LHC@home

maybe i'm wrong, but i think that the program code can be optimized.
If my client runs 4 lhc tasks, my i5 M480 cpu is always at 2,93ghz, the 'estimated app speed' in the task properties shows 2,6 GFLOPs/sec.
If only i.e. einstein tasks run, my cpu is never higher then 2,67ghz, and the properties shows 4,16 GFLOPs/sec
afaik the turbo-boost is only aktive when the cpu's cores have not so much to do

Thanks for that; we have just switched to the Intel
ifort compiler and are more worried about numerical
compatibility. If the current results are OK then
we can relax the constraints and really use SSE2 etc.
Also under the new regime we should be able to have
multiple executables targeted at the hardware available.

Eric.

---

thx for the quick information
my target is to find the things that can be optimized, all the time and everywhere.
The world need good solutions.
throughput and quality rules

Michael