Hello! In about 5 minuets, I might be a statistic, or not. Back-story:
I have had my MSI 890 FXA-GD70 for quite a while. Very early on I noticed the VRM/NB to run insanely hot. It has been actively cooled since about day 2. Just an 80MM Vantec thermoflow hot-glued to the Video card and blowing at it keeps things sane.
Great cooling important to me because, my year-round average ambient temperature is kept near 80. I built my rig to run stock without overheating at full load around 90.
The board has been rock-solid (with some bios-versions dramatically worse than others. The current one is great.) . It is usually run at stock speeds. Today, we shall see. The only fault I can find with it is that no one at MSI will give you a straight answer about the fan headers regarding max wattage. Consequently, I have blown all but one. Heat and noise is a big issue with this board. The new one should be better.
Here is the plan:
I have done this today: http://valid.canardpc.com/show_oc.php?id=1963200
and it has been stable in the desktop (near idle) for a few hours.
I have also the new MSI 990FXA-GD80 on order. It is the same board (MS 7640 I believe, but a newer revision number.)
I have also cooled the ambient temperature to 72. Chip is idling at 31C core 34C. Thermalright trueblack cooler with a moderate fan. Thermal paste is aged ceramique and not fresh.
Now, Prime95. All indications point to spectacular failure almost immediately.
If I am not back in a half hour or so, I'll be camping out in the driveway waiting on UPS.
However, I have my doubts about this, So I am willing to put a (nearly) perfectly good board where my mouth is. My belief is that with adequate cooling, it should be fine.
In fact, I have more concern about the processor at these voltages and speeds and a heavy load than the board. I presume I will hit temperatures approaching 60C on the core within 30 seconds. If I get there and nothing blows up, I will discontinue the test. This assumes it is stable enough for prime.
Why? I would like to know If my confidence in this board was misplaced.
Here we go...
Updates here so as not to make a mess:
Well - that was easy. Instant reboot. Nothing blew. Ram needed adequate voltage- Duh. Also ticked up other voltages. Now at 1.525 or 1.53 depending on software.
Take 2 at 1.525/1.533 Ditto. Reboot.
Now at 1.5375 and 70.7 ambient. Pushing pretty hard here and nothing burnt up yet.
Take 3 at 1.5375 And again, Reboot. However, multicolor static on the screen and a blast of static from the speakers. I recognize that as a sign of too much voltage (with no stability benefit.)
Early conclusion: Crazy voltages and silly overclocks alone do not seem enough to cause a failure.
It must be stable enough to actually bear a full load for a period of time. Throwing up some silly numbers for benching does not seem to be of any harm.
I know from previous experience that this rig will do 4.0GHZ prime stable for at least a half hour at hotter ambient temperatures. (75 or so) because I did it last night.The numbers I think to be the NB/MOS temperatures never went above 48C. It held stable through 3 hours of video games after that.
Seems if you cool the hot stuff and only have 4 cores, the failure is neither violent nor immediate. In fact, I have seen nothing beyond standard overclocking instability for silly numbers and voltages.
I am satisfied that with 4 cores and good VRM cooling at normal room temperature, perhaps only extended periods of burn-in testing, or long term overclocks even have the possibility to cause trouble. I have no reason to believe short-term VRM related damage is even possible with proper cooling and 4 cores.Edited by JohnBulb - 8/21/11 at 6:56pm