Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Honored Contributor I
10 Views

Query to Intel Fortran team on acceptance testing of compiler updates

Consider a contentious thread https://software.intel.com/en-us/comment/reply/754827/1917824?quote=1#comment-form, here's a question for Lorri Menard and the Intel Fortran team:

Lorri Menard (Intel) wrote:

.. And, the gcc/gfortran test suites have been made available to the compiler developers, because as you say, there's nothing special/private/whatever in the test source. ..

Hi Lorri and Intel Fortran team,

With respect to the issues touched upon by several recent posts in connection with compiler regressions, Intel Fortran users will really benefit if the Intel Fortran team can share some information such as:

  1. Some indicators of the size and coverage of Intel's own regression suite relative to various Fortran standard revisions (90, 95, 2003, 2008, etc.) and other value-added facilities such as optimization, vectorization, other parallelization such as OpenMP, etc.
  2. What success rate is considered acceptable for major compiler releases and subsequent updates relative to Intel's own regression suite?  Naturally it can't be 100% in all aspects, otherwise Intel may not be able to issue any updates at any reasonable frequency!!  But hopefully in some critical aspects such as support up to a certain Fortran standard say 2003 or F95, and with respect to regression avoidance of previously reported ICEs, etc., can one hope and expect for Intel team to shoot for zero incidents in each update!?
  3. When users report issues such as compiler bugs at the Intel OSC, what mechanism is in place to extend the regression suite based on such incident requests?  For example, consider this thread: https://software.intel.com/en-us/forums/intel-visual-fortran-compiler-for-windows/topic/755166.  What are the chances the regression case shown therein will be reviewed critically for inclusion into the Intel's test suite if the existing tests do not address the issue?  That is, what level of certainty can users have this particular issue will NOT RECUR in any future Intel Fortran compiler update once the regression in compiler 18.0.1 is fixed?

Appreciate your response greatly. thanks much for your time and attention.

Regards,

 

0 Kudos
1 Reply
Highlighted
Moderator
10 Views

Fortran Fan,

Your input is highly appreciated. We value your feedback greatly and will look to use it to improve our software products. Unfortunately, for various reasons, we cannot discuss our internal testing practices in the public forum. (At least not without consulting our legal team first. ;-) Thank you for your understanding. 

 

0 Kudos