OT: Nasa runs competition to help make old Fortran code faster
Responding: Thanks for the responses. In general, if someone sees some way THEY could do such an exhibit, based on their interests in hardware, software, and to represent some use or users - that's great! More than one person can do such things, as far as I'm concerned. I'll respond with what I'd do, what I had in mind, for my interests. - Herb ---------------- [Jim Scheef:] For this to interest more than just a few people the physical system that is modeled must be something familiar to most people...like Lunar Lander... Is that the type of thing you had in mind? That's not a bad question. No, Jim, I do NOT have in mind something that's "familiar to most people". It's helpful but not a goal. That's because, of 40+ years of personal computing history, "most people" by sheer numbers are familiar with personal computers used as ordinary devices on the ordinary Internet, social media, gaming, Twitter; or previously, to replace paper-tasks of job or business or hobby. Computers used to be extraordinary, and unfamiliar. Those people outnumber the scientists and engineers and other technical-trained people (and their fans) that I'm talking about. I described a very specific set of circumstances. But I'd not do such an exhibit, if there was zero interest. I'm seeing some interest today. Now - for say a continuing exhibit at the VCF Museum, you may well want something that's at least known-of - say, the half-century event of landing humans on the moon. And "lunar lander" programs are about as old as BASIC, if not FORTRAN. And, the landing problem IS a problem in physics; it's not hard to visualize. Of course it's been a game available on many platforms, so it's true history. And the physics and programming may be approachable to the unfamiliar (more on that shortly). also: Lunar Lander may be reasonable for an *analog* computer - I mentioned that hardware. A challenge would be the inputs and outputs, how to make them more than dials and graphs. For instance, the analog outputs could drive a modern visualization on a modern computer. I think this was discussed when analog computer restoration was discussed. -------------------- [John Heritage:] there's actually two pieces to the NASA competition. modify source code to improve performance....and an overall software / hardware solution...[discussion of modern hardware follows] A strong coder and a math guru are part of what's needed.. experienced computer architects, and people good at logic... Of course, I know little about the NASA competition, and little about modern parallel-computing hardware. But, as a digital BSEE engineer with some master's work in computer science, and as someone who hangs out at the Princeton Plasma Physics Lab; I do have a clue about issues of numeric calculation of complex physical phenomena. I know there's some personal-computing hardware around that supports parallel computation. I humbly suggest you add "physicist" and "numeric computational experience" to your list of needed skills. The changes NASA wants, have to produce better results, not just faster ones. But further discussion here, would be outside "vintage computing". I'll end by saying, that limits of hardware and software were also challenges in the era of vintage computing. Something can be learned today, by reviewing solutions of the era. In fact, that's a principle thesis of my vintage computing activities. -------------------- [Tony Bogan:] Herb, I would love to see something like what you describe at VCF next year! I have zero background in fortran...the first thing I thought of was accelerators [for Apple II? Mac? computers]. someone else would need to do the coding.. If using my hardware is feasible, I would be very interested in helping any way I can. Tony, thanks for your encouragement. I have no shortage of vintage computing hardware, thank you. And Apple II's are not my area of primary interest in vintage computing. So I'd not use them in my own exhibit. And, I'd not likely use modern accelerators, those too are not a part of my interests. But there are people today, who ARE interested in those very things. So, you might consider such an exhibit yourself, and find an Apple II programmer who'd support that effort. Or of course, learn some coding and some physics - there were books on Apple II programming for physics and engineering! - and see where that goes. "Lunar lander" isn't the worst place to start. -------------- [Kelly Leavitt:] I have my senior project from [Rutgers?] (1993) . It was a simulation of random particulate transport in groundwater written in FORTRAN. I moved this chunk of FORTRAN IV from the mainframe/mini era to the PC....It ran on my Tandy 1000A with an 8087. The vector mathematics that FORTRAN could do was amazing. ...I had offered to port the FORTRAN code to C++ at the time. The professor just laughed... He had me port this to MS FORTRAN 77 and put a C++ front end on for graphing the results. Aha! Proof! Such things WERE done, even after the 1970's and 80's. And with personal computers. Thanks, Kelly! I'd say a Tandy 1000 with 8087 8088 is "vintage" enough. And the code was even older. Kelly, see if you can resurrect your work and show how it was done to these young'uns! A decent all-in-one machine from the era I'd consider using, is a Heath/Zenith Z-120 series system - 8088 plus an 8085, same MS-DOS (or CP/M), probably similar performance. I have FORTRAN for that, a Zenith/ Microsoft product. I use a Z-120 about once a week, and on many vintage projects. The Z-120's were also bought by the US military for their classrooms, and by many universities; some sold one to EVERY student. Many people over 50 remember them; that's helpful. I could probably add an 8087...768K of memory ought to be enough....vroom vroom!..he he Herb -- Herbert R. Johnson, New Jersey in the USA http://www.retrotechnology.com OR .net
Tony, thanks for your encouragement. I have no shortage of vintage computing hardware, thank you. And Apple II's are not my area of primary interest in vintage computing. So I'd not use them in my own exhibit. And, I'd not likely use modern accelerators, those too are not a part of my interests.
But there are people today, who ARE interested in those very things. So, you might consider such an exhibit yourself, and find an Apple II programmer who'd support that effort. Or of course, learn some coding and some physics - there were books on Apple II programming for physics and engineering! - and see where that goes. "Lunar lander" isn't the worst place to start.
Herb, thanks for the response! My offer of hardware and help was as much for anyone else interested as you as I didn't actually think you'd be interested in an apple display but you never know! :-) Apple micros need love too! Seriously though, For anyone else interested, my offer still stands. As I stated in my previous email the accelerators I have are not modern relative to my hardware, they are from the same period as the Apple II so period correct and appropriate for a display as discussed. Might help move things along for a display for the public....not the Pleiades super computer speed but still ;-) Unfortunately my work and life do not afford me the time to learn basic physics modeling and fortran by VCF next year (too many life things with wife and family happening over the next 6-9 months specifically) nor is My interest in acquiring those skills very high to be in with. But if there are any of you out there with the programming skills, I think Herbs idea or a variation on that theme would be something worth displaying next year. I'm glad that article has sparked some discussion, I figured this group was the place to link to that article! Tony
Since the message below was sent with the "wrong" subject line, I'm copying it with the "right" subject line, so it can be found in the subject thread. I'm pleased to get several responses to the notion of running some kind of physics simulation in FORTRAN, *or* on an analog computer; and some kind of associated "demonstrator" to visualize. That's good! Speaking for myself, my primary interest is to represent 1970's small computing, the microprocessors of the era. But I also want to support the efforts in progress, to repair and restore and operate, analog computers under similar conditions. and I'm not adverse to restoring minicomputers, I have a few myself... I'm not immediately inclined, to engage a bunch of folks in some kind of massive project to these ends. Especially, to operate any more-serious "iron" such as larger minicomputers (PDP-11, small VAX, etc.) - better folks than I are working on those. But the interesting thing about this idea of simulating physical systems, is that MANY forms of "iron" were used to achieve simulations. So my reaction to these encouragements - beyond getting encouraged myself! - is that those interested, can resurrect or poke-at their own sorts of simulations. Show us, and themselves, how to "get 'er done". When they get results - there's some pieces of the puzzle in operation - then there will be skills and tools available, from which some kind of larger project or bigger task can be considered. I'm not in charge, but that's how I see such work. You build the tools you need, then build what you can from them. The skills come along with making the tools work. The tools work YOU! ;) Then you have the basis to plan something more comprehensive. Kelly can do some show-and-tell or Web pages, on the work he did in the era. Tony can work with the Apple II accelerators, see if they can solve some simple physics problems from the Apple books - a lot of computers had accelerators, using them was a challenge! Our analog-computer friends including Bill below, are getting those machines in operation. They will need to solve some simple problems, to produce inputs and outputs that can be "computed" by external logic or display or computers. and mainframes and mini's, well they have to run SOMETHING, right? May as well be FORTRAN! and any floating-point boards they have... and if anyone else is inspired to deal with "real world" data, inputs or outputs or processes, more power to them too. We'll have a vocabulary of experience to speak from together. I'll start poking at some old FORTRAN compilers for some of my "little iron" (bigger than a cell phone, smaller than a PDP-11). And I can manage some of the physics needed - but not all, there's a whole world of numeric programming that people spend a lifetime upon. And I have a 3600 line physics program, I can see if that can be run on a simple computer. But this is what we did in the era, as engineers and techs, to produce the world, that made the world what it is today. It happens I have a modest example, already punched up: http://www.retrotechnology.com/memship/camp/1802_senior_prj.html A modest project, maybe: but see where this person went with his degree! Mighty oaks, from little acorns grow. Herb
SVCatITC at aol.com SVCatITC at aol.com Sat May 6 00:05:56 EDT 2017
BRAVO Herb;
You are another scholar who again brought back memories for me.
When EAI got too big for their britches, they got involved in full scope power plant simulators using SEL 32 mainframes and CDC Winchester drives. FORTRAN was the OS of the day on these mainframes modeling nuclear reactor vessels, and processes.
I have a very good PHD friend living out in Hightstown, NJ who still uses FORTRAN from time to time in his modeling work. We worked together on the ANPP project out in AZ for AZ Public Service (back in the late 80's). I can drop him a line if he might be interested in helping in such a project.
I helped him in showing him real life SCADA instrumentation dynamics so as he could get the software objects to respond and drive the instrumentation as fast as the real life analog processes did. ..... Not such an easy task.
Count me as one who would want to see something similar run on big iron. I might be able to help with the peripheral instrumentation as well.
If you are coming this weekend, we might talk further.
Many Thanks for your continued "long view" insight for the mission statement.
Bill Inderrieden
-- Herbert R. Johnson, New Jersey in the USA http://www.retrotechnology.com OR .net preserve, recover, restore 1970's computing email: hjohnson AT retrotechnology DOT com or try later herbjohnson AT retrotechnology DOT info
From: vcf-midatlantic [mailto:vcf-midatlantic-bounces@lists.vintagecomputerfederation.org] On Behalf Of Herb Johnson via vcf-midatlantic Sent: Friday, May 05, 2017 5:32 PM -------------- [Kelly Leavitt:] I have my senior project from [Rutgers?] (1993) . It was a simulation of random particulate transport in groundwater written in FORTRAN. I moved this chunk of FORTRAN IV from the mainframe/mini era to the PC....It ran on my Tandy 1000A with an 8087. The vector mathematics that FORTRAN could do was amazing. ...I had offered to port the FORTRAN code to C++ at the time. The professor just laughed... He had me port this to MS FORTRAN 77 and put a C++ front end on for graphing the results. {Herb Johnson] Aha! Proof! Such things WERE done, even after the 1970's and 80's. And with personal computers. Thanks, Kelly! I'd say a Tandy 1000 with 8087 8088 is "vintage" enough. And the code was even older. Kelly, see if you can resurrect your work and show how it was done to these young'uns! ------ And my reply to Herb: I can attest that this was done. I'll look around. I saw the printed report just the other day. Quite the tome when you included the original material as an appendix. And yes, Rutgers University, College of Engineering, class of 1993. I believe, either as a second part of this project, or as an independent study, I compared the results to the 3D model presented by modflow92 (see http://gwmodel.blogspot.com/ and https://pubs.usgs.gov/fs/FS-121-97/). I was quite the groundwater modeling nerd for a few years. I look at this math now and am a little depressed in what I've forgotten. Kelly
participants (3)
-
Herb Johnson -
Kelly Leavitt -
Tony Bogan