Re: Large Matrix Handling Issue 64 bit
- From: "Joaquim Luis" <jluis@xxxxxxxxxxx>
- Date: Sun, 4 May 2008 16:11:03 +0000 (UTC)
"Andy Robb" <ajrobb@xxxxxxxxxxx> wrote in message
<fvjqng$8lg$1@xxxxxxxxxxxxxxxxxx>...
I think he has already understood that thrashing swap isa
bad idea. He just did not understand why allocating a16GB
array took so long on a system with 16GB RAM. I explained
why 8GB is the practical limit to array size.
Andy,
You got me scared with that one but as far as I can tell
that copy mechanism is not used. For testing it I created
with no problems a zeros() array only slightly smaller than
the available largest memory chunk. If that copy thing had
been there I would have get an Out of Memory ... bla, bla,
bla error.
J. Luis
.
- References:
- Re: Large Matrix Handling Issue 64 bit
- From: Andy Robb
- Re: Large Matrix Handling Issue 64 bit
- From: Jeremy Greenwald
- Re: Large Matrix Handling Issue 64 bit
- From: Andy Robb
- Re: Large Matrix Handling Issue 64 bit
- Prev by Date: Re: maximum number of variables to be used with fminunc,lsqnonlin
- Next by Date: Re: number WITH insignificant zeros, or matrice with string
- Previous by thread: Re: Large Matrix Handling Issue 64 bit
- Next by thread: Re: Large Matrix Handling Issue 64 bit
- Index(es):
Relevant Pages
|