VMD-L Mailing List

From: Burlen Loring (burlen.loring_at_gmail.com)
Date: Wed Apr 01 2015 - 13:46:17 CDT

I too am curious to see if it's the stack or something else. My 20 core
system has ulimit 8192 kb, much smaller, and no crash. I'm afraid it
would take intervention from a higher power for the sys admin to make
any changes! We're stuck on this front unless one of the user makes a
BIG stink that 4 cores is too slow...

On 4/1/2015 11:34 AM, John Stone wrote:
> Hi,
> I still wonder about this, as the ulimit values you've reported
> don't seem wildly out of whack compared to mine:
> bash-3.00$ ulimit -s -v
> stack size (kbytes, -s) 10240
> virtual memory (kbytes, -v) unlimited
>
> That said, there may very well be some strange behavior in your
> pthreads implementation that relates to the stack size, and if
> limiting your VMD process to 4 threads cures the problem, that
> seems reasonable if your sysadmins aren't willing to adjust the
> process limits otherwise.
>
> Cheers,
> John Stone
> vmd_at_ks.uiuc.edu
>
> On Wed, Apr 01, 2015 at 11:22:20AM -0700, Burlen Loring wrote:
>> It works up to 4 threads! the ulimits are
>>
>> stack size (kbytes, -s) 262144
>> virtual memory (kbytes, -v) 2097152
>>
>> vm is restricted on the login node to 2GB. It's not allowed for us to
>> change these ulimits... I think 4 threads is good enough! We can consider
>> the issue solved if you like.
>>
>> Thanks for your help!
>> Burlen
>>
>> On 4/1/2015 11:01 AM, Axel Kohlmeyer wrote:
>>
>> hmm... what is the stack size limit on those machines?
>> for multi-threaded operations, especially with many CPU cores
>> available, you may run out of stack space.
>>
>> also, you could try limiting the number of available threads to 1
>> using the environment variable VMDFORCECPUCOUNT
>>
>> axel.
>>
>> On Wed, Apr 1, 2015 at 1:09 PM, Burlen Loring [1]<burlen.loring_at_gmail.com> wrote:
>>
>> Hi John,
>>
>> Here is the result of issue and uname.
>>
>> Scientific Linux SL release 5.5 (Boron)
>> Linux cvrsvc03 2.6.32-358.18.1.el6.nersc.x86_64 #1 SMP Wed Aug 28 02:17:42
>> PDT 2013 x86_64 x86_64 x86_64 GNU/Linux
>>
>> I also have the same segv on the newer system image, which is a login
>> option.
>>
>> Scientific Linux release 6.3 (Carbon)
>> Linux cvrsvc03 2.6.32-358.18.1.el6.nersc.x86_64 #1 SMP Wed Aug 28 02:17:42
>> PDT 2013 x86_64 x86_64 x86_64 GNU/Linux
>>
>> The binary I tried is
>>
>> vmd-1.9.2.bin.LINUXAMD64-RHEL5.opengl.tar.gz
>>
>> my build has the following options
>>
>> LINUXAMD64 SURF STRIDE OPENGL TACHYON IMD NOSILENT TCL TK FLTK PYTHON
>> PTHREADS
>>
>> Burlen
>>
>>
>> On 4/1/2015 7:51 AM, John Stone wrote:
>>
>> Burlen,
>> This is helpful. What version of Linux is on the system you're using,
>> and which of the 64-bit VMD Linux packages did you install? I'm
>> particularly
>> interested in the output of these commands:
>> uname -a
>> cat /etc/issue
>>
>> Cheers,
>> John
>>
>> On Wed, Apr 01, 2015 at 01:23:28AM -0700, Burlen Loring wrote:
>>
>> Hi John,
>>
>> Thanks for the help! So I tried text mode as you suggest but the
>> crash is the same (output below).
>>
>> Although the OpenGL window pops up and I can interact with it there
>> may be some OpenGL issue. I'll have to inquire as to the hardware
>> etc. glxinfo command is not installed. This is a cluster, the login
>> nodes may have some crappy graphics cards, folks are using it with
>> NX, and I think this is the source of GLX Indirect. I'm pretty sure
>> that vmd was running here at some point in the past. Again I'll have
>> to ask the guy I'm taking over for ;-)
>>
>> Burlen
>>
>> $vmd -dispdev text
>> Info) VMD for LINUXAMD64, version 1.9.2 (February 19, 2015)
>> Info) [2]http://www.ks.uiuc.edu/Research/vmd/
>> Info) Email questions and bug reports to [3]vmd_at_ks.uiuc.edu
>> Info) Please include this reference in published work using VMD:
>> Info) Humphrey, W., Dalke, A. and Schulten, K., `VMD - Visual
>> Info) Molecular Dynamics', J. Molec. Graphics 1996, 14.1, 33-38.
>> Info) -------------------------------------------------------------
>> Info) Multithreading available, 8 CPUs detected.
>> Info) Free system memory: 38998MB (80%)
>>
>> vmd > mol new 1ubq.pdb
>> Info) Using plugin pdb for structure file 1ubq.pdb
>> Info) Using plugin pdb for coordinates from file 1ubq.pdb
>> Info) Determining bond structure from distance search ...
>> Segmentation fault
>>
>> On 3/31/2015 5:56 PM, John Stone wrote:
>>
>> Hi,
>> From the logs you sent, I noticed this, which is a bit of a red flag:
>> Info) OpenGL renderer: Mesa GLX Indirect
>> What graphics hardware do you have in this machine?
>>
>> If you run VMD in text mode, and load the same file, do you have
>> the same seg fault? Run VMD in text mode like this:
>> vmd -dispdev text
>>
>> If you do NOT have the seg fault in this case, then I am assuming
>> problems with the OpenGL graphics configuration. On the other hand,
>> if you do encounter a seg fault here when running in text mode, then
>> we will need to dig into this in much more detail.
>>
>> If you can run the text mode test and get back to me, that'd be great.
>>
>> Cheers,
>> John Stone
>> [4]vmd_at_ks.uiuc.edu
>>
>>
>> On Tue, Mar 31, 2015 at 02:20:44PM -0700, Burlen Loring wrote:
>>
>> Hi ,
>>
>> I am having some issue with vmd 1.9.2. As soon as I open a file (the
>> first example in the tutorial) there is a crash. I am pasting a
>> stack trace and the output below when running: vmd -debug. First I
>> tried the binary from the vmd site, then I tried building form
>> source, but the crash is the same. Any idea what is going wrong
>> here?
>>
>> Thanks
>> Burlen
>>
>> $vmd -debug
>> ***
>> *** Running VMD in debugger, type 'run' at debugger prompt
>> ***
>> GNU gdb (GDB) Red Hat Enterprise Linux (7.0.1-23.el5_5.1)
>> Copyright (C) 2009 Free Software Foundation, Inc.
>> License GPLv3+: GNU GPL version 3 or later
>> [5]<http://gnu.org/licenses/gpl.html>
>> This is free software: you are free to change and redistribute it.
>> There is NO WARRANTY, to the extent permitted by law. Type "show copying"
>> and "show warranty" for details.
>> This GDB was configured as "x86_64-redhat-linux-gnu".
>> For bug reporting instructions, please see:
>> [6]<http://www.gnu.org/software/gdb/bugs/>...
>> Reading symbols from
>> /global/common/carver/graphics/vmd/1.9.2/lib/vmd/vmd_LINUXAMD64...done.
>> (gdb) run
>> Starting program:
>> /global/common/carver/graphics/vmd/1.9.2/lib/vmd/vmd_LINUXAMD64
>> [Thread debugging using libthread_db enabled]
>> [New Thread 0x50000940 (LWP 26008)]
>> Info) VMD for LINUXAMD64, version 1.9.2 (February 19, 2015)
>> Info) [7]http://www.ks.uiuc.edu/Research/vmd/
>> Info) Email questions and bug reports to [8]vmd_at_ks.uiuc.edu
>> Info) Please include this reference in published work using VMD:
>> Info) Humphrey, W., Dalke, A. and Schulten, K., `VMD - Visual
>> Info) Molecular Dynamics', J. Molec. Graphics 1996, 14.1, 33-38.
>> Info) -------------------------------------------------------------
>> Info) Multithreading available, 8 CPUs detected.
>> Info) Free system memory: 38022MB (78%)
>> Info) OpenGL renderer: Mesa GLX Indirect
>> Info) Features: STENCIL MDE MTX PP PS
>> Info) GLSL rendering mode is NOT available.
>> Info) Textures: 2-D (2048x2048), Multitexture (8)
>> vmd > Info) Using plugin pdb for structure file
>> /global/common/carver/graphics/vmd/tutorial/1ubq.pdb
>> Info) Using plugin pdb for coordinates from file
>> /global/common/carver/graphics/vmd/tutorial/1ubq.pdb
>> Info) Determining bond structure from distance search ...
>> [New Thread 0x60001940 (LWP 26018)]
>> [New Thread 0x70002940 (LWP 26019)]
>> [Thread 0x70002940 (LWP 26019) exited]
>> [New Thread 0x7fffe2fc4940 (LWP 26020)]
>> [Thread 0x60001940 (LWP 26018) exited]
>> [Thread 0x7fffe2fc4940 (LWP 26020) exited]
>> [New Thread 0x7fffd2fc3940 (LWP 26021)]
>> [Thread 0x7fffd2fc3940 (LWP 26021) exited]
>> [New Thread 0x7fffc2fc2940 (LWP 26022)]
>> [New Thread 0x7fffb2fc1940 (LWP 26023)]
>> [Thread 0x7fffc2fc2940 (LWP 26022) exited]
>> [Thread 0x7fffb2fc1940 (LWP 26023) exited]
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x00007fffef3c8a38 in pthread_join () from /lib64/libpthread.so.0
>> (gdb) where
>> #0 0x00007fffef3c8a38 in pthread_join () from /lib64/libpthread.so.0
>> #1 0x00000000004c2a18 in vmd_bondsearch_thr(float const*, float
>> const*, GridSearchPairlist*, int, int**, int*, int**, int, float) ()
>> #2 0x000000004011eb85 in ?? ()
>> #3 0x0000000000000000 in ?? ()
>> (gdb)
>>
>> References
>>
>> Visible links
>> 1. mailto:burlen.loring_at_gmail.com
>> 2. http://www.ks.uiuc.edu/Research/vmd/
>> 3. mailto:vmd_at_ks.uiuc.edu
>> 4. mailto:vmd_at_ks.uiuc.edu
>> 5. http://gnu.org/licenses/gpl.html
>> 6. http://www.gnu.org/software/gdb/bugs/
>> 7. http://www.ks.uiuc.edu/Research/vmd/
>> 8. mailto:vmd_at_ks.uiuc.edu