Computação paralela

MSVC 2013 ICC 15.0: interlocked instructions do not make memory barrier for optimization?

Greetings,

I have my own very fast critical section implementation with interlocked intrinsic functions. It seems to be failing.

I guess ICC IPO optimizer should invalidate program/variable state when it encounters interlocked intrinsic function, even when they are volatile and not so.

Wrong results while solving for eign vectors!

i am trying to use the dsyev routine. Now i modified your default example slightly by solving for matrix:

2,1,3,
1,2,3,
3,3,20	

i am getting :
Eigenvalues
   1.00   2.00  21.00
Eigenvectors (stored columnwise)
   0.71   0.69   0.16
  -0.71   0.69   0.16
   0.00  -0.23   0.97

i again tried with vector:
1,2
2,1

MPI_Request_free() hangs

Hi,

Trying to make use of the MPI persistent communication primitives in our application, I'm ending up with the following sequence of events:

MPI_Ssend_init(msg, msg_length, MPI_BYTE, 0, tag, comm, &req);
MPI_Start(&req);
MPI_Cancel(&req);
MPI_Wait(&req, MPI_STATUS_IGNORE);
MPI_Request_free(&req); // <-- HANGS

The only other node is blocked in an MPI_Barrier(comm);

Multithreading (BLAS sgemm)

Dear Forum,

I am trying to make MKL accelerate a matrix multiplication for me. It works, but MKL insists on doing it with a single thread. I played around a bit. But regardless of what I do - even when multiplying two randomly initialized 10000x10000 matrices - MKL does not use multiple threads. Am I missing something?

Function:

BLAS sgemm, via libmkl_rt.so

 

Environment settings:

Assine o Computação paralela