As to me being "totally mistaken".....these are the data from the foreign language website of people actually doing it with F@H beta, bringing all their drivers up to date and having done 40/50 WU's in a go (pieces of 6 minutes)! Even designed to run 125 Celcius (257 Fahrenheit) constantly? That's a very serious temperature issue building inside any PC.

If......The watt v thru-put is not at discussion....its the very fact that it does eat a lot of juice which is contrary to what the average computer user wants to see from his home PC / entertainment-centre......noise, heat, running cost.

ciao

Errata. Article claims ATI GPU fan not starting until 100 degrees C, but before that the system fans are already majorly abuzz.

It's never going to be too late. Being too early might be fatal, though. Expensive, anyway.

Believe to have read that Dr.Baker/Rosetta was talking to MS and the possible use of the 4 million X360 3 Cell processor 3.2ghz machines already out there. That was posted somewhere a few weeks ago too.

Edit: Here's the post. Oct.11 that Dr.Baker commented: http://boinc.bakerlab.org/rosetta/forum_thread.php?id=1262

Hi ozlynx,
Ideally, the application programs are written by the project scientists, and we just board them for use with distributed computing clients. Unless you program an application right from the start for use with streaming instructions, turning on the SSE2 flag or the SSE3 flag on the compiler will make little difference in speed. Using streaming instructions needs to be part of the program design. It is not an afterthought that can be added on to a program. The easy way is to write your program using the BLAS (Basic Linear Algebra Subroutine) library or a similar library. Then it is easy to produce optimized applications.

We can always hope that we will acquire a project that has been programmed this way. I expect that such things are mentioned when scientists approach us for advice on a potential project.

Lawrence

70-75c on my 1900xt.