pymol-users

[PyMOL] pymol crashes with large pdb files

[<an...@bi...>]

Hi all,

I'm running pymol 0.89 on quite capable machines. When I ray-trace large
pdb files (proteasome; 1pma, 1ryp, 1fnt straight off the pdb), pymol
crashes. Doesn't matter if surface or cartoon representation. When I
decrease the number of subunits from 28 (or 42 in case of 1fnt) to less
than 20 (cut-open or end-on view), ray-tracing works fine.

Is there a limitation to the number of atoms or subunits pymol can handle?
I reset max_triangles from 10^6 to 10^7. Didn't change a thing.

Any suggestions? Just work on Ubiquitin instead??


Thanks

Andreas

Re: [PyMOL] pymol crashes with large pdb files

[Nat E. <na...@bi...>]

> Is there a limitation to the number of atoms or subunits pymol can handle?
> I reset max_triangles from 10^6 to 10^7. Didn't change a thing.

Exactly how much memory does your computer have?  And what OS are you
running?  I've definitely crashed PyMOL before on a Linux system when I
tried to do things with huge structures like bacterial chaperonin.  It's
not that hard to run out of physical memory, or even to hit the
addressible memory limit.

----------------------------------------------------------------------------
Nathaniel Echols                                                  Programmer
na...@bi...                                        Gerstein Lab
203-589-6765                                                 Yale University
----------------------------------------------------------------------------

Re: [PyMOL] pymol crashes with large pdb files

[Andreas <an...@bi...>]

Sorry I didn't mention the operating system. It's WinXPpro. The
computers are a Xeon 2.8GHz (1GB RAM, Nvidia Quadro4 900XGL) and a P4
1.4GHz (768MB RAM, Nvidia GeForce2 GTS). Both run pymol 0.89.

Now there is a twist to the story. We also run linux (redhat with 2.4.18
kernel) on a 2 CPU 930MHz (1GB RAM, Nvidia GeForce2 GTS) PIII. Working
speed is much slower but no crashes when ray-tracing from the exact same
pml script.

The linux computer runs pymol 0.88. When I was using pymol 0.88 on
either of the WinBoxes they were crashing just like now with pymol 0.89.

Can it be that linux makes better use of its physical memory or that the
2GB of swap make the difference?

Andreas


On Wed, 2003-06-25 at 09:40, Nat Echols wrote:
> > Is there a limitation to the number of atoms or subunits pymol can handle?
> > I reset max_triangles from 10^6 to 10^7. Didn't change a thing.
> 
> Exactly how much memory does your computer have?  And what OS are you
> running?  I've definitely crashed PyMOL before on a Linux system when I
> tried to do things with huge structures like bacterial chaperonin.  It's
> not that hard to run out of physical memory, or even to hit the
> addressible memory limit.
> 
> ----------------------------------------------------------------------------
> Nathaniel Echols                                                  Programmer
> na...@bi...                                        Gerstein Lab
> 203-589-6765                                                 Yale University
> ----------------------------------------------------------------------------

-- 
                    Andreas Förster
Dept of Biochem, Univ of Utah, 20N 1900E, #2460 Eccles Bldg.
Salt Lake City, UT 84132, phone: 001.801.585.3919
home: 1736 Kensington, SLC, UT 84108, 001.801.463.3607
        http://www.biochem.utah.edu/~andreas

Re: [PyMOL] pymol crashes with large pdb files

[Nat E. <na...@bi...>]

> Now there is a twist to the story. We also run linux (redhat with 2.4.18
> kernel) on a 2 CPU 930MHz (1GB RAM, Nvidia GeForce2 GTS) PIII. Working
> speed is much slower but no crashes when ray-tracing from the exact same
> pml script.
> Can it be that linux makes better use of its physical memory or that the
> 2GB of swap make the difference?

When you ray-trace an image, PyMOL will tell you exactly how much memory
it's using.  My suspicion is that this is painfully close to the physical
memory limit.  From what I understand, swap space does not give you more
memory but Linux will transfer idle programs to swap, freeing up more
RAM for PyMOL.  I have no idea how Windows works (other than "badly").

'top' in Linux or the Task Manager in WinXP should tell you how much
physical memory is being used at any time.

----------------------------------------------------------------------------
Nathaniel Echols                                                  Programmer
na...@bi...                                        Gerstein Lab
203-589-6765                                                 Yale University
----------------------------------------------------------------------------

Re: [PyMOL] pymol crashes with large pdb files

[<an...@bi...>]

Limited physical memory might indeed be the problem.
When I ray-trace under WinXP, memory usage increases sharply (to about
500MB) and after a second or so, pymol crashes.
When I ray-trace under linux, almost all physical memory (1GB) is being
used. Barely 40MB available.

We should be getting new computers next week. We'll see if 2GB will do the
trick for Windows.


Andreas


>> Now there is a twist to the story. We also run linux (redhat with 2.4.18
>> kernel) on a 2 CPU 930MHz (1GB RAM, Nvidia GeForce2 GTS) PIII. Working
>> speed is much slower but no crashes when ray-tracing from the exact same
>> pml script.
>> Can it be that linux makes better use of its physical memory or that the
>> 2GB of swap make the difference?
>
> When you ray-trace an image, PyMOL will tell you exactly how much memory
> it's using.  My suspicion is that this is painfully close to the physical
> memory limit.  From what I understand, swap space does not give you more
> memory but Linux will transfer idle programs to swap, freeing up more
> RAM for PyMOL.  I have no idea how Windows works (other than "badly").
>
> 'top' in Linux or the Task Manager in WinXP should tell you how much
> physical memory is being used at any time.
>
> ----------------------------------------------------------------------------
> Nathaniel Echols
> Programmer
> na...@bi...                                        Gerstein
> Lab
> 203-589-6765                                                 Yale
> University
> ----------------------------------------------------------------------------

RE: [PyMOL] pymol crashes with large pdb files

[Warren L. D. <wa...@de...>]

That's likely the problem.

If you need to reduce RAM consumption during raytracing (at the cost of
performance), try reducing hash_max to 80 or 60.

set hash_max,60

Cheers,
Warren


--
mailto:wa...@de...
Warren L. DeLano, Ph.D.
Principal Scientist
DeLano Scientific LLC
Voice (650)-346-1154 
Fax   (650)-593-4020

> -----Original Message-----
> From: pym...@li... [mailto:pymol-users-
> ad...@li...] On Behalf Of an...@bi...
> Sent: Wednesday, June 25, 2003 12:28 PM
> To: Nat Echols
> Cc: pymol
> Subject: Re: [PyMOL] pymol crashes with large pdb files
> 
> Limited physical memory might indeed be the problem.
> When I ray-trace under WinXP, memory usage increases sharply (to about
> 500MB) and after a second or so, pymol crashes.
> When I ray-trace under linux, almost all physical memory (1GB) is
being
> used. Barely 40MB available.
> 
> We should be getting new computers next week. We'll see if 2GB will do
the
> trick for Windows.
> 
> 
> Andreas
> 
> 
> >> Now there is a twist to the story. We also run linux (redhat with
> 2.4.18
> >> kernel) on a 2 CPU 930MHz (1GB RAM, Nvidia GeForce2 GTS) PIII.
Working
> >> speed is much slower but no crashes when ray-tracing from the exact
> same
> >> pml script.
> >> Can it be that linux makes better use of its physical memory or
that
> the
> >> 2GB of swap make the difference?
> >
> > When you ray-trace an image, PyMOL will tell you exactly how much
memory
> > it's using.  My suspicion is that this is painfully close to the
> physical
> > memory limit.  From what I understand, swap space does not give you
more
> > memory but Linux will transfer idle programs to swap, freeing up
more
> > RAM for PyMOL.  I have no idea how Windows works (other than
"badly").
> >
> > 'top' in Linux or the Task Manager in WinXP should tell you how much
> > physical memory is being used at any time.
> >
> >
------------------------------------------------------------------------
> ----
> > Nathaniel Echols
> > Programmer
> > na...@bi...
Gerstein
> > Lab
> > 203-589-6765                                                 Yale
> > University
> >
------------------------------------------------------------------------
> ----
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: INetU
> Attention Web Developers & Consultants: Become An INetU Hosting
Partner.
> Refer Dedicated Servers. We Manage Them. You Get 10% Monthly
Commission!
> INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php
> _______________________________________________
> PyMOL-users mailing list
> PyM...@li...
> https://lists.sourceforge.net/lists/listinfo/pymol-users

RE: [PyMOL] pymol crashes with large pdb files

[Andreas <an...@bi...>]

Thanks Warren for pymol 0.90. I started some serious ray-tracing jobs
last night, but stopped this morning when I saw your email about the new
faster version. With pymol 0.88, each image took 2:40h, with pymol 0.90
it takes 40 minutes. That's on a 2CPU PIII Xeon (linux). Fourfold!!

The windows boxes still crash with full proteasomes, even when I
decrease hash_max significantly. But that doesn't matter too much now.


Andreas


On Thu, 2003-06-26 at 09:24, Warren L. DeLano wrote:
> That's likely the problem.
> 
> If you need to reduce RAM consumption during raytracing (at the cost of
> performance), try reducing hash_max to 80 or 60.
> 
> set hash_max,60


-- 
                    Andreas Förster
Dept of Biochem, Univ of Utah, 20N 1900E, #2460 Eccles Bldg.
Salt Lake City, UT 84132, phone: 001.801.585.3919
home: 1736 Kensington, SLC, UT 84108, 001.801.463.3607
        http://www.biochem.utah.edu/~andreas