shared memory in computational chemistry

My quantum mechanical computational software (running
on amd64 etch with 3700mb per node, total ram 16GB) is
implicitly using shared memory segments to speed up
transfer outside the kernel. It is unable to allocate
a 38731776bytes segment, and the computation dies.

In fact, command "ipcsl -l" returns"
------ Shared Memory Limits --------
max number of segments = 4096
max seg size (kbytes) = 32768
max total shared memory (pages) = 2097152
min seg size (bytes) = 1

------ Semaphore Limits --------
max number of arrays = 128
max semaphores per array = 250
max semaphores system wide = 32000
max ops per semop call = 32
semaphore max value = 32767

------ Messages: Limits --------
max queues system wide = 16
max size of message (bytes) = 8192
default max size of queue (bytes) = 16384


QUESTIONS:
1) How to set shmmax in debian?

2) Which is the upper limit for the hardware indicated
abobe, should a relationship to the hardware be? (in
fact, I can't predict the size of segments that will
be tried to be allocated in the future i the MD
simulations I am carrying out).

3)What else - if anything - should be set besides
shmmax.

Thanks for help. I could not get help from the
software people. Probably because they are used at big
business (supercomputer centers) being unfamiliar with
the small business I am presenting here.

Cheers
francesco Pietra



____________________________________________________________________________________
Don't get soaked. Take a quick peek at the forecast
with the Yahoo! Search weather shortcut.
http://tools.search.yahoo.com/shortcuts/#loc_weather


--
To UNSUBSCRIBE, email to debian-user-REQUEST@xxxxxxxxxxxxxxxx
with a subject of "unsubscribe". Trouble? Contact listmaster@xxxxxxxxxxxxxxxx

Relevant Pages

  • Solaris 10: Increasing the process data space; shared memory segments & intimate shared memory p
    ... There is a startup parameter to the database server that will lock the shared memory segments into RAM so they are not be paged out. ... I don't see a resource control to change the SHMSEG segment size like you could in Solaris 9; Can't find anything about MAXUMEM for Solars. ...
    (comp.unix.solaris)
  • RE: onconfig recommendations
    ... almost have to set RESIDENT to -1 to prevent poor performance ... architectures restrictions of only four active shared memory segments ...
    (comp.databases.informix)
  • Re: onconfig recommendations
    ... Don't know HP model numbers, but it this is a PA-RISC machine, you almost have to set RESIDENT to -1 to prevent poor performance due to the architectures restrictions of only four active shared ... VERY important on PA-RISC if you are using shared memory connections. ... Though I had thought that the message segments were less of a problem than the virtual segments since each listener only attaches to one of those segments. ...
    (comp.databases.informix)
  • RE: onconfig recommendations
    ... almost have to set RESIDENT to -1 to prevent poor performance ... shared memory connections. ... For HPUX, I'll have to disagree with you on this one, Art. ... segments were less of a problem than the virtual segments since each ...
    (comp.databases.informix)
  • Re: [RFC][PATCH] /proc/pid/maps doesnt match "ipcs -m" shmid
    ... make sysv ipc shared memory use stacked files ... took away one of the debugging feature for shm segments. ... this shared memory segments. ... Very simply, pmap reports the shmid. ...
    (Linux-Kernel)