topic There are SMP and distributed in Intel® oneAPI Math Kernel Library

How Large is a Large Matrix ?

Vishnu — Tue, 17 May 2016 10:52:46 GMT

I was looking through the LAPACK documentation and came across various instances where there are different routines recommended for large and small matrices. For eg., SYEVD vs SYEVR ( https://software.intel.com/en-us/node/469180 ) - { last paragraph of the description section } .

But nowhere that I could find, is there any mention of how large, large exactly is. Does someone here know?

Thanks!

There are SMP and distributed

Gennady_F_Intel — Tue, 17 May 2016 14:15:09 GMT

There are SMP and distributed versions for both of these routines. Therefore, the problem sizes will be very large ( fit with RAM available on SMP or Cluster Systems correspondingly ).

Quote:Gennady Fedorov (Intel)

Vishnu — Tue, 17 May 2016 15:04:06 GMT

Gennady Fedorov (Intel) wrote:

There are SMP and distributed versions for both of these routines. Therefore, the problem sizes will be very large ( fit with RAM available on SMP or Cluster Systems correspondingly ).

I am currently using it on an SMP system. But what is the relative qualification in input matrix sizes that the documentation makes? You say that "the problem sizes will be very large". Do you mean to say that both of them take very large inputs and that it doesn't matter? If so, what does the doc mean?