|
forums.silverfrost.com Welcome to the Silverfrost forums
|
View previous topic :: View next topic |
Author |
Message |
JohnHorspool
Joined: 26 Sep 2005 Posts: 270 Location: Gloucestershire UK
|
Posted: Thu Oct 15, 2009 1:53 pm Post subject: |
|
|
jjgermis
Quote: | Even though we have many of our (essential) software written in Fortran and still continues to improve and expand it, many colleagues still believe (and unfortunately spread it as well) that Fortran is outdated. |
In the world of finite element analysis, consider these well known solvers, Nastran, Abaqus, Lusas, Ansys, Algor, CalculiX...... they all have solvers written in Fortran. Perhaps this might help to change your work colleagues attitude! |
|
Back to top |
|
|
silverfrost Site Admin
Joined: 29 Nov 2006 Posts: 191 Location: Manchester
|
Posted: Fri Oct 16, 2009 1:17 am Post subject: |
|
|
> 2010 (besides the soccer worldcup in South Africa) Silverfrost ???
I don't think Silverfrost can be blamed or congratulated for the soccer world cup being in South Africa. |
|
Back to top |
|
|
DanRRight
Joined: 10 Mar 2008 Posts: 2815 Location: South Pole, Antarctica
|
Posted: Sun Oct 18, 2009 5:34 pm Post subject: Re: |
|
|
silverfrost wrote: | > 2010 (besides the soccer worldcup in South Africa) Silverfrost ???
I don't think Silverfrost can be blamed or congratulated for the soccer world cup being in South Africa. |
If we start congratulation & blame game...
Oh yea, Silverfrost/Salford for sure can be congratulated and blamed for many things....Just few of them: Salford can be congratulated for the first compiler which opened 32bit address space for PC Fortran users (specifically shocking was virtual common), while Silverfrost can be blamed for potentially being the last in 64 bits...
Silverfrost can be congratulated for being first in debugging capabilities and blamed for being last in code execution speed in Polyhedron examples (excluding may be couple cases where it is first)
Silverfrost can be one more time congratulated for superb debugging capabilities (it offers such super-hack like undefined variable check for example for decades) but again blamed for the most primitive Windows 3.1-ish look and sloppy gray windows of its debugger. People touching it will never believe this is a king of debugging and not a rat.
Silverfrost can be congratulated for hell a lot of such hacks but blamed for not supporting those damn primitive VAX extensions like <n>X for n spaces in format statement or not allowing X (for one space) requiring writing 1X instead.
Silverfrost can be congratulated for amazing compilation speed and blamed for the same poor execution code efficiency
Silverfrost can be congratulated for rich graphics capabilities and blamed for not being able to handle Matlab or Mathematica graphics
Silverfrost can be congratulated for supporting OpenGL, but blamed that the examples they provide don't have a smallest artistic touch. This is great selling point completely not addressed by Silverfrost
Silverfrost can be congratulated for multithreading cabilities and blamed for not supporting OpenMP or parallel linear algebra libraries of www.equation.com (and when asked to help by its developer, Silverfrost became the only Fortran developer which did not care)
Silverfrost can be congratulated for Clearwin in Windows and blamed for weak pushing it as a standard for Windows/Unix Fortran users where hyperactive in advertisement Winteracter eats its breakfast, lunch and dinner and due to that continuing growing further (The amazing example that Solverfrost does not care how it looks in the eyes of Fortran users is that Polyhedron for more then a decade writes here http://polyhedron.com/clearwin "Salford Clearwin+ is very concise, 'Hello World' requires only 4 lines of Fortran code..." while to be exact actually "Hello World" requires just the same 1 line of Fortran text and even the same amount symbols. Compare Code: | i=winio@('Hello World')
end
with
write(*,*)'Hello World'
end
| . I can not imagine this would be written about my own software)
Silverfrost can be congratulated for the ability to directly write C code in the Fortran source and blamed for not advertising it
Silverfrost can be congratulated for giving users ability to write also HTML code in the Fortran source and blamed that nobody even knows that
Silverfrost can be congratulated for being most feature-rich Fortran compiler and blamed for having most boring look and wrap because probably decided that advertisement is not important. I remember DB had video presentation for FTN95. Repeat this inviting similar guy like one from www.slapchop.com and sell FTN95 like hot cakes on your own stadium after buying back the football team from Abramovich. (Btw these folks from Slapchop are so damn crooky, they've easily made me for 63 bucks instead of 19.95!)
Silverfrost can be congratulated for making free Fortran for personal use which beats absolutely each and every other compilers including C in downloads per week and blamed for not being aggressive in advertisement and not smashing instead other Fortran companies with ste
Last edited by DanRRight on Sun Oct 25, 2009 7:16 pm; edited 10 times in total |
|
Back to top |
|
|
jjgermis
Joined: 21 Jun 2006 Posts: 404 Location: Nürnberg, Germany
|
Posted: Wed Oct 21, 2009 3:23 pm Post subject: |
|
|
Quote: | Silverfrost can be congratulated for giving users ability to write also HTML code in the Fortran source and blamed that nobody even knows that |
Quote: | Silverfrost can be congratulated for the .NET and blamed for 99.999999% of FTN95 users have no clue what the hell it can be used for...
|
The comments from Robert are indeed interesting and to I can assent myself to almost all of them. I certainly am one of the nobodies in the above two statements.
I assume that many of the blames can be traced back to the available capacities at Silverfrost. This still is one of the lacking information in the history. Over the years there must have been some increase in the number of employees |
|
Back to top |
|
|
jjgermis
Joined: 21 Jun 2006 Posts: 404 Location: Nürnberg, Germany
|
|
Back to top |
|
|
DanRRight
Joined: 10 Mar 2008 Posts: 2815 Location: South Pole, Antarctica
|
Posted: Sun Oct 25, 2009 4:35 pm Post subject: Re: |
|
|
JohnHorspool wrote: | jjgermis
Quote: | Even though we have many of our (essential) software written in Fortran and still continues to improve and expand it, many colleagues still believe (and unfortunately spread it as well) that Fortran is outdated. |
In the world of finite element analysis, consider these well known solvers, Nastran, Abaqus, Lusas, Ansys, Algor, CalculiX...... they all have solvers written in Fortran. Perhaps this might help to change your work colleagues attitude! |
...or in the modern world of microprocessor benchmarks like FP Component of SPEC CPU2006 which 50:50 consists of Fortran and C. No, do not worry, scientific and engineering applications will be mostly written in Fortran and there is already no tendency for Fortran to retreat further like it was one-two decades ago driven by hysteria of networking, computer graphics, server, Internet commerce, web development and universities computer science departments people.
Seems things go now exactly opposite. Fortran standards get it more modern. Fortran compiler developers add useful extensions like with all other languages almost eliminating your need for C (and guess which Fortran compiler is first here ). Fortran newsgroups are one of the most active on the net beating all others besides C. Silverfrost is literally surging beating downloads per week records -- even i did not believe my eyes several months ago seeing that FTN95 was beating all other C, C#, VB, VBnet, Pascals and long bunch of others in downloads on CNet. And it still is in first 3-5.
Of course there exist good reasons for that. For me major in using Fortran is its ultimate simplicity with the syntax almost as simple and natural as in calculator, the minimal source code size and hence highest development speed and hassle-free manageability,its unparalleled debugging capabilities and guaranteed high execution speed which you can fine-tune to the absolute limit of you like. All that is most important specifically for large codes.
And even more specifically for Salford/Sliverfrost, though this sounds a bit as an extreme, i do not believe to any large code results if the code was written not in this compiler. I can beat i will find so many bugs in other codes that can claim that almost for sure all previous published results of these codes are wrong . Lahey and probably recently Intel only approach Silverfrost in this matter. Last year i was adapting to FTN95 someone's large code developed during last 25 years which according to its authors was "working well" with different Unix and some PC compilers. The code has large users base and produces many megabytes of data per run. I gave up with it finding thousands of bugs and spurious shocking 10^4 difference on some specific data. It was literally impossible to fix the bugs without authors sitting near you. And this was mostly Fortran 77. I wouldn't even touch any large C code, only fire can fix its hidden bugs. So if somebody does not like this my extreme claim, they just calming themselves that the last bug in the code anyway is never found which might be true but here nothing we can do, higher then FTN95 there is no authority. |
|
Back to top |
|
|
jjgermis
Joined: 21 Jun 2006 Posts: 404 Location: Nürnberg, Germany
|
Posted: Wed Nov 04, 2009 11:28 am Post subject: |
|
|
Until now there much more (and valuable) information that goes beyond the history. However, should one call the FTN compiler a Salford or a Silverfrost compiler? What else does Silverfrost do besides the FTN compiler. When I go to www.silverfrost.com I find information on FTN95 and Solar Kingdom. Yes, the "About us" tab gives a little bit information which then set a link to Salford. Does this means that Salford "oder" the FTN95 compiler from Silverfrost? |
|
Back to top |
|
|
jjgermis
Joined: 21 Jun 2006 Posts: 404 Location: Nürnberg, Germany
|
Posted: Sun Dec 13, 2009 4:36 pm Post subject: |
|
|
Quote: | I would take me half an hour to dig out an acurate history and I don't see how it will help you. |
Paul, a few weeks ago this was your reaction on this topic. In the meanwhile I have discussed many of the ideas in the forum with colleagues - very interesting!
I am amazed by the number of views. What is your opinion on this - do you still believe it is not helpful?
Our history teacher tought us: If you would like to plan for the future, you should know the past. |
|
Back to top |
|
|
PaulLaidler Site Admin
Joined: 21 Feb 2005 Posts: 7924 Location: Salford, UK
|
Posted: Sun Dec 13, 2009 6:56 pm Post subject: |
|
|
I am happy to be proved wrong. |
|
Back to top |
|
|
jjgermis
Joined: 21 Jun 2006 Posts: 404 Location: Nürnberg, Germany
|
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Mon Mar 16, 2015 8:10 pm Post subject: |
|
|
Well, I enjoyed reading that !
Came across this old post by accident and
I even can connect to it .... as ... I learned my Fortran at the University of Manchester Regional Computing Centre (which was mentioned above) way back in 1977 when I started at the University of Manchester (Maths tower next door). Is the computing centre still there, I saw only very recently they had pulled down the Maths tower ... sad ....
I also chuckled at the comment about 4Mb of memory being considered as a radical upgrade !
Linking to up-to-date events, it was interesting reading this in John Horspools comment above ....
"I don't believe the Polyhedron benchmarks either, on a 64bit machine a 32bit silverfrost compiled program easily out performs a full 64bit gfortran version on a numerically intensive task."
... to me a layman as far as concerns performance of these things go, does this mean that the activities concerning the 64-bit version are effectively purely 'strategic' then (a sort of 'if they do it then we've got to do it' approach,and won't bring any real performance benefits ?
Seems to me like development of 64bit is a bit of a white effalump , at least for the vast majority of users, after all here we are 20 years down the road from when I first used a PC for work porpoises (yes it's only 20 years most of us have been single digit tapping our lives away) , and I keep asking myself the question, I was running FORTRAN with the luxury of with 32Mb memory on a mainframe at that point, and my problems haven't grown that much up to today, as I imagine most peoples haven't, and yet I'm looking at a laptop with 4Gb memory ... 4000 / 32 .... lets call it a 1000 mor epowerful machine (and no tapes to be mounted ) .... is there any need for it all ?
... and then there's that interesting comment at the end about the .NET development of Simpleplot ! , especially bearing in mind the discussions over the last few months concerning the difficulties of working with an 'un--openable', hence undebugable code... now another white effalump ? ... over to you on that one Dan :O) !!! |
|
Back to top |
|
|
JohnCampbell
Joined: 16 Feb 2006 Posts: 2554 Location: Sydney
|
Posted: Tue Mar 17, 2015 1:57 am Post subject: |
|
|
John,
Thanks for bringing this thread to light.
To comment on "I don't believe the Polyhedron benchmarks either, on a 64bit machine a 32bit silverfrost compiled program easily out performs a full 64bit gfortran version on a numerically intensive task."
I have not found the 64-bit .exe perform any slower than their 32-bit equivalent. There are a number of aspects of 64-bit that usually make them faster.
FTN95 performance in Polyhedron benchmarks is an interesting topic. I have actually looked into a few of the tests and used /timing to find where FTN95 is going slow. Typically the cause is bad coding approaches in the tests that FTN95 does not optimise. Things like x**2.0d0; I just changed it to x**2 or x*x and the run times reduced substantially. Poor memory management and Stack usage is another common problem.
I suspect that other compilers perform much better as they have included optimisation that targets the specific bad coding identified in the tests. Not much good for those who do not use these bad coding structures.
As a user of FTN77 on the Pr1me, I have benefited from the many good features of the Salford/Silverfrost Fortran over this period.
John |
|
Back to top |
|
|
DanRRight
Joined: 10 Mar 2008 Posts: 2815 Location: South Pole, Antarctica
|
Posted: Tue Mar 17, 2015 3:38 am Post subject: |
|
|
John-Silver,
Yes, over time this forum gathered a lot of interesting stuff.
JohnCampbell,
Great that you are looking at this and already found some shortcuts. Finally someone has been doing that with FTN95, because first time we discussed these tests in comp.lang.fortran almost two decades ago. Salford Fortran was in the state of switch to Fortran 90 at that time and cleaning just the bugs took more then a decade so the optimization was not a first priority. Right now probably it's good time for that.
Have you tried one test on Polyhedron which gave whopping, hairs-rising 30x slowdown? Was this due to a lot of underflows? Or another John (Polyhedron's John Applegate) made something somewhere with the compiler settings? |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|