>What do you mean by "one process" when it is clearly 3<br>
>processes are running ? When you say 2 processes of tst,<br>
>do you mean you launched two instances of<br>
>"mpiexec -n 3 ./tst" simultaneously.<br><br>Yes, I meant of launching two instances of "mpiexec -n 3 ./tst" simultaneously.<br><br>I could not reach the MPICH cluster until monday, so that I'll send the modified programme on Monday.<br>
<br>Thank You,<br>Waruna<br><br><div class="gmail_quote">2008/11/29 Anthony Chan <span dir="ltr"><<a href="mailto:chan@mcs.anl.gov">chan@mcs.anl.gov</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
>From your Readme file:<br>
<br>
> mpiexec -n 3 <path><br>
><br>
> output:<br>
> when only one process runs<br>
><br>
> 0:Total: 499999999500000000 : Time: 9.272023<br>
> 2:Total: 499999999500000000 : Time: 10.722239<br>
> 1:Total: 499999999500000000 : Time: 11.324907<br>
><br>
> When two processes of 'tst' runs in the same time<br>
><br>
> 0:Total: 499999999500000000 : Time: 9.538206<br>
> 2:Total: 499999999500000000 : Time: 16.045104<br>
> 1:Total: 499999999500000000 : Time: 22.400754<br>
<br>
What do you mean by "one process" when it is clearly 3<br>
processes are running ? When you say 2 processes of tst,<br>
do you mean you launched two instances of<br>
"mpiexec -n 3 ./tst" simultaneously.<br>
<br>
Could you add the following lines to your tst.cpp<br>
<br>
char host_name[ MPI_MAX_PROCESSOR_NAME ];<br>
int namelen;<br>
MPI_Get_processor_name( host_name, &namelen );<br>
printf("rank %d running on %s\n", rank, host_name);<br>
<br>
before first fflush(stdout) to show the location of each process.<br>
Rerun your experiments and let us know the result.<br>
<br>
Also, you can use MPI_Wtime() instead of Duration.cpp that may<br>
simplify your test program.<br>
<div><div></div><div class="Wj3C7c"><br>
A.Chan<br>
<br>
----- "Waruna Ranasinghe" <<a href="mailto:warunapww@gmail.com">warunapww@gmail.com</a>> wrote:<br>
<br>
> Hi Rajeev,<br>
> Here's test programme. It's just for testing. The results I got, are<br>
> in the<br>
> read me file in the attachment<br>
><br>
> Thank You<br>
> Waruna<br>
><br>
> 2008/11/27 Rajeev Thakur <<a href="mailto:thakur@mcs.anl.gov">thakur@mcs.anl.gov</a>><br>
><br>
> > In that case, what kind of program are you running? Can you send us<br>
> a<br>
> > small test program.<br>
> ><br>
> > Rajeev<br>
> ><br>
> > ------------------------------<br>
> > *From:* <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a> [mailto:<br>
> > <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a>] *On Behalf Of *Waruna Ranasinghe<br>
> > *Sent:* Wednesday, November 26, 2008 9:05 PM<br>
> ><br>
> > *To:* <a href="mailto:mpich-discuss@mcs.anl.gov">mpich-discuss@mcs.anl.gov</a><br>
> > *Subject:* Re: [mpich-discuss] Clarification needed on<br>
> > runningprocessesonMPICH2<br>
> ><br>
> > Hi Rajeev,<br>
> > There's no doubt that process runs on all 3 machines.<br>
> > I have tried cpi example and it prints the hostnames of 3 machines<br>
> ><br>
> > mpiexec -l -n 3 <path to process><br>
> > (the same path is available in all 3 machines)<br>
> ><br>
> > Thank you<br>
> > Waruna<br>
> ><br>
> ><br>
> > 2008/11/26 Rajeev Thakur <<a href="mailto:thakur@mcs.anl.gov">thakur@mcs.anl.gov</a>><br>
> ><br>
> >> Make sure the processes are actually running on the 3 machines.<br>
> Try the<br>
> >> cpi example in the examples directory. It prints out the hostname.<br>
> How are<br>
> >> you running the job?<br>
> >><br>
> >> Rajeev<br>
> >><br>
> >> ------------------------------<br>
> >> *From:* <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a> [mailto:<br>
> >> <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a>] *On Behalf Of *Waruna<br>
> Ranasinghe<br>
> >> *Sent:* Wednesday, November 26, 2008 11:07 AM<br>
> >><br>
> >> *To:* <a href="mailto:mpich-discuss@mcs.anl.gov">mpich-discuss@mcs.anl.gov</a><br>
> >> *Subject:* Re: [mpich-discuss] Clarification needed on running<br>
> >> processesonMPICH2<br>
> >><br>
> >> Fedora 8 - MPICH2 one machine with core-2-duo - master<br>
> >> two machines with one core<br>
> >> all together 3 nodes<br>
> >><br>
> >> 2008/11/26 Rajeev Thakur <<a href="mailto:thakur@mcs.anl.gov">thakur@mcs.anl.gov</a>><br>
> >><br>
> >>> What kind of environment are you running on (how many machines,<br>
> how<br>
> >>> many cores each)?<br>
> >>><br>
> >>> Rajeev<br>
> >>><br>
> >>> ------------------------------<br>
> >>> *From:* <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a> [mailto:<br>
> >>> <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a>] *On Behalf Of *Waruna<br>
> Ranasinghe<br>
> >>> *Sent:* Wednesday, November 26, 2008 2:15 AM<br>
> >>> *To:* <a href="mailto:mpich-discuss@mcs.anl.gov">mpich-discuss@mcs.anl.gov</a><br>
> >>> *Subject:* Re: [mpich-discuss] Clarification needed on running<br>
> processes<br>
> >>> onMPICH2<br>
> >>><br>
> >>> Hi Anthony,<br>
> >>> I'm sorry, in this case I have to disagree with you.<br>
> >>> Because it is not max(t1,t2) but t1+t2 (this is the real result I<br>
> got)<br>
> >>> I use fflush(stdout) also<br>
> >>><br>
> >>> What I think is that, Cluster runs two programmes alternatively.<br>
> >>> i.e. Process A runs for a while (say t3 seconds) then process B<br>
> for<br>
> >>> sometime and so on.<br>
> >>> Therefore, ultimately both the processes will run for t1+t2<br>
> >>><br>
> >>> Cluster: MPICH2<br>
> >>> Fedora 8<br>
> >>><br>
> >>> 2008/11/26 Anthony Chan <<a href="mailto:chan@mcs.anl.gov">chan@mcs.anl.gov</a>><br>
> >>><br>
> >>>><br>
> >>>> If process A and B are launched by mpiexec, the time taken<br>
> >>>> by mpiexec should be max(t1,t2) not t1 + t2. As Rajeev said,<br>
> >>>> calling fflush(stdout) after each printf() is the fastest way<br>
> >>>> to get each process's stdout printed to your console.<br>
> >>>><br>
> >>>> A.Chan<br>
> >>>> ----- "Waruna Ranasinghe" <<a href="mailto:warunapww@gmail.com">warunapww@gmail.com</a>> wrote:<br>
> >>>><br>
> >>>> > Hi Rajeev,<br>
> >>>> > Actually this is not what I'm talking about.<br>
> >>>> > Say there are processes A and B<br>
> >>>> > Process A utilize 100% of the cpu while it runs. The answer<br>
> will<br>
> >>>> > appear in<br>
> >>>> > t1 seconds<br>
> >>>> > Process B also utilize 100% of the cpu while it runs. The<br>
> answer will<br>
> >>>> > appear<br>
> >>>> > in t2 seconds<br>
> >>>> ><br>
> >>>> > When I run both the process A and B at the same time, The both<br>
> answers<br>
> >>>> > will<br>
> >>>> > appear in t1 + t2 seconds. where as I want to get the Process<br>
> A's<br>
> >>>> > answer<br>
> >>>> > first. (Here Process A is submitted before Process B)<br>
> >>>> ><br>
> >>>> > Is there anything that I can do to make this happen.<br>
> >>>> ><br>
> >>>> > Thank You,<br>
> >>>> > Waruna<br>
> >>>> ><br>
> >>>> > 2008/11/26 Rajeev Thakur <<a href="mailto:thakur@mcs.anl.gov">thakur@mcs.anl.gov</a>><br>
> >>>> ><br>
> >>>> > > If you are refering to the output of "printf", you can try<br>
> adding<br>
> >>>> > an<br>
> >>>> > > fflush(stdout) after the printf. You don't have much control<br>
> over<br>
> >>>> > the order<br>
> >>>> > > in which it is printed from different processes.<br>
> >>>> > ><br>
> >>>> > > Rajeev<br>
> >>>> > ><br>
> >>>> > > ------------------------------<br>
> >>>> > > *From:* <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a> [mailto:<br>
> >>>> > > <a href="mailto:mpich-discuss-bounces@mcs.anl.gov">mpich-discuss-bounces@mcs.anl.gov</a>] *On Behalf Of *Waruna<br>
> Ranasinghe<br>
> >>>> > > *Sent:* Tuesday, November 25, 2008 1:35 AM<br>
> >>>> > > *To:* <a href="mailto:mpich-discuss@mcs.anl.gov">mpich-discuss@mcs.anl.gov</a><br>
> >>>> > > *Subject:* [mpich-discuss] Clarification needed on running<br>
> processes<br>
> >>>> > on<br>
> >>>> > > MPICH2<br>
> >>>> > ><br>
> >>>> > > Hi all,<br>
> >>>> > > I submitted 3 processes at the same time using mpiexec. the<br>
> results<br>
> >>>> > of each<br>
> >>>> > > process appears only after all the processes have finished.<br>
> (I guess<br>
> >>>> > that<br>
> >>>> > > the way mpich schedule it).<br>
> >>>> > > What if I want to get the result of the process, which was<br>
> submitted<br>
> >>>> > first,<br>
> >>>> > > first. (like first in first out).<br>
> >>>> > ><br>
> >>>> > > I use MPICH2 cluster in Fedora 8 (with mpd)<br>
> >>>> > ><br>
> >>>> > > Thank you.<br>
> >>>> > > Waruna Ranasinghe<br>
> >>>> > ><br>
> >>>> > ><br>
> >>>><br>
> >>><br>
> >>><br>
> >><br>
> ><br>
</div></div></blockquote></div><br>