[Chimera-users] Memory allocation
pett at cgl.ucsf.edu
Wed Sep 13 10:49:55 PDT 2006
I guess when I put out the announcement for the 1.2255 snapshot
release I forgot to mention that the session code got rewritten, so
you should try that release. Using 1.2255, sessions are about 40%
smaller, use 15% less memory and take half the time to start up.
Unfortunately, Brittany's problem is that Chimera isn't very
efficient when handling many thousands of small molecules (compared
to a large molecule with about the same number of atoms as the small
molecules combined). We're looking into that.
On Sep 12, 2006, at 4:08 PM, Jonathan LS Esguerra wrote:
> I also encounter memory problems (quite often) because I'm working
> large structures such as the ribosome.
> When I try to restore .py sessions as large as 40 Mb, Chimera
> after around 5-10 minutes and throws a memory error.
> It doesn't help that my machine runs on dual pentium 4 processors
> (2.8 GHz
> each) and 2.0 Gb RAM.
> I just gave up saving too large sessions in my PC when working with
> ribosome structures.
> I have had some correspondence a long time ago with Eric Pettersen
> He suggested me the following:
> "....if you have access to another Windows machine with more that 1
> GB of
> memory, you might try restoring the .py session file there and see
> if it
> works and produces a .pyc file. If it does, you could copy
> the .pyc file
> back to your machine and be able to restore the session while
> bypassing the
> compilation step.
> Another alternative if you are a little Python-savvy is to open
> Python shell (Tools->Programming->IDLE) and compile the .pyc file
> import py_compile
> py_compile.compile('full path to .py file', 'same path except with
> [remembering that backslashes in quoted strings need to be
> then stop Chimera and start a new one in which you try to restore the
> The important difference between .py and .pyc files is that .py
> files are
> portable between different computer architectures and therefore can
> be distributed to other people and/or used on other machine types
> .pyc files are not guaranteed to be portable."
> I have not really tried the trick above since at the moment, I'm
> the only
> one here at the lab with a PC with 2 GB RAM...and still I couldn't
> my large sessions. I have been trying to install Chimera in our linux
> cluster here but have had some setbacks with missing
> libraries,,,and some
> other things to do... :(
> Jonathan LS Esguerra, MSc Bioinformatics
> PhD student, Microbiology/Functional Genomics
> Cell and Molecular Biology-Microbiology
> Göteborgs University
> Box 462, Medicinaregatan 9C
> 40530 Gothenburg City
> Tel:+46 (0)31-7733738
> Fax:+46 (0)31-7732599
> ----- Original Message -----
> From: "Thomas Goddard" <goddard at cgl.ucsf.edu>
> To: <brmorgan at clarku.edu>
> Cc: <chimera-users at cgl.ucsf.edu>
> Sent: Tuesday, September 12, 2006 10:39 PM
> Subject: Re: [Chimera-users] Memory allocation
>> Chimera does not have any fixed amount of memory allocated. It will
>> use whatever is available. If you load large molecular systems
>> (100,000 atoms) or large volume data sets (> 100 Mbytes) your
>> may start swapping (transfering data back and forth to disk
>> because it
>> does not fit in memory). That can dramatically slow down any
>> computation by factors of 10 or 100. The only real solution is to
>> a machine with more memory, or load smaller systems in Chimera.
>> Are you working with molecules or volume data? There are tricks to
>> reducing the data size depending on the type of data.
> Chimera-users mailing list
> Chimera-users at cgl.ucsf.edu
More information about the Chimera-users