[OEIS] Re: Core 2 Duo and speed of computations. Quick machine needed.

Antti Karttunen antti.karttunen at gmail.com
Thu Oct 12 15:17:59 CEST 2006


Lßbos ElemÚr wrote:

>Dear Seqfans,
>Did somebody [among you] worked
>with PC including new Core 2 Duo CPU
>claimed to be quick.
>Did somebody compared spewcifically
>for mathenatical computations
>PC with P4-3GHz
>and
>PC with Core2Duo 2.1GHz processzor..
>
>Is this new CPU quicker despite of lower frequency
>at mathematical purposes or it is quicker
>only in games or media performances...?
>
>  
>
I think it depends largely on, whether the programs you are using
(Mathematica, PARI?) can effectively partition their computations
to multiple threads (Which I doubt in general...)
If you are using the same computer heavily for same daily
purpose, then the duo core is somewhat useful,
as the other core can take care of the mathematical
calculations and the other core of your more mundane
tasks (or that how it should in principle be allotted...)

But if you intend to dedicate that new machine exclusively
for long mathematical computations, and the ability of
Mathematica and PARI to utilize two cores is limited,
then I would suggest the single core CPU with higher frequency.
(Don't buy laptop, they get hot when running heavy!)
On the other hand, if you are writing some C-code by yourself,
then the multicore thing might be better for some applications.

Caveat emptor: These are just my theoretical speculations, having no 
experience
of actually using any multicore PC!

Actually I would need myself a semi-dedicated machine
with 1 - 4 gigabytes of RAM, moderately fast CPU,
and some sort of Unix (e.g. Linux or NetBSD) to which
I could connect with ssh, and compile C-programs with gcc,
and download and install there some needed libraries (for the
other users' joy as well!), compile my programs, and then
let them run for weeks.

It would be nice to find a mesenate that could offer
such a machine on the net for us the fanaticians' sequence work!
("SeqFan's creative computer lab". The users are required
to publish their results on OEIS.)


Yours,

Antti







More information about the SeqFan mailing list