[mpich-discuss] comparison of execution times:

[Jayesh Krishna]

Hi,
 Also try the nemesis channel (mpiexec -n 2 -channel nemesis MYMPIPGM.exe), if you are not already doing so. Nemesis is going to be the default channel on windows in the upcoming MPICH2 1.3 series.

(PS: Nemesis provides better performance than the old shm channel.)
Regards,
Jayesh
----- Original Message -----
From: "Rajeev Thakur" <thakur at mcs.anl.gov>
To: mpich-discuss at mcs.anl.gov
Sent: Sunday, May 23, 2010 5:37:44 PM GMT -06:00 US/Canada Central
Subject: Re: [mpich-discuss] comparison of execution times:



It is possible that there is better cache utilization in the parallel program than in the sequential one. As an experiment, run the sequential program with only 250 iterations and see if the time is less than half the time taken by the sequential program with 500 iterations. 

Rajeev 





From: mpich-discuss-bounces at mcs.anl.gov [mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of sankar s 
Sent: Sunday, May 23, 2010 6:02 AM 
To: mpich-discuss at mcs.anl.gov 
Subject: [mpich-discuss] comparison of execution times: 


Hi, 


I have done a simulation program in C, which contains 500 Replications. In sequential code these 500 replications will be done one by one and the combined results will be posted in output at last. 


Same code I have written for MPI. first 250 replications in processor 0 and next 250 replications in processor 1. after completion of processor 0, it will send these results to processor 1. Processor 1 combines all the results and gives the output. 


I expected, because of doing these parallelism, the execution time of MPI should be half or some what more than half than sequential code. 


But parallel code is taking less than half execution time of sequential code. Some times (1/3rd also). Results point of view it is good. But I don't understand why it is taking less than half time? 


am using DELL core2duo processor and windows 7 OS. 




Thank you 
_______________________________________________
mpich-discuss mailing list
mpich-discuss at mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss