View previous topic :: View next topic |
Author |
Message |
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Fri Feb 08, 2013 2:26 pm Post subject: 64 bit ClearWin+ |
|
|
A beta version of 64 bit ClearWin+ is now available to users who have a Silverfrost support contract.
I will send you details upon receiving a message in my Forum message box. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
DanRRight
Joined: 10 Mar 2008 Posts: 2816 Location: South Pole, Antarctica
|
Posted: Wed Feb 13, 2013 6:58 pm Post subject: |
|
|
Paul,
I expect this is a step in right direction (and probably together with making FTN95 64bit-capable would be the greatest achievement of a decade+), but to inform us more can you please drop couple words what it specifically is, what are the differences with our current CWP, how it was implemented, is it Windows-only or also for Linux and Mac OS, does it go with Visual Clearwin, can you also give us examples how specifically to compile it together with 64-bit IVF, Absoft Fortran and GNU Fortran, how debugging process goes, what examples are included, do you have teaser website for it, how do you expect it to distribute, prices etc. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Wed Feb 13, 2013 8:29 pm Post subject: |
|
|
First of all there are no plans to make FTN95 64-bit capable in a direct way.
64 bit ClearWin+ is just the existing ClearWin+ ported to a 64 bit DLL and made available to third party 64 bit compilers. The visual result is identical and the code is the same with a few cosmetic changes. For example, the @ symbol is not accepted by most compilers so winio@ becomes winio$ and so on. The necessary changes have been kept to a bare minumum.
The main idea is that, users who need 64 bit executables will be able to use their existing FTN95 programs and compile them using a 64 bit compiler such as 64 bit gFortran.
The existing 32 bit ClearWin+ library has been extracted from salflibc.dll and ported using a 64 bit compiler (which happens to be Microsoft C++).
The result is a Windows DLL. Its use from Linux would be via Wine etc.
The new Plato has a basic facility to compile and link using gFortran and I have tested this myself. One user has worked successfully with NAGfor but that's as far as it goes at the moment.
As far as debugging is concerned, development can continue as at present using FTN95. A basic test suite is included in an existing gFortran download and will be part of a future release.
It is expected that the 64 bit ClearWin+ DLL will be included in the standard FTN95 release. That being the case, there will be no additional charge. Distribution will be free like salflibc.dll. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
DanRRight
Joined: 10 Mar 2008 Posts: 2816 Location: South Pole, Antarctica
|
Posted: Fri Feb 15, 2013 5:49 am Post subject: |
|
|
Though great that you have made CWP 64bit but it also a bit pity. Yes, now more users will polish it and demand for new features which is always good for both sides - developers and users. It also provides escape path to 64bits from FTN95.
But abandoning FTN95 to get 64bitness with other Fortran compilers would be sad day for many of us. 25 years ago when this compiler was the best no one would expect that.
(We all still guessing why no 64bits...Nobody will tell that the developers of this company lack needed experience for that. Even GNU fortran somehow managed without funding and much talent behind it to get there: but GNU is that bad that might sometimes run even missing enddo, endif or with truncated by 73 char variables! Still it is 64bit while FTN95 not)
Adding 64bit support to FTN95/CWP, possibility to run parallel applications like MPI or CUDA would extend the life of this compiler for at least next 30 years when later on you may add OO features and prolong it for 30 more!
Didn't you guys have tights with Salford and other universities where you might find both motivated industrial app designers and artists as well as talented young programmers willing to develop further this compiler which 25 years ago was a decade ahead of competition. And it is still ahead in many respects, though also motivated copycats succeeded not only slowly erode its advantages but even surpass in some areas (more optimization and hence faster code, support of parallelization, some F2K+ features, not even mentioning 64bits).
Again, Paul, thanks for your great effort. But I'm still scratching my head who is taking key decisions in your company? Hearing for decade that 64 bits are not coming and decently always do not believe in that. Always hope it is just the "no comments on unannounced products". But if it is true, i strongly feel it's insane - who kills the stuff which may work great for decades to come? |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
dpannhorst
Joined: 29 Aug 2005 Posts: 165 Location: Berlin, Germany
|
Posted: Fri Feb 15, 2013 8:30 am Post subject: |
|
|
Hi Paul,
the decision to plan no 64bit version for FTN95 is a shame and far away to to see the reality of the development in modern software.
Why should anyone buy FTN95, if you recommend to use third party compilers? Clearwin+ has been a good tool for fast design of GUIs in Fortran,
but you can do this also in other ways.
I fully agree with Dan, that you are loosing any leadership Silverfrost/Salford has had for many years.
Please, think once again to revise your (or your companies) decision.
Detlef |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Fri Feb 15, 2013 10:06 am Post subject: |
|
|
It is not for me to comment on our sanity or lack of it, but 64 bit ClearWin+ is being released as immediate response to requests for 64 bit functionality. Users who develop their applications using FTN95 will now have the option of creating 64 bit executables at the final release/production stage.
I am confident that this new feature will be a welcome addition for may of our users. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
LitusSaxonicum
Joined: 23 Aug 2005 Posts: 2388 Location: Yateley, Hants, UK
|
Posted: Sat Feb 16, 2013 8:44 pm Post subject: |
|
|
It's not fair to criticise Silverfrost for responding to a real user need with the simplest and easiest solution to the problem. Of course it is clear to them that it is a stopgap - but it does save people abandoning Clearwin ...
If FTN95 ever goes 64 bit the work on Clearwin+ won't be wasted.
Of course, if it were up to me, I would have the 64-bit Clearwin+ as a separate product available at appropriate cost and fund (as far as possible) the future development of a 64 bit FTN95 out of the proceeds. It does seem to me that a 64-bit FTN95 (insofar as Windows is 64-bit!) is a smaller next step than addressing the problems of multithreading or using the newer opcodes to speed up execution. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
DanRRight
Joined: 10 Mar 2008 Posts: 2816 Location: South Pole, Antarctica
|
Posted: Mon Feb 18, 2013 8:48 pm Post subject: |
|
|
By the way, current FTN95 finds some major errors in the user's code written with CWP. It checks consistency of types of arguments, their amount etc. Say, you added by mistake an extra %rb or forgot one more %rd, for example, and compiler hints you there. Despite it is not the brutal error extermination like with FTN95 true Fortran texts but that still helps a lot. You are not left alone to find what the heck. Will GNU or IVF also check for these user mistakes in CWP? |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
LitusSaxonicum
Joined: 23 Aug 2005 Posts: 2388 Location: Yateley, Hants, UK
|
Posted: Mon Feb 18, 2013 10:31 pm Post subject: |
|
|
Dan, I think the point is that the code has to work in 32-bit FTN95 first, before you transfer it to another compiler - i.e. do all development in the Silverfrost system. Eddie |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Mon Feb 18, 2013 10:34 pm Post subject: |
|
|
ClearWin+ error reporting occurs at run time from the DLL.
So on the face of it, error reporting will be unchanged.
However, we are assuming that development will take place using FTN95 and ClearWin+ error messages will be blocked for users who do not have an FTN95 licence. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
qt
Joined: 23 Aug 2005 Posts: 46 Location: Berlin, Germany
|
Posted: Fri May 31, 2013 11:32 am Post subject: Re: 64 bit ClearWin+ |
|
|
PaulLaidler wrote: | A beta version of 64 bit ClearWin+ is now available to users who have a Silverfrost support contract.
I will send you details upon receiving a message in my Forum message box. |
I guess, just a .dll won't be sufficient. Do you also supply a MODULE which declares INTERFACEs such that the different variants and uses of WINIO@ are correctly mapped by gFortran, for example? Or, is this not necessary? |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Fri May 31, 2013 2:52 pm Post subject: |
|
|
The beta download includes a 64bit dll together with source code for all related modules. For gFortran users there is also a dll containing the object code for the supplied gFortran modules.
There is a separate gFortran download containing the 64 bit gFortran compiler together with a set of ClearWin+ sample files.
Readme instructions are provided with both downloads. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
DietmarSiepmann
Joined: 03 Jun 2013 Posts: 279
|
Posted: Mon Jun 10, 2013 3:14 pm Post subject: |
|
|
Once that I have been able to compile a small Fortran test application with gFortran calling into the 64 bit ClearWin+ dll, I would like to ask if there are any plans to support other Fortran compilers, as well, especially the INTEL 64 bit compiler (ifort). |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Mon Jun 10, 2013 6:45 pm Post subject: |
|
|
In theory 64 bit ClearWin+ will run with any third party Fortran compiler.
The new DLL has been used with gFortran and NAG Fortran but not, to my knowledge, with Intel Fortran as yet.
I am not sure that we can guarantee to "Support" the DLL (in the normal sense) for any third party compiler. We expect it to work and so far it does.
Hopefully, someone will have the time to try it out with Intel and provide some feedback for other users. |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
DanRRight
Joined: 10 Mar 2008 Posts: 2816 Location: South Pole, Antarctica
|
Posted: Tue Jun 11, 2013 3:41 am Post subject: Re: 64 bit ClearWin+ |
|
|
PaulLaidler wrote: | A beta version of 64 bit ClearWin+ is now available to users who have a Silverfrost support contract. |
I think this needs a reference for more information. What specifically support contract is? |
|
Back to top |
|
![](templates/subSilver/images/spacer.gif) |
|