[BUG REPORT] Array index started with 1 in LAPACKE_dgeqp3()

[BUG REPORT] Array index started with 1 in LAPACKE_dgeqp3()

Wei W.'s picture

Sorry I failed to find a proper place to report bug on www.intel.com.

The sixth parameter of LAPACKE_dgeqp3() ("jpvt") which is an array storing sorting indices of another array. However it wrongly contains 

Fortran-style indices (starting from 1) but not C-style indices. 

This bug appears  in both MKL 11.0 and MKL 10.0.

Thanks.

5 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.
Tim Prince's picture

MKL implementations of netlib functions are required to work the same as the netlib reference source. If it doesn't do so, you could file a bug report on premier.intel.com. If you didn't register your license, thus creating a support account, you can do so at https://registrationcenter.intel.com.
If you are calling the Fortran-callable function, it should not be surprising if subscripts are Fortran compatible.

Gennady Fedorov (Intel)'s picture

no no, you don't need submit the bug report on premier,intel.com because of if this is the real problem, we will take and submit it from this forum.
of course if this is a real problem but I am not sure about this case. We need to check it.

Gennady Fedorov (Intel)'s picture

Hello Wei,
yes this is the real problem and it has been escalated into the mkl team. we will let you know when the status of the problem will be updated.
--Gennady

Wei W.'s picture

Hi Gennady,

Thanks a lot!

Login to leave a comment.