Loss of IFORT after installation of version 2025.0.1

[Stephen_Sutcliffe]

After getting an e-mail regarding the availablity of new Fortran compilers:

I can no longer see IFORT in the VS compilers. The e-mail (above) mentions BOTH Intel Fortran Compiler (IFX) AND Intel Fortran Compiler Classic for Windows so I expected both to be installed. I suspect this is not the case.

I tried repairing previous OneAPI Toolkit but this didn't help. We are recommended to do these updates but they always seem to cause users more grief. I was hoping IFX might finally compile my program without ICE's at last but until this is the case I need IFORT to maintain a usuable and reliable executable. 

Does anyone know a workround for this?

Thanks

[Devorah_H_Intel]

I have reported this issue with an email update mentioning ifort before, but unfortunately, the issue was not resolved. There is no ifort in any of the 2025 oneAPI releases. 

In addition, there are no changes to the ifx compiler in the 2025.0.1 minor update release. I hope this clarifies things. 

[Stephen_Sutcliffe]

It doesn't clarify anything and only adds to the confusion.

If there is no difference to previous version, why was it made available with recommendation to update?

The e-mail link is misleading as mentions Classic (IFORT) had it not mentioned this I probably would have not bothered and waited until next full One API version. 

So question remains, how do I reinstall IFORT? I do have Premier Support.

I've only tried this new install on my laptop which is more of a backup which I try the new IFX versions in the hope that 

[andrew_4619]

The previous ifort install should still be there and selectable.

[Stephen_Sutcliffe]

Thanks Andrew,

I can only see the latest versions 2025.0.1 on the download site. No IFORT or OneAPI 2024 or whichever was last to include IFORT.

I will continue searching..

[andrew_4619]

I meant in VS, the 2025 install does not remove previous ifort

[Stephen_Sutcliffe · Accepted solution]

Ah I see maybe it was my fault after all. I uninstalled the older version 2024.2 to create space after installing the latest compiler based upon the description Intel used in their e-mail stating both compilers. I had forgotten that was previously mentioned that 2024 was the last to include IFORT. I have found the 2024 installation files in my dropbox so will hopefully can resolve the issue.

Intel should take more care with their product descriptions to avoid such confusion.

Thanks for your time, much appreciated.

[Stephen_Sutcliffe]

Sorry got a bit trigger happy there, I'll try again.

It doesn't clarify anything and only adds to the confusion.

If there is no difference to previous version, why was it made available with recommendation to update?

The e-mail link is misleading as mentions Classic (IFORT) had it not mentioned this I probably would have not bothered and waited until next full One API version. 

If issue is unresolved the software update e-mail should mention this and warn users that IFORT will get wiped from the IDE Integration which is what I assume has happened. Intel really don't like IFORT do they and seem too keen to banish it into obscurity ASAP.

So question remains, how do I reinstall IFORT or resolve IDE integrations? I do have Premier Support but only tend to use it for reporting issues where I can send reproducers.

Luckily I've only tried this new install on my laptop which is more of a backup system where I try the new IFX versions in the hope that they will resolve the ICEs relating to Optimisations (which I reported over a year ago with reproducers) in Release configuration. I have to turn optimisations off in some files to avoid the ICE.

There is also an issue with IFX Debug linker which fails with a message relating a usually couple of projects (out of a total of 32) saying "DYNA3D.lib(d3d_mat21.obj) : fatal error LNK1220: 'resolving static reference symbol' requires '/WOWA64' specification". I've tried rebuilding the project(s) but this does not always work first time though sometimes it will eventually link after repeated rebuilds (no changes to any codeor compiler settings) which suggests to me a bug in linker somewhere. What are possible causes of this error? 

[Mark_Lewy]

I agree that the email link is causing confusion because oneAPI 2025.0.x no longer contains ifort.

To reinstall ifort, I'd uninstall oneAPI 2025.0 then install 2024.2 and then 2025.0.

Also, it's not as if you haven't had sufficient warning that ifort was going to disappear, ifx has been available since oneAPI 2021 and IMO has been good for production use since 2024.0. Yes, there are issues with it, but no probably more than in ifort when it was the only compiler.

[Stephen_Sutcliffe]

Thanks for the tip Mark. I did manage to successfully retrieve the IFORT by reinstalling 2024.2. It only affected one computer. 

I agree that we have had plenty of prior warnings about the deprecation of IFORT but the e-mail I referred to does indicate a lack of attention to detail on Intel's part, which is worrying and might explain why there have been so many issues still cropping up in IFX such that it remains unusable for my purposes (fact not opinion. 

I have experience of fortran compilers on Windows, over past 26 years, switching from Microsoft Powerstation to Digital then to Compaq (CVF), and finally to Intel (including introduction of 64 bit compilers and parallel processing)  and each change generally went without any problems. In fact most of those changes resulted in a noticable improvement be it faster compile & run times, better compiler diagnostic checks (compiler and runtime) and new and useful languages features.  The only negative issues I remember were the loss of a really neat Array Viewer feature and a spell where debugging (watching) of array variables of user defined data types did not work for a few years. It took a number of forum users putting in support tickets about this matter to finally get this resolved (almost anyway). At the time we were originally told it would be difficult if not impossible to fix as it was something within the integration with VS and that was controlled by Microsoft.

When I see each new release of IFX I always hope that it will finally deliver but so far that has not been the case.  It has only taken 4 years to get this far but I remain hopefully that, with help of all their army of unsung volunteer Beta-Testers, Intel finally come up with the goods. Maybe next time, fingers crossed.

Intel marked this thread as resolved even though there was no acknowlegement of their misstep with the e-mail and didn't attempt to answer any of my questions regarding the IFX compiler/linker issues mentioned in the entries in this post.