f@h causing shutdown

[Jesse_V]

Have you run any big games on this machine? I wonder if they would cause issues, although F@h often stresses your hardware more thoroughly than games.

[BrandonVeiga]

I'm not sure why the PSU would be bad.

I've run Crysis, Starcraft 2, Borderlands 2, and Diablo 3 at highest settings (except crysis) without any problems for hours and hours on end. I can't imagine f@h causing more of a load especially since I have it down to 50% on SMP and it still causes issues

[Joe_H]

If by 50% on SMP you mean that the slider has been set at 50%, that has little effect on SMP folding. That slider really only works on uniprocessor cores. If you want to cut down CPU utilization, reduce the number of cores set to be used. Changing the setting from the default of -1 which uses all available threads, 8 in your case, to 4 will cut down the heat generated and use half the cores and CPU time.

As for those games, none of them will use all cores at full blast like folding will.

[Napoleon]

Also keep in mind that the CPU utilization readouts in Task Manager can be very misleading. FAH SMP is all about floating point / SSE, and your CPU has only four such units anyway. So simply dropping to SMP:4 may still stress your CPU almost as hard as SMP:8. Seeing an actual drop in power draw and temperatures is much more reliable metric.

As for the PSU, I wasn't able to find a review at a quick glance, but I assume you're talking about this one? If so, the PSU is something of a concern, IMHO. For example, it has two 12V rails, 12V x 22A == 264W maximum each. That's not much, and if the PSU manufacturer happened to be foolish enough to put external PCIE power and CPU power on the same 12V rail, you're definitely in trouble. At stock clocks, maximum rated wattage for your GTX460 + FX8120 == 160W + 125W == 285W. Ooops.. "as if someone pulled the plug" kind of sounds like some protective circuit is getting tripped in the PSU.

OK, let's assume that the PSU manufacturer hasn't been quite that reckless, but then again, the description on the manufacturer's website seems pretty vague. I for one would like to have some hard facts about the PSU, say jonnyguru.com style, before ruling it out as the culprit... especially when I come across forum posts like this: http://www.jonnyguru.com/forums/showthr ... #post83759. If I understood it right, the modular plugs aren't properly keyed and it's actually possible to connect them incorrectly. Yes, external PCIE power is yellow-and-black only (+12V and GND wires).

Sorry to say, but this doesn't sound like the kind of PSU I'd like to have anywhere near my setups, let alone inside them...

[Nathan_P]

I'm not sure why the PSU would be bad.

I've run Crysis, Starcraft 2, Borderlands 2, and Diablo 3 at highest settings (except crysis) without any problems for hours and hours on end. I can't imagine f@h causing more of a load especially since I have it down to 50% on SMP and it still causes issues

F@H pushes hardware harder and faster than just about anything out there -it has been stated on these very forums that nothing will stress an overclock or break poor hardware more effectively than f@h

The PSU cannot produce its rated capacity. According to the sticker on the side it can supply 22a on each 12v rail - I'm guessing that due to the design of the PSU, all of the cpu power and at least half for the GPU is coming off the same rail and is thus overloading it and causing the psu to shut down.

To expand a bit further quite cheap lower quality PSU's have a label on them that states that they can supply far more power than they actually can - they know that most people think that a bigger number is automatically better and won't push a PSU to anything like its real capacity, let alone the number printed on the shiny sticker, and we are assuming that what the sticker says is accurate. If it isn't then anything is possible.

There are lots of horror stories on the web about crap PSU's, a good site to start with would be jonnyguru.com, he reviews the good, the bad and the ugly in PSU's

However you are loading a very power hungry processor, a gpu, motherboard and 8Gb of Ram at the same time, if the PSU is not upto it it will shut down end - People have been successfully folding on a Fx8120 and Gpu's since they came out so there are no setup problems. I know its not cheap but it will rule out the PSU as a possible problem and may save you the cost of more components going when the PSU does go pop

[art_l_j_PlanetAMD64]

I am running a Gigabyte GA-990FXA-UD3 motherboard, AMD FX-8120 Processor
I think that's the best i'm going to get with my older case which doesn't circulate air too well, and stock coolers on my CPU and GPU.

I think your stock CPU cooler is a problem. I just bought the parts to build a 6th FAH computer, with an AMD FX-8150 Processor, and when I was reading the Customer Reviews here, I found comments like this:

Awesome benchmarks, easy overclock, ultra fast processor. If you have the budget to buy this, then you will not regret it, but make sure you buy a heatsink because the stock heatsink is crap.

I thought I would try the stock CPU cooler just to see how well it worked, but even during the BIOS setup, the CPU temp was 59C. That's when it is doing almost nothing. The maximum CPU temperature is only 61C, so I had to install a much bigger aftermarket cooler to get the temperature down.

The FX-8120 and the FX-8150 both have the same 125W power rating, and the same 61C maximum temperature, so it is very likely that your CPU is overheating too. You should check the CPU temperature with a program like Speccy:
Speccy - Download

[Napoleon]

FX-8120 with boxed cooling, folding? Built a Bulldozer for a friend, did a quick burn in, then turned it to FAH for 24/7 operation for a couple of days. The WWWHHHHHRRRRR.... the WWWHHHIIIRR!

Stock cooling for BD just barely works. If you appreciate quiet and peace (of mind) at all, look for aftermarket cooling options. FX-8120 with stock cooling is so very... budget. OK CPU in my book, but the boxed cooling... far from OK.

[art_l_j_PlanetAMD64]

Speccy shows that my FX-8150 runs at 46C on a P8028 WU, with this cooling setup:
- Nexus AXM-8200 CPU Cooler
- Arctic Cooling MX-4 Thermal Compound

[art_l_j_PlanetAMD64]

There have been a number of discussions in several topics here, about the effect of higher temperatures on the lifetime of components/cards. For example, see here and here and here.

There is a short article about that here, for solid capacitors. There's even a chart for the lifetime at different temperatures:


Lifetime @ 65C = 57 years
Lifetime @ 105C = 0.57 years

And most semiconductor products will have a similar drop in lifetime at higher temperatures (ie ratio of ( lifetime @ 65C / lifetime @ 105C ) ). So it's a good idea to keep all electronic components running at the lowest temperature possible, that is compatible with reasonable cooling methods. If your cooling method is not able to keep the temperature down to a reasonable value, either get a better cooling method, or reduce the cause of the problem (eg lower clock rates on the CPU/GPU, etc.).