- Hi all:

I am working on a number theory problem and am calculating billions and

billions of iterations related to the problem to get an understanding

of how the primes are behaving under various constraints. I can now

get sufficient results to establish certain trends related to my

problem, but I would like to be able to get into the trillions of

iterations as added confirmation of my findings. Each run now takes

one or two days running on my one year old HP laptop with a dual

processor speed of 1.9 Ghz. I have 2 Gb of ram, and that seems to be

sufficient so far. My software is old, a 1990's era copy of Borland C++

3.0 and it is running on DOS. My code I am sure is not very efficient

speed wise, but I doubt I can make it much faster (my limitations).

Can anyone recommend what I might do to get the speed up by something

close to an order of magnitude without too much cost ($500 or so)?

Thanks for any suggestions.

Regards, Bill - --- On Sat, 2/14/09, bill2math <bill2math@...> wrote:
> Hi all:

The first thing to do is to work on improving your algorithm. Try to ensure that you don't calculate anything expensive twice, and deduce things from results of previous calculations. There are some pretty hot algorithmists and mathematicians on the list, feel free to describe your program and algorithm and we can see if there are improvements that can be made.

> I am working on a number theory problem and am calculating

> billions and

> billions of iterations related to the problem to get an

> understanding

> of how the primes are behaving under various constraints.

> I can now

> get sufficient results to establish certain trends related

> to my

> problem, but I would like to be able to get into the

> trillions of

> iterations as added confirmation of my findings. Each run

> now takes

> one or two days running on my one year old HP laptop with a

> dual

> processor speed of 1.9 Ghz. I have 2 Gb of ram, and that

> seems to be

> sufficient so far. My software is old, a 1990's era

> copy of Borland C++

> 3.0 and it is running on DOS. My code I am sure is not very

> efficient

> speed wise, but I doubt I can make it much faster (my

> limitations).

> Can anyone recommend what I might do to get the speed up by

> something

> close to an order of magnitude without too much cost ($500

> or so)?

> Thanks for any suggestions.

Phil