Update 4 of Intel® Parallel Studio XE Composer Edition for Fortran macOS* 2019 Release, support for macOS* X 10.14.4

[Ron_Green]

Today we released Intel® Parallel Studio XE Composer Edition for Fortran macOS* 2019 Update 4.  This version adds support for macOS* X 10.14.4

You can download the 2019 Update 4 compiler from Intel Registration Service

[may_ka]

Downloaded, installed, run .......................... and crash with ICE.

Nice. Seems as "usable" as all releases since 17.08. So I got stuck with that version and I am scratching my head whether I should redirect our licenses fees to NAG. Intel's ifort has become a real source of pain. very sad.

cheers

[Juergen_R_R]

may.ka wrote:

..... and crash with ICE.

Nice. Seems as "usable" as all releases since 17.08.

So are you saying that with all v18 and all v19 of ifort you get an ICE with your code? Did you report this to the support? What is the status of that support request? ifort 18 came out as beta version I guess in  late 2017. So that would be a bug unfixed since more than 1.5 years. What is this ICE? Can you program around this part?

[may_ka]

Yes. I am saying that:

https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/783339

https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/796692

https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/796693

and of course other bugs which made progression from 17.08 impossible (e.g.):

https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/781714

https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/753808

With respect to the reporting ................ I think I remember you complaining when 19 came out that you boiled down ICEs to 700 lines of code, and Intel still couldn't find it. I had a similar thing with this

https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/796693

I think I handed in about 500 lines of code but Intel couldn't find it. Then I invested 48 hours and cooked it down to 20 lines with a precise error description and a workaround .................... But I don't think that that is my job.

Almost all ICEs I have seen came out of the blue (just reshaping code e.g. splitting function headers and bodies etc.) and occurred inside huge libraries with heavily interrelated code. So finding a workaround is not that easy/trivial, especially when you have problems such that you call a%c%fun() or b%c%fun(), and the compiler eats the former but gets nausea from the latter. Then you run everything in gfortran hoping that it will report some standard violations ............ but it is happy. etc. etc.

So if my time allows I'll try to narrow down the ICE but, you know, life is short and maybe I should just wait for ifort 21.

cheers

[Ron_Green]

may.ka:  make clean, tar and compress, attach and give me instructions to build so I can reproduce it.  I can cut it down to a small reproducer, I'm quite proficient at this sort of thing.  I'll get the bug filed.

Ron

[hannah__nic]

I am unable to download this product because the download link is broken. I have tried several times over the past two days and the download always gets cancelled within a minute or so. The url is the following:

http://registrationcenter-download.intel.com/akdlm/irc_nas/tec/15533/parallel_studio_xe_2019_update4_cluster_edition.tgz

If I try to go to the "Customizable Package" page at:

http://registrationcenter-download.intel.com/akdlm/irc_nas/tec/15533/parallel_studio_xe_2019_update4_cluster_edition_online.tgz

Then I receive a 404 error.

Do you have an alternative link that I can use? I am very keen try out this compiler on our new Skylake cluster.

Regards,

Nic

[Alex9]

I downloaded and run the software and it is very much useful to me for creating and building faster code for my Python project.

Thanks and Regards

[FortranFan]

may.ka wrote:

Yes. I am saying that:

.. Then you run everything in gfortran hoping that it will report some standard violations ............ but it is happy. etc. etc. ..

@may.ka,

I think your attempts at compiler comparisons are neither accurate nor appropriate for this forum.  Your comments in Quote #2 also appear to be neither useful nor relevant to this forum given a lack of specificity with the ICE and a tacit suggestion of some other regression(s) in Update 4 but with no details.

To the extent you are concerned about your own time and any "waste" thereof, you may take note

1. your issues are largely with certain aspects of Fortran 2003 and 2008 standard revisions for which the standard itself has had "bugs" too and so has every processor implementation out there of these features and this is similar with other developments in other computing aspects (C++, Python, .NET, software and hardware) specifically and all other fields involving humans generally; Fortran given its history struggles with resources everywhere including with Intel Fortran which presents its own challenges but
2. if you're not keen to locate and identify and assist in reporting specific new issues in newer Intel Fortran versions,you may wait until Intel support informs you all your reported issues e.g., your list in Quote #4 are resolved before you do any further "Downloaded, installed, run .."

A user can find some regression, perhaps a regression leading to the insidious ICE even, in a given compiler version, say 17.0 Update 5 and notice it takes several update cycles, say up to 20.0 Update 1 before the regression is resolved.  I think it will be better if the users work directly with Intel support (via OSC?) to address the elapsed time of resolution (e.g., 17.0 Update 5 to 20.0 Update 1 which can be >2 years) and how to shorten that.

But using the public forum to comment in the kind of manner in Quote #2 and #4 above of incidents which are of that commenter's own interest but that have not been resolved in particular Intel Fortran updates is something I personally do not find being inclusive of my interest as a reader of this forum.  I hope Intel Forum administrators address posts suitably.

[may_ka]

@FortranFan

thanks for you wisdom, greatly appreciated!!! You are a real mentor!!! I'll try hard to get to a more novel style of posting to be more inclusive of your interest and to give you a pleasureful bed time reading and a nice sleep.

Yours sincerely.

[FortranFan]

may.ka wrote:

@FortranFan

thanks for you wisdom, greatly appreciated!!! You are a real mentor!!! I'll try hard to get to a more novel style of posting to be more inclusive of your interest and to give you a pleasureful bed time reading and a nice sleep.

Yours sincerely.

@may.ka,

You're welcome but actually all thanks go to Intel team and Intel OSC staff for Intel Fortran product and related services; and the forum admin(s) too who support a great platform for all things Fortran in addition to your infantile sarcasm.

And yes, "nice sleep" is very important for everyone including Intel Fortran team and Intel OSC staff who too have to work with their own growing requirements of all kinds with every increasing complexity and entanglements while being constrained with limited resources to deliver products and services toward the Fortran "ecosystem" that are so precious for the technical and scientific and engineering computing domains.

[Anagnostou__Georgios]

It seems that the debugger does not work with fortran (see small program and xproj-file in the attachment).

The products I use are:

MacOS Mojave 10.14

Xcode 10.0

Intel Parallel Studio XE 2019.4.233

Is this a known issue?

Kind regards

George

[Ron_Green]

Georgios - you are using xcode 10.0 correct?

I am using 10.2.1 and am able to set and run to a breakpoint.  Are you able to do this?

what issue are you seeing?

I will see if I can find an older xcode 10.0 and try Update 4 install on it.  You did install the compiler after Xcode correct?

[Anagnostou__Georgios]

Dear Ronald

Thanks you very much for your message. 

Yes, at the the time of my question I was using Xcode  10.0. In the meantime I updated to Xcode 10.2.1 (and reinstalled the compiler) and am able, too,  to set and run to a breakpoint. This was possible also with Xcode 10.0. The problem is that I cannot inspect the variables. 

Kind regards

George

PS: Please note that in the meantime I asked for "official" support via ETH Zurich's IT department ([ID#18145729] Intel Parallel Studio XE 2019 - Support).