again

Posted: Wed Sep 26, 2012 8:43 pm
by Napoleon
The "problem" certainly is heat related... like I said, I noticed the problem when I recently heated the card well beyond folding temps using Furmark. Thing is, when the card was fairly new, it could stand higher Furmark temps than now, and without problems too. After folding practically nonstop with an OC and 80+ degrees Celsius temperature (spec max is 98C) for at least a year, it seems a fair assumption that the card no longer simply is "as good as new". I'm pretty regular when it comes to checking for any dust buildup and increased temperatures. The card has never actually overheated, although in the past I've intentionally pushed it pretty close to spec maximum during burn-in tests.

Re: Long term stability - lesson learned again

Posted: Fri Sep 28, 2012 8:56 am
by rpmouton
You OC'd guys are usually pretty good about heatsink compounds but if you haven't removed the heatsink and replaced the compound with a fresh, viscous, preferably silver filled layer you might try that out. 6 to 10 K hours at 85C shouldn't be cooking the semiconductors noticeably..

Re: Long term stability - lesson learned again

Posted: Sat Sep 29, 2012 9:24 pm
by Napoleon
Sorry folks, I can resist just about everything except temptation. Here's a hilarious comparison. http://www.dansdata.com/goop.htm

Re: Long term stability - lesson learned again

Posted: Sat Sep 29, 2012 11:28 pm
by k1wi
Napoleon - that post does note that better thermal grease may lead to longer durability - which may very well be contributing to your problem? If you're still using the standard grease perhaps it has dried out as a result of the higher than normal average temperature.

I liked the accuracy goal, but found that article a little lacking across a number of values - for example, how many watts were being driven through the heater block, what temperatures were being measured? I suspect a lot of inaccuracy was not accounted for.

A lot has happened since 2002, although I do believe that Vegemite is still made to the traditional recipe.

A good, more current image is here: http://www.hardwaresecrets.com/imagevie ... mage=43088, from http://www.hardwaresecrets.com/article/ ... 012/1490/5.

Yes thermal grease is only a matter of degrees, but that is often measured when very recently applied and there are numerous instances where marginal improvements are very important...

Re: Long term stability - lesson learned again

Posted: Sun Sep 30, 2012 1:48 am
by Napoleon
Thanks for the links. I didn't even realize there are *that* many choices. I use Arctic Silver 5 myself, I've had good results with it whenever I've been asked to "fine-tune" computers. I must admit I've been too lazy to do that for my own rig. No doubt it'd do good for the card. But temperature is much better now, after reducing OC and improving airflow, so I don't see any urgent need to tamper with the stock cooling.

As for the link I posted, I liked the tongue-in-cheek style of writing. And the main point was that some goop - almost any goop - is way better than no goop at all. Very evident in the tests you referred to, as well. The toothpaste comparison has some fun value to me, and it makes sense that evaporating goop does a decent job - as long there's something to vaporize. :lol:

I recently installed a new mobo, graphics card etc to a friend's computer, and I was a bit hamfisted when removing the old graphics card. I grabbed it by the heatsink, and there was an audible crack as the heatsink came loose. Well, the springs keep it in place and the card still works, but I think it's pretty obvious that whatever goop there is has gone dry a long time ago and should be replaced.

I don't think any decent thermal compound needs to be reapplied frequently, but I suppose it might do some good to reapply after a couple of years or so.

Re: Long term stability - lesson learned again

Posted: Sun Sep 30, 2012 9:49 am
by rpmouton
Napoleon wrote: I don't think any decent thermal compound needs to be reapplied frequently, but I suppose it might do some good to reapply after a couple of years or so.
I totally agree with this. It is certainly my experience with laptops that after about two years the factory grease is solid and the movement of/in the laptop has separated the heat sink from the chip.

EDIT: Almost forgot to mention that one such netbook with thermal problems from its delivery had a sticker on the CPU underneath the heatsink. Definately got improved performance after redoing that one!

Certainly appreciated the link you provided for what it was, I guess my equivalent to Vegamite would be creamy peanut butter..

Re: Long term stability - lesson learned again

Posted: Sun Jun 29, 2014 2:06 pm
by user123
user123 wrote:The failures could have been caused by overheating.
Dust can build up on the heatsinks and fans and hinder the cooling process.
Yesterday, I cleaned my PC after running FAH (SMP and GPU clients) on it for more than a year.
Removed dust from the casing vents, fliter sponge at the front of casing, CPU and GPU heatsinks etc.

Also took the opportunity to upgrade the thermal paste (from the thermal paste that came with the card to Arctic Cooling MX-4) between the GPU and its heatsink.

Tedious but worth the effort. :D

At folding load,
CPU (SMP client) temperature decreased by about 3* Celsius.
GPU (FAH6) temperatures decreased by about 9* Celsius (checked GPU temp via MSI Afterburner). GPU fan speed is also down so it's less noisy :)

Re: Long term stability - lesson learned again

Posted: Sun Jun 29, 2014 2:12 pm
by 7im
Nice Necro.