[wellylug] diskless headless

[John Durham]

On 4/11/2010 2:44 p.m., E Chalaron wrote:
>
> Hrm... parallel programming does not scale linearly with the number of
> processes. There is some added complexity to split jobs up and
> reassemble them. It's hard to generalise in a single email. Wikipedia is
> a good resource[1] However, parallel processing can perform much better
> than a single CPU if three broad features exist with the data:
>>
>>   - massive amounts of data are required to be processed
>>   - the job can be split into many smaller jobs
>>   - shared state is not required between the smaller jobs
>>
> what do you call shared state ?? sorry  lack of knowledge here...
> cheers
> E
If this is what it looks like, I used to think about real parallel 
processing networks with shared bus in up to 6 directions (up, down, 
left, right, front, back). The hard part is the job sharing software, 
which has to cope with requests from surrounding processors while 
running its own jobs. They can be seriously cool if made right, but a 
pain in the neck if the users don't understand the system well.

Much depends on the way jobs are farmed out to other processors. 
software required would be unique to the computer structure, which could 
be unique in itself. Still, for me it was all theory.
-- 
Regards, John Durham
Fax/Phone 64 4 5286786
PC-HELPERS list subscribe/unsub at http://modecideas.com/discuss.htm?sig
Inventor of multi-spectral fibreoptics.
Good advice is like good paint- it only works if applied.
Quote: What I cannot create, I do not understand. (Richard Feynman)