Multiple MPI task offload to the same MIC.

Hi all,

I am currently looking to offload from different MPI processor to the same MIC.

The offloads are asynchronous in nature.

What I would like to know is there a queue implemented by default that makes sure only a given MPI task at a time executes on the MIC?

Is there a partition of the MIC we can carry out such that each task offloads to a subset of cores?

How do I synchronise if no such thing exists?

Whats the solution?

Profile events on xeon by using perf


I am doing some experiments using xeon and make a comparison between xeon and AMD, I am using perf in both machines. My concern is that the results of my events in xeon are thousand times higher then the results from AMD, but the runtime on xeon is much better than the AMD. I am measuring cache, instructions and cpu-clock in both machines.

My application is a matrix multiplication, size 1000x1000 and I am running a sequential execution, not parallel yet. 


RH-6.4 + MPSS-3.6.1 + Intel 16.0.1 missing libcoi_device



    I have recently upgraded to MPSS-3.6.1 to make use of gcc 5.1 on the Phi but the linker of the offload code now fails trying to find libcoi_device

   It appears that this library is not installed on the host by MPSS-3.6.1 and  /opt/intel/mic only contains the host the libcoi_host

   Do I need to compile this library from the source code?



Error Message 

Cat not open file on Phi

This is my program,it is very simple.

 1 #include <stdio.h>
  2 #include <fcntl.h>
  3 #include <unistd.h>
  5 int
  6 main(){
  8         int fd;
  9         fd=open("./p",O_RDWR);
 10         if(fd<0)
 11                 printf("+++++\n");
 12         close(fd);
 14         return 0;
 15 }

I compile it .   icc open.c -o open -mmic

Then I forward it to phi.  scp open mic0:/home/wj/test/data/soc

SCIF error

We have application in python which call to C++/C to scif to communicate with Fortran/C++ server on Xeon Phi (CPU <--> Xeon Phi on the same computer). This application runs one a month. In the past this works fine, however  recently we got the following error in kernel (see the end).
To resolve this we needs to reboot the computer, reboot Xeon Phi(s) alone did not resolve the problem.  We have like to ask if there is any way to
resolve this issue without reboot the computer.
   Many thanks,

Subscribe to Unix*