[Maxima] ram problem

Vadim V. Zhytnikov vvzhy at mail.ru
Sat May 19 09:42:29 CDT 2007


Vadim writes:
>>> Dear all,
>>> I am using the latest windows build of Maxima and trying to solve 
>>> nonlinear equations. But as I understand during the build process of gcl 
>>> there is a ram bottleneck and the build I use can not continue analysis 
>>> and stops from the ram problem. I tried to build the windows version 
>> >from source but when trying to ./configure the gcl it stops as "cant 
>>> find c compiler" althoug I got the correct version for gcc and binutils 
>>> but dont (and cant find but the latest) for gcl.
>>>
> 
> I see that GCL's configure issues 'uname -s', what does your 'uname -s'
> reports to you?

MINGW32_NT-5.1

> 
> Can you please share knowledge about maximum memory for other LISP
> implementations?
> 

On 32-bit Linux I was able to get the following numbers:

clisp (no memory control parameters AFIK) - ~1Gb
cmucl with -dynamic-space-size - ~1.5Gb
sbcl with -dynamic-space-size - ~1.5Gb
gcl 2.6 with --enable-maxpages=1024*1024 - ~2Gb

Gcl 2.6 is less memory efficient since it has
3-word cons-cells while other lisps have 2-word
cons cells.  Gcl 2.7 which should be released later
this year will have 2-word cons as well.

I don't know exact limits on 64-bit.
I heard success stories about 8Gb heap
size with sbcl and very large heap
sizes (16Gb?) with gcl on IA64 (Itanium).
Here results of my personal tests:

clsip can allocate at least 4Gb (I don't have more).
sbcl is able to start with --dynamic-space-size 32Gb !
I also was able to build and start gcl 2.6.8 with
--enable-maxpage=8*1024*1024 and (room) reports
8Gb of available pages = 32Gb of RAM !




-- 
      Vadim V. Zhytnikov

       <vvzhy at mail.ru>
      <vvzhy at netorn.ru>


More information about the Maxima mailing list