Re: using scheme for large time-series datasets?
- From: pjb@xxxxxxxxxxxxxxxxx (Pascal J. Bourguignon)
- Date: Thu, 02 Apr 2009 19:44:53 +0200
Jason <jason.lillywhite@xxxxxxxxx> writes:
Have you tried running your stuff and seeing if things are too slow?
I have not. But you talked me into it! My colleagues tell me FORTRAN
might be the best when getting into the numerical analysis and number
crunching on large datasets. Looks like I need to do some more
investigation into Lisp because it is becoming very enjoyable (I
didn't think it was possible coming from Ruby). Thanks for the
pointers. I'm going to study the tail recursion!
What was a large data set a few years ago is not a large data set anymore.
It was large when you couldn't keep it in RAM, you had to use swap or
I/O, and when processing it meant the computer was busy for at least a
whole night, if not weeks.
But thanks to Moore law, suddenly computers are 10,000 times faster
with 10,000 times greater RAM sizes, so what was a large data set is
not anymore, and holds only a small fraction of the RAM, and can be
processed interactively even by the worse algorithms.
And we've reached the point where it's even hard to come with "large
datasets", unless you have video data, and even that, soon won't be
large anymore.
Even the human genome is less than 4GB...
--
__Pascal Bourguignon__
.
- Follow-Ups:
- Re: using scheme for large time-series datasets?
- From: Jason
- Re: using scheme for large time-series datasets?
- References:
- using scheme for large time-series datasets?
- From: Jason
- Re: using scheme for large time-series datasets?
- From: Aaron W . Hsu
- Re: using scheme for large time-series datasets?
- From: Jason
- using scheme for large time-series datasets?
- Prev by Date: Re: using scheme for large time-series datasets?
- Next by Date: Re: using scheme for large time-series datasets?
- Previous by thread: Re: using scheme for large time-series datasets?
- Next by thread: Re: using scheme for large time-series datasets?
- Index(es):
Relevant Pages
|
