Intel® oneAPI Math Kernel Library
Ask questions and share information with other developers who use Intel® Math Kernel Library.

DSS parallel performance

shesheni
Beginner
455 Views

Hello! Now I try to migrate from CVF6.6/MKL 9.0 to IVF 9.1/MKL 9.0. A queston arose as follows. If I call DSS from a code compiled in CVF6.6, solution time is about 23 sec on one core of E6600or 13 sec on two cores. If I run the same program compiled in IVF the solution timesare about 23 sec and 26 seccorrespondingly.I really appreciate if anybody explain what I do wrong?Thanks in advance.

0 Kudos
1 Reply
Intel_C_Intel
Employee
455 Views

Well, the one thing we know has not changed is MKL. So changes in behavior, assuming the same set of parameters are used in both cases, has to be traced back to something different that is happening in the compiled environment.

Can you provide a simple example? The only thing that comes to mind is that somehow the data is laid out differently but there might be other causes for such dramatic differences in the behavior of your program as a function of the compiler. I might also suggest that you post to the compiler forum.

Bruce

0 Kudos
Reply