[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ScanTesla7 (was: ScanTesla7 for OSX)



Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

At 04:12 AM 6/22/2005, you wrote:
At 19:18 21/06/05 -0600, you wrote:
>I am sure you could run two models at once...
I have no idea how to have "one" program use the two CPUs....

Why don't you just split the search space into two and assign one half to each CPU.

That works pretty good! There is not a lot of extra power left over for web cruising but it does ok.


One thing to really watch is to use "real" values. There is no reason to scan Cp from 0 to 100nF in 0.1 steps. Since MMCs come in step sizes, just scan for those. In my case, I am just going from 75nF to 525 nF in 75nF steps since that is the smallest step size I have for the real MMCs. Same for Lp. You can only tune to say 1/10th of a turn, so no reason to scan every /10 inch ;-) So if you keep the increments fairly big and practical, you can speed things up about 100X!

Cheers,

        Terry






Say you wanted to search Lp from 10 to 20uH and k from 0.1 to 0.4.

Have one copy of ScanTesla search Lp from 10-20uH, k from 0.1 to 0.2..
Have the other search Lp from 10-20uH, k from 0.20001 to 0.4

and run one on each CPU. I'm sure it would be easy to write a script to do this and combine the output, for two CPUs or indeed for any number of CPUs that need not even be in the same machine. "Tesla@Home" here we come ;))))))

Steve Conner