Originally Posted By: Volvohead
Ahhh . . . my favorite computer HW topic, PSUs. The one thing that causes all the most mysterious problems. The most important component in the box, IMO.
First off, what make is yours? The power ratings on the cheaper ones can be overstated, depending on the test parameters used. Temperature affects PSU performance. As they age, output capacity drops. This accelerates with higher temperatures. In a box running higher ambients, the capacity and stability can be compromised in the lesser quality PSUs. A hot GPU can add to the heat load. A higher efficiency model will run cooler and use less power. If you use the PSU as the primary exhaust fan you're asking for trouble. Good PSUs test at rating at 50C. A cheapie may derate by 25-40% at that temperature.
Not knowing the full system build, I can only guess at your load. That CPU is not all that power hungry. Normally, a 350 watter is plenty for that older MB/CPU combo. But multiple HDDs and a full ram fill may be pushing the limits of the PSU from a long-term perspective. Just because it's rated at 350 watts doesn't mean you want to draw that, or a majority of that, on an ongoing basis. Running at 50% of rating will provide maximum stability and PSU life.
There are some decent PSU calculators on the web that can give some general guidance as to your loads.
The other issue is not just the wattage, but the rail loads. While all PSUs draw from a common rail, those with multiple "virtual" rails spread the OCP across several branches. This can become important if you have loaded down one branch with too many HDDs, fans, GPUs and other equipment. If you overload a rail, the PSU will cause freeze ups or reboots. Cold boot up is when the greatest current draw occurs, and that is when the marginal PSUs poop out or show problems.
The best way to check out the video is to place it in a known good working system. If the issues repeat, then you've found the problem.
Otherwise, I have seen older MBs with tiring caps that work fine until you start loading up the slots, and then they start collapsing. And tired MBs will stress PSUs and vice-versa. The other thing to make sure is that you are using all your auxiliary ATX supply sockets. GPUs without discrete supply taps can draw heavily off the MBs supply bus.
Believe me, PSUs and good power distribution can be very vexing. I just had to ditch an otherwise very good 650 watter on a new Xeon server build. The 3.06 ghz Nehalem, 12 gb DDR3, and a full HW RAID array (with a flock of other PCI/X cards) was more than it could handle on the involved rails (even with staged drive spool ups). Plunking down $200 for a new one that could handle it was not fun.
Probably more than you wanted to know, but it might help.
Similar scenario here, q-core Xeon, 8GB of RAM, 6x 500GB HDD's in mirrors and fitted with a 1200W PSU because the board supports dual XEON's and we've only got one in there right now..... But plan on upgrading it.
I don't think it's possible to get "too good" a PSU.
And I agree completely with your sentiments on those that comes in cases (except AOpen and Enlight cases, which come with PSU's made by SPI) belong in one spot: the trash.