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

Documentation Bug: mkl_dcsrbsr

bamberg
초급자
783 조회수

The documentation for mkl_dcsrbsr is incorrect. (See: http://software.intel.com/sites/products/documentation/hpc/compilerpro/en-us/cpp/win/mkl/refman/index.htm)

The description of the job array states thtat:

- If job(6)=0, all output arrays acsr, ja, and ia are filled in for the output storage.

- If job(6)0, only arrays ja, ia are filled in for the output storage.

But the behaviour of the library is exactly opposite.

Also, it might be helpful to state that this function is actually implementing the transpose operation, if used with two CSR or two CSC inputs.

best regards,

Brian Amberg

0 포인트
3 응답
Gennady_F_Intel
중재자
783 조회수

This problem has been already fixed in the latest 10.3 beta:

see, How it looks right now in 10.3 beta Reference Manual

job(6)- job indicator.

For conversion to the BSR format:

Ifjob(6)=0, only arraysjab,iabare generated for the output storage.

Ifjob(6)>0, all output arraysabsr,jab, andiabare filled in for the output storage.

Ifjob(6)=-1,iab(1)returns the number of non-zero blocks.

--Gennady
0 포인트
bamberg
초급자
783 조회수
Dear Gennady,
thank you for the reply. You actually copied the documentation of the wrong function. It reads:
For conversion to the CSC format:
If job(6)=0, only arrays ja1, ia1 are filled in for the output storage.
If job(6)!=0, all output arrays acsc, ja1, and ia1 are filled in for the output storage.
For conversion to the CSR format:
If job(6)=0, only arrays ja, ia are filled in for the output storage.
If job(6)!=0, all output arrays acsr, ja, and ia are filled in for the output storage.
It would be helpful to put the MKL 10.3 documentation online, because I just used the online documentation.
best regards,

Brian
0 포인트
Gennady_F_Intel
중재자
783 조회수
we can do it later with 10.3 Gold, but at this time you can register and download 10.3 beta here.
--Gennady
0 포인트
응답