Intel® Fortran Compiler
Build applications that can scale for the future with optimized code designed for Intel® Xeon® and compatible processors.
29487 Discussions

Webinar May 13 - What's New in Intel Fortran 16.0

Steven_L_Intel1
Employee
2,146 Views

Wed, May 13, 2015 12:00 PM - 1:00 PM EDT

This session will cover new features in the Intel® Fortran compiler version 16, part of Intel® Parallel Studio XE 2016. Topics include new features from Fortran standards, new OpenMP* features and changes for users of Microsoft Visual Studio*.

Register here

This session will be recorded and the recording made available sometime afterward.

0 Kudos
13 Replies
FortranFan
Honored Contributor III
2,146 Views

Thanks, Steve, for your notification.

Intel Fortran 16.0 might be quite an enhancement and I'm waiting with bated breadth! :-)

 

0 Kudos
Steven_L_Intel1
Employee
2,146 Views

I think many will find things to like there.

0 Kudos
andrew_4619
Honored Contributor III
2,146 Views

We get sub-modules, I think ? :-)

Will there be some tools to create the interface sections for converting existing modules? 

0 Kudos
Steven_L_Intel1
Employee
2,146 Views

We're not working on such tools. An interesting idea....

0 Kudos
우왕희_WooWangHee
2,146 Views

Support Visual Studio 2013?

Can Debug In 2 Byte Character Name Folder? (Korean, Chinese, Japanese)

Can Jump to Defined Variable or Subroutine(Function)?

 

0 Kudos
andrew_4619
Honored Contributor III
2,146 Views

Steve Lionel (Intel) wrote:

We're not working on such tools. An interesting idea....

With the existing functionality e.g. generated interfaces checking and the need to robustly put some form of the same date in .mod files one would hope doing this would be a relatively simple job based on a compiler switch... It would be a neat feature that could save a lot of work for some users.

0 Kudos
Steven_L_Intel1
Employee
2,146 Views

VS2013 is already supported. If you meant VS2015, we support new VS versions as soon as practical after retail release.

Go To Definition has been in the product for a couple of years now. You do have to enable "Advanced" text editor features in Options - the documentation explains this.

0 Kudos
FortranFan
Honored Contributor III
2,146 Views
This comment has been moved to its own thread
0 Kudos
WSinc
New Contributor I
2,146 Views
This comment has been moved to its own thread
0 Kudos
Steven_L_Intel1
Employee
2,146 Views

A recording of the webinar will be made available later - I will update this thread when that happens.

Here are the questions received during the webinar:

Any Co-Array Fortran new features or optimizations? Some optimizations, no new features.

Are there improvements to the vectorization capabilities of the compiler? Each new release brings improvements in optimization.

For -init=snan, the man page for ifort 16 seems to imply that allocatables aren't supported. Is this a case of the man page lagging the release notes? Yes. I'll make sure the man page gets updated.

In your Submodule code example, assuming this code is located in a DLL project, how do I export(expose) subroutine Sub1? Do I put the DLLEXPORT directive in the interface located in the module? Do I also have to put this directive in the submodule? Is this directive going to "magically" become DLLIMPORT on procedure/module use in the main referencing program? It looks as if this is something we missed and I have reported it to the developers. My recommendation is that you'll need DLLEXPORT in the interface. If you don't use MODULE PROCEDURE, you'll also need it in the submodule.

Steve briefly mentioned that Visual Studio 2015 might be supported with an update to the Fortran 15.0 Visual Studio integration. When might this happen? Sometime this summer - it will be "latent" support for VS2015 CTP6.

Will the Intel Fortran samples include cases for new Fortran features such as submodules and enhanced interoperability with C (Fortran 2015 feature)? We'll try to get these in.

Will Fortran 16 have any new optimization features/capabilities? See above answer.

With new C interoperability features, do you foresee existing C code to change?  That is, existing C code that is working correctly with Fortran 2003/2008 definitions (ISO_C_BINDING and BIND(C)) for interoperable arguments? Existing code would not have to change, but I do see the opportunity for simplification and new benefits from the added features.

Can you comment on the ability to create new Fortran solutions in Visual Studio from existing projects? Not sure if I understand the question. Maybe you meant "new projects from existing source code"? That we already have. Otherwise, you can create an empty VS solution and add projects to it. Simply copy the project folder if you want to duplicate the project.

Is interprocedural optimization (IPO) in Intel compiler fully compatible with submodules? Yes.

Has Intel established a target timeline to become a full-feature Fortran 2008 compiler? Can't comment on a timeline, but we're getting closer.

Any changes in 16.0 version with respect to Visual Studio integration features such as editor for Fortran coders. Not this time.

 

 

0 Kudos
FortranFan
Honored Contributor III
2,146 Views

Thanks, Steve, for the great presentation and for posting the questions here - very informative and useful.

0 Kudos
Steven_L_Intel1
Employee
2,146 Views

Recording is available at https://software.intel.com/en-us/videos/what-s-new-in-intel-fortran-160-for-intel-parallel-studio-xe-2016

0 Kudos
TimP
Honored Contributor III
2,146 Views

I had already started installing VS2015 before getting the notice and watching the video.  Install is still running, after 7 hours.  In view of the information above (thanks), I'll try to integrate the 2015 update 4 rather than the beta ifort.  I'm mainly interested in whether CL itself optimizes better or worse than its predecessors when called from ifort.

ifort 16 beta has eliminated the requirement for all but one legacy !dir$ simd directive (in place of !$omp simd or no directive) in my benchmarks.  This is a good trend, even if at odds with the slogan about directive based vectorization.  There is still a lot of dependence on !dir$ vector aligned which accounts for much of the advantage over gfortran.  Also interesting to me is the ability to vectorize (more effectively, on a limited number of cores) without the simd clause under !$omp parallel (an ability not shared by any C or non-Intel compiler).

0 Kudos
Reply