Well, so far a couple of the systems in there have been running for several years almost non stop. The only time they are not running is during a power outage or once a year when I open them and blow out the dust. The longest running one has been going since May 2012, with an OS upgrade in 2013 and a chip upgrade a couple of years ago. It has a little more than 33 years of cpu time. So far so good.
Cheers

Arrrgh! My best cruncher went down sometime in the middle of the night. It won't even power on now. It is a homebuilt machine with an I7. When I get some time I will need to pull it apart. I suspect the power supply, but it was a brand new one, not something I got from another machine. Temperatures had been running in the low 40's(C) so I know it was not overheating. Hmmph!
Edit: Good news. I have it back up and running. It turns out there are two little plastic clips that hold the on/off switch in place. One of the clips gave out and then the switch moved out of position so the front button would no longer engage it. Once I zip tied the switch back into place, it started right away. So I guess when I went to restart it the other night I did not even pay attention to whether or not it started, I just presumed it did without checking. So all is once again well, at least for the moment.
Cheers

Supposedly the new AMD chips are pretty close to on par with the upper end Intel chips. We won't know for sure until some get out in the public and we can see how they benchmark in a production mode. The big question mark is how much power do they use and how much will they cost. If they can come in at 30 to 40% of the comparable Intel offerings on price that will put pressure on Intel to kick their prices down. We will have to see if they live up to their hype. It would be nice if they did.
Cheeers

This would be nice in a dual processor setup....Depending on price though.
http://overclock3d.net/news/cpu_mainboard/amd..._naples_server_platform/1

Boinc/WCG may have to rethink some of their limits. As the thread/core counts increase, the WCG 35 wu per thread limit becomes irrelevant. With that many threads one would hit the BOINC limit of 1000 work units in progress per host maximum long before the 35 per thread. On a 128 thread machine, that would be about 8 work units in progress per thread, counting the one executing, which would be OK for longer running tasks like FAH2 or HSTB but wouldn't be much of a buffer for shorter units like FAHV. Any hiccup in the feeder and your machine would run dry in less than a day.