Vorticon

Create a blank disk image (I use TI99Dir) and set it up as DSK3. Start up XB with the compiler disk in DSK1 and one of the library files in DSK2. It will autoload Fortran 99. Press 1 for Edit from the menu then press 2 to start the editor. Paste the program source I posted earlier using Paste (not Paste XB to preserve formatting) and press Fctn-9 when done. Press 3 from the menu to save the source file to DSK3. Now press 7 to exit the edit menu, then go back into it again by pressing 1. Load the source file from disk by pressing 1 then press 2 to edit again. The file will be truncated and sometimes "rearranged". None of this is an issue when using a FIAD.

So I played with the Fortran 99 package today. In emulation under Classic 99, you have to use a FIAD disk as your source file disk. If you use a disk image instead, the source file you edit will get corrupted. I am using the latest Classic 99 version. Also Fortran 99 will not run with js99er at all. Not sure why. I was able to successfully edit, compile, link and run the sample program included in Beard's introduction, so I thought OK let me try something a little more interesting and went ahead a made a Fortran version of the simple PI approximation program we discussed in another thread. Unfortunately the compiled and linked program would not run. I have absolutely no idea where the issue is. When I compile, I get 4 warnings about using mixed variable types in computations which is fine, but no errors. The program uses the Fortran and Math libraries (for the IRAND function) and the linker does not throw any errors either and produces the proper executable files. I tried this in Classic 99 as well as real iron with similar results. Here's the program listing: anything looks out of order? C CALCULATING AN APPROXIMATION OF PI PROGRAM PICALCFT INTEGER *4 N, I, TOTAL REAL *8 X, Y, D, PI C GET NUMBER OF ITERATIONS 10 FORMAT('+', C12, M1.1, 'Number of iterations? ', I8) READ(6, 10) N C MAIN PROGRAM LOOP 20 FORMAT('+', M12.1, 'Iteration # ', I8) I = 1 DO WHILE(I .LE. N) X = IRAND(10) / 10 Y = IRAND(10) / 10 D = X**2 + Y**2 IF (D .LE. 1) THEN TOTAL = TOTAL + 1 ENDIF WRITE(6, 20) I I = I + 1 REPEAT C FINALIZE RESULT AND DISPLAY 30 FORMAT('+', M23.1, 'Approximate PI = ', F1.6) PI = TOTAL / N * 4 WRITE(6, 30) PI STOP END

Thanks for posting that. I'm going to be trying it out once I go through the manual. Here are the disk images for Fortran 9640 created from my physical disks. FORTCOMP.dsk FORTLIBR1.dsk FORTLIBR2.dsk

This is NOT hysteria. Please educate yourself for your sake and the sake of those around you. We are only at the tip of the iceberg unfortunately...

I will try to make a copy of the disks sometime this week and post them here.

It must have been quite a ride to experience computing from the early pioneering days all the way to the present age! When I took the Fortran course in high school circa 1982, I already was programming my TI in Basic which was my first exposure to computer programming. Compared to the looseness of Basic, Fortran seemed like being in jail! To make matters worse, the manual we were using, which I still have, was from 1978 and expected us to use punched cards for program entry... Kill me now... I passed that course by the skin of my teeth... Incidentally, it looks like Fortran 9640 seems to be based mostly on the 1977 standard, with some earlier elements retained as well. I am determined to get familiar again with Fortran however as I am intrigued by its mathematical capabilities

I wrote a chaos exploration program in assembly in 2010, which includes the logistic equation above (from the tigameshelf.net in Edutainment). Chaos Musings By Walid Maalouli In my constant strive to stress my TI 99/4A computer to the limit, I decided to explore the realms of Chaos. And no, this is not Warhammer, but a mathematical exploration of chaos visualization. This is a collection of 3 separate programs that create graphical representations of various chaotic equations, which also happen to be fractal. Be forewarned that it can take hours or even days of computations to come up with an image depending on the zoom level requested. The first program looks at the logistic equation, the second investigates the Henon equation, and the last explores Lyapunov space. Not for the faint of heart, but interesting nonetheless if you are into that sort of thing. Requirements: Editor/Assembler cartridge, 32K RAM, Disk Drive Executable to run: LOGISTIC, HENON or LYAPUNOV (E/A Option 3 - Program name is START) Chaos.dsk Chaos.pdf

Preditor is pretty much the only programming text editor I use on the TI. As I am reading through the Fortran 9640 manual, it's quite evident that the 99/4A version of the included editor is very primitive. And quite frankly, Fortran as a whole seemed pretty archaic in syntax, like the need to use a Format command in addition to a Write command to get anything on screen. And don't get me started with the DO loop structure: what a nightmare... Attached is a cartridge image for the 80-column version which works with Classic99 as well as a scan of the manual. PREDIT80-8.bin preditor.pdf

I'm planning to attend the International European Treff in Vienna this October since I'll be visiting my son in Germany around the same time. Hopefully Covid-19 will be a just a bad memory by then. If not, the world is in deep trouble... ☹

I dug up my Fortran 9640 package and gave it a quick cursory look. Yeah, it will take a little time to go through it. What Fortran version is this based on? This may not be a bad environment for my mathematical explorations on the TI after all... I also found a manual errata document which I scanned. Unfortunately, my printer's tray feed only scans single sided, so there are 2 PDF scans attached: one for the front pages and the second for the back pages, the latter starting at the last page. They will need to be printed out and assembled manually unless someone knows how to merge them digitally. Perhaps they might have an answer as to dhe's issues. Fortran 9640 Errata front pages.pdf Fortran 9640 Errata back pages.pdf Edit: the errata pdf is actually already included in the zip file in the Development Resources pinned thread, so you can ignore this post.

Some 4.2 billion iterations and we have the first 3 decimals of PI=3.141 using Guillaume's program! It took about 3 days to get this far under Classic99 in Overdrive mode and TI XB. I'm letting it run for a while longer and see if we can score a 4th decimal

Both my nephews are PhD candidates in aerospace engineering specializing in fluid dynamics at the University of Minnesota, and all their computational work for modeling is done in Fortran. It apparently is still the preferred language in this field! Who knew...

It never ceases to amaze me to see the ungodly amount of effort some people put into digitally emulating old computing devices with no real practical use anymore and with a very limited and slowly shrinking audience. Kudos.

I nearly failed Fortran in high school, and have since developed an aversion to it. That said, I should probably give it another chance. I do have the TI Fortran package, so if you post your source code I can try it out and see if I can duplicate your issue. Alternatively, try downloading the linker disk again. Sometimes the process corrupts the disks....

Here is the RXB versus XB RND function comparison using Guillaume's program for an N of 5000. Again expected values are mean= 4.5 and SD=2.87 RXB: XB: Not a huge difference for all practical purposes unless a very large set of random numbers are needed such as in the PI approximation program in which case the XB RND will be superior.

Try using Guillaume's version with a larger n, perhaps 5000, and see what you get.

Yes of course! Did not think of it that way. Well done.

Here you go, with the caveat that the array size is limited to 1500 in TI Basic: 10 CALL CLEAR 20 RANDOMIZE 30 OPTION BASE 1 40 DIM N(1500) 50 PRINT "PHASE 1:" 60 FOR I=1 TO 1500 70 N(I)=(RND*10)+1 80 T=T+N(I) 90 NEXT I 100 PRINT "PHASE 2:" 110 MEAN=T/1500 120 FOR I=1 TO 1500 130 T1=T1+(N(I)-MEAN)^2 140 PRINT I 150 NEXT I 160 SD=SQR(T1/1500) 170 PRINT "MEAN:";MEAN:"SD:";SD Using the same array size in XB gives: XB is still better, but not my much, although I do expect the gap to widen with a larger number of iterations.

Excellent! According to Dewdney's article, it took about 4 billion iterations to get the first 4 decimals correct! Keep it running

Impressive results! I went ahead and did a quick and dirty evaluation of the XB RND function generating 3000 numbers from 1-10, the maximum array size I could fit in XB. This is the ideal result: and this is the actual result: Really not bad at all for only 3000 data points.

Looks like I missed that: what HRD 4000+ sales?

So 1 million iterations in plain XB yielded 3.1434, clearly a very very slow conversion. The MLC version gives about the same answer at 1 million iterations. OK so now I'm going to do a trial run for 1 billion under Classic99 to see if I can gain another decimal. This will take a while (just kidding. This will take about 3 years to complete! I need a supercomputer cluster. On the other hand, with overdrive on in Classic 99, I might be able to bring it down to about 3.5 months. That's feasible).

Fascinating tidbit. Jim, the range and depth of your collector knowledge is impressive. Perhaps you should consider putting a compendium together with color pictures and all and sell it in bookstores. I'll be the first one in line to purchase it!

I let the MLC version run for a while, and it seems that the Pi value is getting stuck around the 1.427 value +/- 0.005 or so after over 119 million iterations. I wonder if this is a limitation of the RND algorithm. The 1 million iterations trial with the XB version is still running, and I'll be curious what kind of result I will get (should be complete within the next 48hrs).

Great... Now the next time a TI Invaders cart is up for sale, the seller will want $100 for it 😁 This auction was clearly a battle between collectors as I have never seen the Gamevision carts.