Announcing Intel Visual Fortran 8.0

[Steven_L_Intel1]

Intel is announcing today the availability of Intel Visual Fortran 8.0 for Windows, the successor to Compaq Visual Fortran. As soon as I can find the press release on the Intel web site (probably later today), I'll link to it here. You can read more about the release at http://developer.intel.com/software/products/compilers/fwin/

Intel Fortran 8.0 combines the best technologies from Compaq Fortran and Intel Fortran to create a new line of compatible compilers for Windows and Linux that deliver outstanding performance on IA-32 and Itanium processor systems.

At this time, Intel Visual Fortran Standard Edition is available, with a US suggested retail price of $499. (Microsoft Visual C++.NET Standard or higher must be installed in order to use Intel Visual Fortran.) A Professional edition, with the new IMSL 5 library optimized for Intel processors, will be available in early 2004. Note that Array Visualizer, which had been available only in the Pro edition of CVF, is now in the Standard Edition of Intel Visual Fortran.

Users of Intel Fortran with current support licenses can download the 8.0 release from their Premier Support account. For a limited time (but I don't know the end date), CVF users can upgrade to Intel Visual Fortran 8.0 Standard for only $200. An upgrade from CVF Pro will be made available later (and if you're interested in that, I suggest you wait for it.)

On a related note: Many of you are aware that I've been handling the bulk of CVF support requests for the past couple of years. As of today, HP engineers will take over responding to CVF support requests sent to the CVF support address CVF users are still welcome in this forum, of course.

If you have questions, ask them here, I'll do my best to answer.

[Steven_L_Intel1]

Please do not call VNI aboutPro upgrades- this doesn't help. Just let us know here (or to me by e-mail) what your needs/wants are in the area of IMSL upgrades and we'll discuss it with VNI. We understand that IMSL is important to many of our users.

[durisinm]

I apologize, but I already had a reply call from VNI. I told him (I think his name was Rick)that I wanted an IVF8 standard to IVF8 pro upgrade for $650. That $650 plus the $200 I already paid earlier this year to upgrade from CVF pro to IF7 would be the same as the people who are going to wait until next year to pay $850 to upgrade directly from CVF pro to IVF8 pro.

I will stay out of this and let Intel and VNI work things out.

Mike D.

[billaustralia]

I have downloaded the User Manual for IVF8, but can find nothing on how to interface with Visual Basic, other than it is supported.

I need to know

) can I call a fortran dll from vb6?

) will the same calls work as in CVF6.6?

) is the CVF6.6 call back procedure available?

Thanks

[ahasan]

IVF 8 supports API32 calls, I believe. Is this the same as CVF 6.6, or has there been an updating of new calls for XP?

[Steven_L_Intel1]

Pretty much the same as for CVF6.6 for now.

[davidspurr]

Yet again I get to find out about a VF upgrade only by chance (again ex clf). I gather from one post in this forum that emails were sent out but I did not receive one. Possibly due to changing my email address? My address is correct in my profile for this forum, but my CVF registration is probably under my old (now defunct) email address. Where / how can I change that?

re: the IVF 8 upgrade. Not entirely clear to me what has changed, other than getting the Intel rather than CVF optimiser
I need some clarification:

- I take it that there has been very little change wrt fortran source code. Correct? (ie c. zero benefit to me there?)

- I take it that in addition to the US$200 u/g cost, I will now also have to purchase Visual C++ .NET ??? (ie. the true u/g cost is more like $300 min?). No benefit there either since I have never used C et al, nor are likely to.

(Or can I still use my existing Visual Studio installation with IVF? Does not read like it.)

- re existing Professional versions. It take it that my existing IMSL libraries will not run if I "upgrade" to IVF8 Std? Ie. if I want to retain this facility, then I HAVE to wait till the future release of IVF8 Pro version - which will cost US$850 (+ c $100 for V++.NET). Again many $$$'s for no net benefit relative to my current CVF?

- A LOT of talk about IA-32 / Itanium / Pentium / ... optimisations etc, but by the total lack of references, I take it that there will be little, no, or negative improvement on my Athlon system. ie. I presumably can at best count on zero gain in terms of execution speed (& quite likely a significant negative impact?).

I would be very interested in other users feed back on this issue.


(I do trust that the inclusion of only Intel products in the "Minimum Hardware Requirements" section is purely a marketing thing - I assume that the new compiler, and code compiled with it will run on my Athlon PC?)

At this point CVF is looking like it may be a dead end product for me (the 2nd one for me, having originally started out with Watcom). I am struggling to see how I will benefit by "upgrading" to IVF8, but I sure would be glad to hear otherwise. Otherwise my original decision to buy CVF than on seems to have been a considerable mistake.

David

[tonyjay68]

David.

As a happy CVF/DVF user I see no current reason to move to IFC8.0 (i'm currently evaluating IFC)

because.

1. It costs money to do so (however if an eventual migration to IFC is expected it may be sensible take the pre march offer - which is obviously what they want - and not actually use the compiler for a while)

2. 64 bit : I'm really interseted in this as our software needs large arrays > 2GB and I do not really understand what is happenning in the marketplace. As far as I can understand Itaniums are not readily available as PC workstations and perhaps we will actually go the way of AMD (and I do not know what compilers will support AMD). I am not sure I want to put my eggs in the itanium basket if our end users (and us) end up prefering AMD

3. IFC 8.0 is not compatiable (yet) with CVF - There are many bugs in the CVF to IFC project conversion (I have currently9 open bug reports filed from 4 weeks of evaluation). There does not seem to have been extensive testing of migrating CVF to IFC (expecially if any mixed language projects)

4. The compiler has bugs (reports submitted). SO what, all compilers have bugs, but this is a different compiler so you come across new (to you ) bugs.

5. I'm looking for a compiler / environment which inteligently recompiles modules so that if my interface has not changed, my files which use the interface (.mod) do not have to recompile. (IFC does not seem to have the dependancy bugs of the old visual studio, but it is still painful using modules)

6. I appear to have to use the new C++ compiler in the .net studio, which means repairing lots of c++ code (where the microsoft c++ compiler now conforms to ANSI standards where it did not use to !)

7. The new studio in .net takes time to learn (I have, after 4 weeks not yet learned how to do things which I used to in the old one - Ithink some of the old things are not possible )

Why not let other mugs (me ??) use IFC for a while, report lots of bugs and then everyone else can use the product when it is finally wearing long trousers.

Tony

[Steven_L_Intel1]

David (and others),

I'm sorry you weren't able to learn about the new Intel Visual Fortran directly - there is a holdup regarding permission to use the CVF registration database. I hope that this can be resolved soon.

We won't be doing support for VC6, and there is no way to do this manually.We are investigating ways of not requiring a separate VC++ purchase. I will note, though, that the combined cost of the VF upgrade and VC++ Standard is less than the CVF version upgrade price ($280 vs. $329).

Regarding non-Intel processors - we know of many who use the Intel compiler on Athlons, including AMD themselves for their SPEC submission. While some of the most advanced optimizations are supported for Intel processors only, I think that you will find that the Intel compiler is the best you can find for any IA-32 compatible processor. Try it and see - there's a 30-day free trial.

[durisinm]

Any news ona revisedprice for adding IMSL 5 to IVF8for those of us who previously upgraded from CVF Pro to IF7 in 2003 and received IVF8 Standard automatically as part of our one year of support?

By the way, is the official product name for the current Intel Fortran now Intel Visual Fortran 8 (IVF8) or Intel Fortran Compiler 8 (IFC8)? I've seen both mentioned in this forum.

Mike D.

[Steven_L_Intel1]

There is good news on the IMSL front, but I'm not yet free to share it. Stay tuned.

The official name of the Windows compiler is Intel Visual Fortran, but you will see Intel Fortran Compiler often used, especially when a general statement including the Linux compiler is made.

[bradlepc]

The Standard Edition of VC++ doesn't include an optimizer. If you're doing mixed language programming (C and Fortran), you care about performance,and you want to use Intel 8, you're gonna pay.

I'm hoping to see an option to at least be able to use Intel Fortran from the command line without requiring VC++.NET.

Pete

[Steven_L_Intel1]

I don't understand how the VC standard's lack of an optimizer is relevant to Fortran. Yes, as you say, if you're doing mixed-language applications, you'll have unoptimized C, but how would that be different if Intel Fortran was self-sufficient? It wouldn't include a C compiler.

If you want optimized C, buy an optimizing C compiler.

[eddie_breeveld]

We would also love to see a command line Intel Fortran that does not require any Visual C++. As I understand it, the problem is with the Microsoft linker and associated libraries.

Our main requirement is to reduce the number of installation discs, not just the cost.Ideally wewould provide our product, which needs a Fortran compiler, on just one disk. With the Intel compiler, we now need 3 - our product, the Intel compiler, and Microsoft Visual C++.

[Steven_L_Intel1]

Eddie,

We understand - we don't consider the current situation ideal either. Thanks for your comments.

[svani123]

Hi,

I am currently using CVF6.6, I have come to know that CVF development is going to be stopped in near future. Should I have to move all my source code to IVF8.0, In moving this source code, will there be any issues and risks related.

Could you please explain me the issues and risks and then clarify whether there will be problem or not.

Thanks.

[svani123]

Hi,

I am presently working on CVF6.6 Quickwin application which is a very big application with complex calculations and lot of graphics involved.

Now As CVF is stopping its development in future, there is need to move to IVF8.0. As of Now I am not sure if I need to change anything in the source Code so that It would be compatible to IVF8.0.

Could u please let me know if I need to make any new changes to the modules which I am using in CVF6.6 when moved to IVF8.0 or the same code would work in IVF8.0 also with out any changes.

Any comments on this would be much helpful to me.

sree

[Steven_L_Intel1]

You should be able to rebuild without changes.

[md25]

Hi,

What are the prospects for CXML to be supported by IVF 8.0. Do the old DLL still work ?

PS : I bought 9 additional licences of CVF 6.6 standard and 5 updates to CVF 6.6 Pro end of november 2003 because we could not wait any longer for IVF 8.0 to use the budget. Nobody told me about the one year service solution for IF 7 !!!

[Steven_L_Intel1]

No CXML. Intel has its Math Kernel Library which overlaps a lot of CXML and is optimized for Intel processors and multiprocessing.

[jkrob]

sree,

I too am working on a large Quickwin scientific application in CVF & decided to take the plunge & switch to IVF. Here are my lessons learned.

The biggest thing is concerning the SAVE statement. In both CVF & IVF documentation, it states that "...certain variables are given the SAVE attribute, or not, by default." One would think that would mean that both compilers should behave the same way. Well...CVF does follow that statement but IVF doesn't. I've had to go back through my code which worked in CVF without a SAVE and add a Save to make it work in IVF.

I had problems converting the CVF Project to a IVF Project. Everything seemed to be there but when I went to compile the first time, the Make routine in IVF complained about no routines to build. I just deleted the IVF "auto" project & rebuilt it from scratch to mirror the CVF project. I may not have followed all the steps in the conversion process.

If you are using the PEEKCHARQQ function, CVF needs "USE DFLIB" but IVF needs "USE IFCORE" to work.

The size of the executable is much larger in IVF. In CVF, my executable was about 1.2 Mb where in IVF it is 6.8 Mb (!). Looking at it with a hex editor, it is mainly all Zeros. I've compiled for the smallest executable size but it made no difference. It runs fine but...

That is all I've run into so far. Hope this helps. Good luck!

Jeff

[durisinm]

There is good news on the IMSL front, but I'm not yet free to share it. Stay tuned.

Steve, you posted that on January 8 in response to my question about information for users who already upgraded from CVF to IVF before March of last year. Can you add anything yet?

Mike D.