[gmx-users] Re: High frequency output slow down the simulation?
This is one reason why I developed the checkpointing script that is
available in the wiki under beginners -> checkpointing_jobs
Often a cluster node allows you to write to ${TMPDIR} which creates a
unique directory that is actually local and then you can copy the
contents to your final storage space.
Chris.
-- original message --
Yang Ye wrote:
Writing on local or remote certainly makes a difference. How about
the speed with no output at all?
Absolutely. Depending on the cluster setup, the best option is often
along the lines of
1) Write to disk physically associated with the CPUs (IIRC for GROMACS,
preferably the disk associated with MPI node 0, where applicable).
2) Write minimally often for your needs - using XTC rather than TRR
output wherever high precision is not required.
3) Arrange for the file to be returned to more convenient or permanent
file space at the end of the job.
This last step might be as simple as adding some cp commands to your
batch script, or perhaps issuing some "stage file out" instructions to
your batch system.
Your first port of call should be the cluster administrator /
documentation for how best you should be doing these steps. We can't
help much there.
Mark
___
gmx-users mailing listgmx-users@gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/search before posting!
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to [EMAIL PROTECTED]
Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
Yang Ye wrote: Writing on local or remote certainly makes a difference. How about the speed with no output at all? Absolutely. Depending on the cluster setup, the best option is often along the lines of 1) Write to disk physically associated with the CPUs (IIRC for GROMACS, preferably the disk associated with MPI node 0, where applicable). 2) Write minimally often for your needs - using XTC rather than TRR output wherever high precision is not required. 3) Arrange for the file to be returned to more convenient or permanent file space at the end of the job. This last step might be as simple as adding some cp commands to your batch script, or perhaps issuing some "stage file out" instructions to your batch system. Your first port of call should be the cluster administrator / documentation for how best you should be doing these steps. We can't help much there. Mark ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
Writing on local or remote certainly makes a difference. How about the speed with no output at all? Regards, Yang Ye From: xianghong qi <[EMAIL PROTECTED]> To: Discussion list for GROMACS users Sent: Sunday, November 9, 2008 12:36:40 AM Subject: Re: [gmx-users] Re: High frequency output slow down the simulation? The time difference is : for simulation with every 100ps output, 160ns/day; for simulation with every 0.1ps output, It has been running for 4 days, only wrote to 2300ps . I am not sure the local hard disk since I run my simulation in some cluster. Thanks. -Xianghong Qi On Sat, Nov 8, 2008 at 11:09 AM, Yang Ye <[EMAIL PROTECTED]> wrote: More details please. Local hard disk? How much the time difference? YY On 11/8/08 12:41 PM, xianghong qi wrote: Hello, all: I compare the two simulations with different output frequency for .xtc file in same machine. One with low frequency runs much faster than the one with high frequency. Is that reasonable? I think the frequency shouldn't affect the simulation speed. Anyone has idea about this situation? Thanks. -Xianghong Qi -- Some people make the world more special just by being in it. ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php -- Some people make the world more special just by being in it. ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
The time difference is : for simulation with every 100ps output, 160ns/day; for simulation with every 0.1ps output, It has been running for 4 days, only wrote to 2300ps . I am not sure the local hard disk since I run my simulation in some cluster. Thanks. -Xianghong Qi On Sat, Nov 8, 2008 at 11:09 AM, Yang Ye <[EMAIL PROTECTED]> wrote: > More details please. > Local hard disk? > How much the time difference? > > YY > > > On 11/8/08 12:41 PM, xianghong qi wrote: > >> Hello, all: >> I compare the two simulations with different output frequency for .xtc >> file in same machine. One with low frequency runs much faster than the one >> with high frequency. Is that reasonable? I think the frequency shouldn't >> affect the simulation speed. Anyone has idea about this situation? >> Thanks. -Xianghong Qi >> -- >> Some people make the world more special just by being in it. >> >> >> ___ >> gmx-users mailing listgmx-users@gromacs.org >> http://www.gromacs.org/mailman/listinfo/gmx-users >> Please search the archive at http://www.gromacs.org/search before >> posting! >> Please don't post (un)subscribe requests to the list. Use the www >> interface or send it to [EMAIL PROTECTED] >> Can't post? Read http://www.gromacs.org/mailing_lists/users.php >> > ___ > gmx-users mailing listgmx-users@gromacs.org > http://www.gromacs.org/mailman/listinfo/gmx-users > Please search the archive at http://www.gromacs.org/search before posting! > Please don't post (un)subscribe requests to the list. Use the www interface > or send it to [EMAIL PROTECTED] > Can't post? Read http://www.gromacs.org/mailing_lists/users.php > -- Some people make the world more special just by being in it. ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
thanks lot for your reply. My high frequency is every 0.1ps, low frequency is every 100ps. I am not sure whether every 0.1ps is extremly high or not. I am running with Gromacs 4.0. Up to now, I can see the only difference is the speed between high frequency simulation and low frequency simulation. Thanks. -Xianghongh Qi On Sat, Nov 8, 2008 at 4:31 AM, Jussi Lehtola <[EMAIL PROTECTED]>wrote: > On Sat, 2008-11-08 at 16:13 +1100, Mark Abraham wrote: > > xianghong qi wrote: > > > Hello, all: > > > > > > I compare the two simulations with different output frequency for .xtc > > > file in same machine. One with low frequency runs much faster than the > > > one with high frequency. > > > Is that reasonable? I think the frequency shouldn't affect the > > > simulation speed. > > > > > > Anyone has idea about this situation? Thanks. > > > > If that's really the only difference between the two simulations, then > > it suggests your file system has severe problems. With modern buffered > > I/O it's inconceivable that writing output could rate-limit GROMACS. > > (Unless something's broken in 4.x, but you haven't said what you're > using.) > > It is reasonable, if you have a ridiculously high output frequency, e.g. > you write the output on each step. Also the performance of parallel > simulations may be hurt more than of serial simulations, since there is > an extra communication step to collect the data. > -- > -- > Jussi Lehtola, FM, Tohtorikoulutettava > Fysiikan laitos, Helsingin Yliopisto > [EMAIL PROTECTED], p. 191 50623 > -- > Mr. Jussi Lehtola, M. Sc., Doctoral Student > Department of Physics, University of Helsinki, Finland > [EMAIL PROTECTED] > -- > > ___ > gmx-users mailing listgmx-users@gromacs.org > http://www.gromacs.org/mailman/listinfo/gmx-users > Please search the archive at http://www.gromacs.org/search before posting! > Please don't post (un)subscribe requests to the list. Use the > www interface or send it to [EMAIL PROTECTED] > Can't post? Read http://www.gromacs.org/mailing_lists/users.php > -- Some people make the world more special just by being in it. ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
More details please. Local hard disk? How much the time difference? YY On 11/8/08 12:41 PM, xianghong qi wrote: Hello, all: I compare the two simulations with different output frequency for .xtc file in same machine. One with low frequency runs much faster than the one with high frequency. Is that reasonable? I think the frequency shouldn't affect the simulation speed. Anyone has idea about this situation? Thanks. -Xianghong Qi -- Some people make the world more special just by being in it. ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
On Sat, 2008-11-08 at 16:13 +1100, Mark Abraham wrote: > xianghong qi wrote: > > Hello, all: > > > > I compare the two simulations with different output frequency for .xtc > > file in same machine. One with low frequency runs much faster than the > > one with high frequency. > > Is that reasonable? I think the frequency shouldn't affect the > > simulation speed. > > > > Anyone has idea about this situation? Thanks. > > If that's really the only difference between the two simulations, then > it suggests your file system has severe problems. With modern buffered > I/O it's inconceivable that writing output could rate-limit GROMACS. > (Unless something's broken in 4.x, but you haven't said what you're using.) It is reasonable, if you have a ridiculously high output frequency, e.g. you write the output on each step. Also the performance of parallel simulations may be hurt more than of serial simulations, since there is an extra communication step to collect the data. -- -- Jussi Lehtola, FM, Tohtorikoulutettava Fysiikan laitos, Helsingin Yliopisto [EMAIL PROTECTED], p. 191 50623 -- Mr. Jussi Lehtola, M. Sc., Doctoral Student Department of Physics, University of Helsinki, Finland [EMAIL PROTECTED] -- ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] Re: High frequency output slow down the simulation?
xianghong qi wrote: Hello, all: I compare the two simulations with different output frequency for .xtc file in same machine. One with low frequency runs much faster than the one with high frequency. Is that reasonable? I think the frequency shouldn't affect the simulation speed. Anyone has idea about this situation? Thanks. If that's really the only difference between the two simulations, then it suggests your file system has severe problems. With modern buffered I/O it's inconceivable that writing output could rate-limit GROMACS. (Unless something's broken in 4.x, but you haven't said what you're using.) Mark ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
[gmx-users] Re: High frequency output slow down the simulation?
Hello, all: I compare the two simulations with different output frequency for .xtc file in same machine. One with low frequency runs much faster than the one with high frequency. Is that reasonable? I think the frequency shouldn't affect the simulation speed. Anyone has idea about this situation? Thanks. -Xianghong Qi -- Some people make the world more special just by being in it. ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php
