Excel files

[DMacPherson]

All I can do is speak for our own practices, but validation is really our "end game". Our software products (and consulting services) are about what you actually see in service. The theoretical is only relevant as it provides the underpinning to developing a prediction method. Put another way, I don't care what the resistance of a ship or boat "theoretically" might be according to some algorithm or code. I care about what is actually seen in service. So, I lean heavily on the empirical versus the purely numerical. If a prediction method is ill behaved or does not pass our validation studies, then it does not go into our software. How can you have comfort in a code that gives ten different answers when put in front of ten different engineers? Or takes so long to run that there is the inevitable temptation to simplify the problem to the point that the model is no longer valid?

Don

[troy2000]

My first computer was an Atari 400, with a dinky little monochrome green monitor. I wound up using it as a word processor for years, even after we had other computers.

And of course I played Asteroids on it for countless hours....

[yipster]

hi california troy2000, i designed promo material for the first ping-pong computers in palo alto
even before the time Bill G made dos right next door in menlo park, must have missed the man

love to have a peek into your computers at the mets DMacPherson

[cthippo]

286? BAH!

I started on an 8086 and a TI99-4A. We still have the Osborne Elite "luggable" with the 4" monochrome screen somewhere.

[marshmat]

Anyone else remember the IBM PCjr? Intel 8088 core (just different enough from the 8086 to piss off programmers), a whopping 128 kB of main memory, full 16-colour graphics, cartridge BASIC, etc. The thing was older than me.... but I've never come across a newer computer that was anywhere near as reliable as the PCjr.

Of course, even though we still have 5.25" drives kicking around here and there, it's darn near impossible to find anything that can decipher the file formats used back then. At least with Egyptian hieroglyphs, the data's human-readable. An archaeologist 1500 years from now will have a devil of a time. Once they figure out what a hard disk is.... then how to read the patterns of alternating magnetic polarity.... then how to figure out what this sequence represents.... then how file systems are structured.... then how to reverse engineer the JPEG algorithm.... the enlightened archaeologist will find, after years of work, a photo of a cat overlaid with a stupid caption.

(Are we getting off topic? )

[Leo Lazauskas]

I still have a slide rule. Not a plain linear one, but a fancy circular one made up of concentric rings.

Engineers are well-advised to be wary of fluid dynamics programs, even my own contributions. A famous CFD practitioner, Peter Bradshaw, said at a meeting of the AIAA that "CFD was an accident waiting to happen". That worries me in the light of the recent problems with Rolls-Royce engines.

These days, I have taken to the idea of programs that are no longer than 100 lines, run in less than 10 seconds, and that give an answer with 10 figure accuracy. I can't remember who came up with the idea, but it makes for codes that are easily comprehended, easily verified with exact solutions, and that don't bore you when they run.

Not much use to engineers, I admit, but I find it an excellent system for numerical mathematics.

I better toddle off and have my cocoa now, oldtimers.

[daiquiri]

Originally Posted by Leo Lazauskas

These days, I have taken to the idea of programs that are no longer than 100 lines, run in less than 10 seconds, and that give an answer with 10 figure accuracy. I can't remember who came up with the idea, but it makes for codes that are easily comprehended, easily verified with exact solutions, and that don't bore you when they run.

The Apollo Guidance Computer (AGC) had about 4 kb of RAM and around 70 kb of ROM. That was enough for Engineers (capital "E" intentional) of that era to make it run in multitasking mode and to safely guide the Apollo spaceship to the Moon and back to Earth again.

Today we need megabytes and gigabytes of memory and ultra-high speed processors just to make Windows (or whatever other OS) running - in a hope that that the bluescreen won't show up. At least not before we managed to save our data.

Somehow, in the time period from 60's to present days, engineers have lost the capital "E" and have started creating uncontrollable memory-eating software monsters.

Hope that at least you Leo will be able to find your way out of that insane way of doing things...

Cheers!

[Leo Lazauskas]

Originally Posted by daiquiri

The Apollo Guidance Computer (AGC) had about 4 kb of RAM and around 70 kb of ROM. That was enough for Engineers (capital "E" intentional) of that era to make it run in multitasking mode and to safely guide the Apollo spaceship to the Moon and back to Earth again.

I did a lot of very good work on a Commodore 64. It really forced you to be economical with memory and I felt like a real power user when I found that I could use the 2k of printer memory.

[NoEyeDeer]

Originally Posted by marshmat

An archaeologist 1500 years from now will have a devil of a time. Once they figure out what a hard disk is.... then how to read the patterns of alternating magnetic polarity.... then how to figure out what this sequence represents.... then how file systems are structured.... then how to reverse engineer the JPEG algorithm.... the enlightened archaeologist will find, after years of work, a photo of a cat overlaid with a stupid caption.

I can just see that happening. I quoted it to a friend who is a software developer and we had a good laugh over it. Thanks.

[MikeJohns]

Originally Posted by daiquiri

........

Somehow, in the time period from 60's to present days, engineers have lost the capital "E" and have started creating uncontrollable memory-eating software monsters.............

I've run the whole course with computers, my first was a programmable HP calculator.

Now thanks to poor, sloppy, memory hungry OS's and particularly gamers, we have the irony of really decent engineering workstations as a side effect of lazy programmers linking giant libraries of code, it's great really for us.

The really hungry engineering apps are usually very well coded. FEA models with medium meshes run in under 5 minutes, it's real computing power on your desk.

It sure beats Roarks and a slide rule.

As for CFD, it has some good specific uses but it's probably better to say it's indicative rather than accurately predictive. You really need to tie the models to real data.

[Leo Lazauskas]

Originally Posted by MikeJohns

Now thanks to poor, sloppy, memory hungry OS's and particularly gamers, we have the irony of really decent engineering workstations as a side effect of lazy programmers linking giant libraries of code, it's great really for us.

I've long believed that the gaming industry has driven many facets of the computer industry. (I treat most of my work as a sort of game -it's more fun to do it that way!) Unfortunately, modelling fluids is still very poor in computer games.

Originally Posted by MikeJohns

As for CFD, it has some good specific uses but it's probably better to say it's indicative rather than accurately predictive. You really need to tie the models to real data.

A very healthy trend is the emphasis on validation and verification introduced by (among others) Fred Stern at the Uni of Iowa and workers at INSEAN.

Hopefully some of the contributions to the Gothenburg 2010 Workshop will show us the current state of affairs.

[MikeJohns]

Originally Posted by Leo Lazauskas

I've long believed that the gaming industry has driven many facets of the computer industry. (I treat most of my work as a sort of game -it's more fun to do it that way!) Unfortunately, modelling fluids is still very poor in computer games.........
Hopefully some of the contributions to the Gothenburg 2010 Workshop will show us the current state of affairs.

3D CAD is just second nature to me now, I have two decent screens a fast graphics card and I can run a complex FE model and still use the CAD packages or come here to waste some time.


CFD will eventually come of age, it's quite good in pipes

[Replica]

Originally Posted by marshmat

Anyone else remember the IBM PCjr? Intel 8088 core (just different enough from the 8086 to piss off programmers), a whopping 128 kB of main memory, full 16-colour graphics, cartridge BASIC, etc. The thing was older than me.... but I've never come across a newer computer that was anywhere near as reliable as the PCjr.

Does it give my age away if I say I say I still have one. My wife wanted to throw it out a number of years ago, but, needless to say it is now safely ensconced in our ceiling (attic). Don't have the heart to throw it away.

What about the Sinclairs ZX-81 connected to a TV-set, tape cassette player for storage and an optional 16k Ram pack (if you bumped it, you lost all your work). Those were the days ;-)

[DGreenwood]

Gotcha all beat. 1970-71 I worked for RCA service company at Jet Propulsion Lab as a programmer and operator on the Mariner 9 Mars Orbiter project. While there we took delivery of a state of the art processor about the size of a large refrigerator laid on its side. A whopping, jaw dropping 15k!!! We thought surely science had reached its physical limits.We had IBM 360s and Univacs that took up entire air conditioned floors that had less capacity than blister packed toys you see while waiting in line at the grocery register.