99/8 MESS & disk drive cards

[+mizapf]

1. Did you use single density or double density? The SD disks should be the compatible. If you say "99/2-formatted disk", do you mean that you did a FORMAT MEDIA on the disk? This way does not allow for choosing the density. I'll try to find out what happens in TIMT; could you probably provide a simple Howto to cause the corruption? You can also create disks in TIMT.

 

2. No, since I am lacking information about how this expansion is used or connected, and whether the 99/2 BASIC can really make use of it. Actually, some early docs state that the standard RAM size is 2K, which does not work at all.

[+hloberg]

I was using the .DSK you posted to this thread, attached.

I switched to using SD 90k disk and it seems to work now.

 

2nd, darn. Any way you can goose up the memory to at least 16k?

 

BLANK1-992.dsk

[+mizapf]

Could you send me (or post here) the corrupted disk image and tell me what operations caused it to became broken? I tried to copy the HELLO file with TIImageTool to the same disk (as HELLO1, HELLO2) which was still readable in the 99/2 emulation.

[+hloberg]

51 minutes ago, mizapf said:

Could you send me (or post here) the corrupted disk image and tell me what operations caused it to became broken? I tried to copy the HELLO file with TIImageTool to the same disk (as HELLO1, HELLO2) which was still readable in the 99/2 emulation.

It was using the larger .dsk format that corrupted. when I use the smaller SDSS format everything was OK. I'm sure that what is was. unfortunately I deleted the corrupt files.  Oh, one thing about the corruption is that it only seems to happen on larger files than HELLO. may be when it jumps sectors? That may be why you're not getting an issue. Anyway, like I said, if you use the small SDSS files there doesn't seem to be an issue. My bad.

One thing I do get problems with is if i try to import BASIC code using TIIT in 99/2 disk. I type in the programs in text then import the files using the TIIT. All kinds of weird things happen when I read the file into the 99/2 like most of the file is corrupt or such. The file itself doesn't seem to be corrupt on the disk though. So far i haven't gotten it to work. I guess there might be some slight differences between TIBASIC  and 99/2 BASIC?

And I plan to, maybe, demo the MAME 99/8 and 99/2 emulators in August at the Austin TI fest. So if you have any thoughts about the emulators that I could pass along it be appreciated.

 

Edited June 23, 2019 by hloberg

[+mizapf]

I tried to load some TI BASIC programs in the 99/2 but failed for most of them. First, they must not be longer than the free space; second, it seems as if the BASIC interpreter has some troubles processing the unknown commands or subprograms. Many subprograms are missing, like CHAR, SCREEN, COLOR etc.

[+hloberg]

thanks.

As for loading in TI BASIC into the 99/2 the files i created it didn't look to have anything not in the 99/2 command set but still had issues. I'll keep playing with it though, I may have missed something.

 

Like i said, I want to demo MAME TI99 compatibles for Austin TI Fest in August so I'll be reading up on it and expect some questions. People are interested in using MAME for the 99 (it's my emulator of choice) but seem to be intimidated by it for some reason. 

 

[+hloberg]

Ran some test and got a better idea what is happening with the 99/2 Mizapf.

What looks to be happening on the MAME 99/2 is that if you write more than 4 or 5 files of average program size to any type of .DSK ( SD, DD 40 track, 80 track tried them all) the files become messed up from that point on. the 1 though 3 or 4 files are fine but files 5 onward get weird problems like corrupt loads or locking up the emulator.

Doesn't matter if files where previously on the .DSK or not. It's the files that the 99/2 writes to .DSK.

oddly if you take the same .DSK the files load fine in the MESS TI-99/4a emulator. And in TIIT the files look fine.

attached are some files I tested with. The files I wrote with the 99/2 are named TEST(number). try loading TEST4 and beyond on the disk.

Maybe it's something I'm doing or setup as it's fairly consistent?

disk.zip

[+mizapf]

As far as I could find out by debugging, the 99/2 locks up while reading the Hexbus. It waits for the slave to release the line, which does not happen. This may be a bug, but I have to take a closer look. Since the 5102 works correctly with the 99/8, chances are that this is a problem within the Hexbus emulation inside the 99/2.

 

Edit: Looking into the source code, I remember how much guessing I had to do to get a working Hexbus interface in the 99/2. The problem is that TI put everything into a gate array, and I had to guess from the operating system code how it is supposed to behave.

Edited June 24, 2019 by mizapf

[+hloberg]

considering 99/2 was still an unreleased item and therefore maybe, maybe not yet finished I was surprised you got it working this good with the Hexbus. 

 

Note; ran another test of just 1 large file on a disk and so it's a file size issue. Since the test files were getting larger that's why the later ones was error-ing out. I see that now. Also explains what you just posted about the hexbus locking up. It must read in a chunk before locking up. Oh well.

 thanks for looking into this.

EDIT: I removed the hexbus from the startup batch and just using tape and the 99/2 is  now really FAST ! wow! mostly what i was made for anyway. 

 

Edited June 25, 2019 by hloberg

[+mizapf]

Got faster? Was your PC too slow for the emulation of both 99/2 and HX5102? What did it return as emulation speed (x%)? (You can see it when starting MAME on the command line.)

 

I'll eventually fix the Hexbus in the 99/2, but this may take some more time; definitely not a thing to be fixed in a couple of days. I'm quite sure that it can be done.

[+hloberg]

the 99/2 was fairly fast with the HX5102 setup, about 100%. but when I removed the HX5102 from the start batch it started peaking above 100%. Yes, I'm running this on a high end Atom based laptop (I like the small size) so it's not as fast as say an i5. But only the 99/8 with HX setup is slow on it though. Everything else is at speed.

If these speeds are realistic of what the real 99/2 was then it's a whole lot faster then the 99/4a console BASIC and a little faster then even Ex BASIC. Sad it never made it to market.

No rush on fixing the Hexbus, I'm using the tape just fine. I'm going to write a little game to demo for the Austin TI fest in August and tape will do for that. When I give my talk on the MAME emulation I'll just point out the 99/2 Hexbus implementation is still a work in progress.  Like I said, I'm amazed you got this much working. So far the HX was the only issue I run across on the 99/2 so kudos for the good work!