Re: freeBSD 5.3 compile problems

> Adam> btw a simple way to get multi-cpu support would be to write a
> Adam> simple shell script to spawn as many instances of hashcash as 
> your
> Adam> hyperthreads x cpus and wait for the first one to terminate then
> Adam> kill the rest.  As there is no need for communication between the
> Adam> instances this should basically be just as effective.  An 
> exercise
> Adam> for the reader...
>
> I was just thinking about this yesterday.  I think this is a good way 
> to
> go.  I don't know if a shell script would be able to cut it, but it
> should be simple to write a C program with some fork(2)s, wait(2)s and
> kill(2)s.  I should be able to whip up something soon (say, next week).
>
> BTW, this is also a good solution from the point of view of clusters
> (and distributed computing) because threads don't work too well under
> those environments.  Some (most?) clustering software can't even 
> migrate
> programs that use threads.

Eric mentioned the clustering question some time ago, and I thought up 
a basic protocol which could distribute the work among ordinary desktop 
PCs (without clustering software).  I think I even managed to make it 
DoS resistant, in that performance would never get worse than if the 
PCs remained standalone.

All you need to do is broadcast a request packet every minute when you 
have work to do, while working on it yourself.  Any PCs that hear it 
can choose to help.  Requests that are not refreshed for more than 5 
minutes are cancelled.  Each PC *always* works on it's own requests in 
preference to others', but may then work on other requests in random 
order.  Once a solution is found by any PC, it is also broadcast, being 
a signal for all PCs who hear it to cancel that request.

--------------------------------------------------------------
from:     Jonathan "Chromatix" Morton
mail:     [email protected]
website:  http://www.chromatix.uklinux.net/
tagline:  The key to knowledge is not to rely on people to teach you it.