2 Replies Latest reply on Aug 5, 2011 5:42 PM by drodo

    Two-Step rccerun ?


      Hello everyone,


      I have observed the following when running both the pingpong example, as well as some RCCE pieces of code of my own writting (the following examples come from a pingpong execution). When I call the rccerun with the usual arguments, there is a first brief round of execution of the following form:


      pssh -h PSSH_HOST_FILE.25174 -t -1 -p 2 /shared/drodo/mpb.25174 < /dev/null


      and afterwards, the actual code is being executed:


      pssh -h PSSH_HOST_FILE.24803 -t -1 -P -p 2 /shared/drodo/pingpong 2 1.0 00 01 < /dev/null


      On that observation I have two questions:


      1. What is the use of the first round of execution? Is it vital? Can it be omitted in any way?

      2. Sometimes, the first round fails (giving some error, most usually the 127 one), whereas the second (the actual thing I want to run) is completed without any problems. Do I have to worry that the mpb.XXXXX is not found/executed?


      Thank you for your time.





        • 1. Re: Two-Step rccerun ?

          You can look inside the rccerun script to see what is happening. But what rccerun is doing is running another program first .... it cleans up the MPB because the last time a rcce program may not have left things in a good state. Look at Section 2.2 in the RCCE Specification.

          There is no guarantee that the MPBs are in a clean state, at the beginning of a RCCE execution. You can explicitly wipe the  MPBs by executing mpb –c on the cores. Run it on each core whose MPB you want to clear.


          That's what the first run you see is doing. It's safer to leave it there. I don't know why the first one sometimes fails. 127 usually means that it can't find the program. This utility is put in your shared directory when you run a ./configure SCC_LINUX. It might have just gotten deleted. Try reconfiguring RCCE.

          1 of 1 people found this helpful
          • 2. Re: Two-Step rccerun ?

            Thanks for the immediate reply!


            I will report if I come across anything more specific!