TI-994a How-To

rhester72 · Unread post by **rhester72** » Sat Jul 31, 2021 7:08 pm

@Flandango - Nope, that' s not the case. Steps to reproduce:

- Ensure boot.rom (as a 99/4A full ROM) is in /media/fat/games/ti-99_4a
- Power off MiSTer
- Power on and select the TI core
- Note that you get a black screen instead of the the expected 99/4A start screen

For the weird part (keep going)...

- Now hand-load boot.rom as full ROM - note it comes up as expected
- Reload the TI core, and behold! The 99/4A appears

This last part is because whatever memory space in SDRAM you're loading *whatever* ROM into persists across core (re)loads, so if you last had a full ROM of, say, Parsec, that's what will be there on core reload, so boot.rom is being flat-out ignored at all times.

Flandango · Unread post by **Flandango** » Sat Jul 31, 2021 7:53 pm

Ah...forgot that SDRAM only gets wiped when Main is loaded.
Ok I made a quick fix and it should be working now.
If you are not using a boot.rom file, you will still have the previous rom in SDRAM after a core change since the core itself doesn't wipe the SDRAM unless you do a reboot.

arcadeshopper · Thu Aug 12, 2021 6:56 pm

I went to load a new game yesterday that has a - in the filename, on my keyboard it looks like the -_ key is mapped to =+ and also the =+ is that and also the \| is that.. so all 3 keys map to =+ .. makes it very difficult to type a -

Greg

Flandango · Unread post by **Flandango** » Fri Aug 13, 2021 5:32 pm

The forward slash (/) and minus key (-) is on the left bracket key ([). It's were it's physically located on a real TI-994a keyboard.

akeley · Unread post by **akeley** » Fri Aug 13, 2021 8:16 pm

I've tried the latest version and the whole screen is nicely filled again, the top/bottom bars gone. Great stuff.

tmop · Unread post by **tmop** » Thu Aug 19, 2021 11:27 am

Flandango wrote: ↑Sat Jul 31, 2021 7:53 pm Ah...forgot that SDRAM only gets wiped when Main is loaded.
Ok I made a quick fix and it should be working now.
If you are not using a boot.rom file, you will still have the previous rom in SDRAM after a core change since the core itself doesn't wipe the SDRAM unless you do a reboot.

@Flandango,

I was finally able to test your last update of the core and it's simply great!

Now it boots perfectly into the title screen. The support for large roms is working fine. Now these games are all working fine:
- Alex Kidd On Miracle Island;
- Flying_Shark V1.2.1;
- Knight Lore;
- Rasmus 8in1 Cart;
- Warzone 2;
- AtariSoft 14 games compilation (256KB and 512KB versions).

These SSS are now working:
- Console Calc / TI Planner
- MiniWriter 1.6
- Homework Helper

and also the "Disk Catalog", "Disk Copy" and the "Disk 2 Cassette" from RAG Software are not crashing anymore.

The MiniMemory seems to work perfectly!

I've tested some MM games that were tranferted into disk:

- Light Race (uses assembler calls from TI Basic);
- 9900 Break-Thru! Loaded using Option 1 menu ("Load and Run");
- D-Station. Loaded using MM2DSK util ("Cartridge RAM Loader/Saver)";
- Magic Places. Loaded using MM2DSK util ("Cartridge RAM Loader/Saver). Run from Easy Bug.
- Sneaky. Loaded using Option 1 menu. Run using the "run" menu.

I've tried to save/load the NVRAM file with these games and it works without any problem.

Of course, since there is no tape support, it was not possible to test the specific tape features of the module.

Just one note: the "Mini Memory Plus 2.0" (1986, by Sebasoft) is not able to load the NVRAM file, probably because it uses an 8K RAM.
Another minor thing, in the Cart Type menu there is a typo error for "paged 378", that is only "paged 7".

I'm really impatient to see the next improvements for this core (and the fix for the NTSC setting), however I think it really now deserve to be promoted in the official repository...

Flandango · Unread post by **Flandango** » Thu Aug 19, 2021 1:26 pm

Thanks Tmop.
The menu option for Paged 378 is call UberGrom since I've seen that referenced more often than Paged 378.
If folks are more familiar with Paged378 vise UberGrom I can change that.
So far, it should support Normal, MBX, Paged 7, Paged 378 (UberGrom), Paged 379 and MiniMem.

I'll look into the Mini Memory Plus 2.0 and see what the differences are other than 4k-8k mem.

tmop · Unread post by **tmop** » Sat Aug 21, 2021 2:52 pm

@Flandango,

it seems that the Atari games carts support is broken in this release if loaded as a full rom image. E.g. Donkey Kong was working fine if loaded as C and D parts, but not as a single file. It was working previously. Could you please check?

rhester72 · Unread post by **rhester72** » Sat Aug 21, 2021 4:31 pm

Confirmed re: Donkey Kong...that's odd.

Flandango · Unread post by **Flandango** » Sat Aug 21, 2021 7:56 pm

Hmmm....weird. I'll look into it.

Flandango · Unread post by **Flandango** » Mon Aug 23, 2021 12:38 am

Think I got the issue with the Atarisoft Full Image roms not working.
This version also utilizes the Caps Lock LED on the keyboard to indicate if the Alpha Lock is on/off and the Scroll Lock LED for Floppy Drive Activity, for those who don't have an IO Card with the fancy large LEDs.
Also an option to have the Alpha Lock key On when starting the core. I didn't include that ability for soft resets or resets from OSD to turn on the Alpha key in the event folks didn't want it changing on them when they reset it. I could also add that capability to the OSD Reset if folks care for it.

As for Mini Memory Plus 2.0, I couldn't find any information on it. If anybody can point me in the right direction, I would appreciate it.
Also I am still trying to figure out the issue that is causing the PALTEST to report NTSC timings as PAL. I can get exactly 60hz and it will report PAL, If speed it up to about 65-70hz, it will come back as NTSC but it will most likely cause issues with folks using VGA monitors.

As always, please let me know what I broke now.

rhester72 · Unread post by **rhester72** » Mon Aug 23, 2021 2:33 pm

@Flandango - Again, thank you SO much!

If you don't mind my asking, just because I find this so fascinating and I'm trying to learn, what did you fix/tweak to get Atarisoft going again?

Flandango · Unread post by **Flandango** » Mon Aug 23, 2021 3:03 pm

Long story short, I had to tweak a rom/memory mask that is used to adjust the TI's CPU memory address and the SDRAM's address for the various locations of each chunk of rom. When a full rom in loaded, it has to be broken up into various memory locations for each part (C/D/G/System Grom/Speechrom...etc).

tmop · Unread post by **tmop** » Mon Aug 23, 2021 7:41 pm

@Flandando

Tested (all working):
4A Flyer, Buck Rogers, Computer War, Congo Bongo, Jungle Hunt, Donkey Kong, Mancala, Micro Tennis, Moon Patrol, Ms. Pac Man, Pac Man, Picnic Paranoia, Protector II, Return to Pirate's Isle, Shamus, Super Storm, Tutankham, Parsec, Munch Man.

TI Calc, Terminal Emulator II, Extended Basic v2.5

Tested with MBX cart type (all working):
Big Foot, Honey Hunt, Meteor Belt, Sewermania, Sound Track Trolley, Space Bandits, Superfly.

All in "Full image", all is working fine!

I've checked on whtech for the Mini Memory Plus, but no docs, so I've asked on AA. Hope to have some answers.

tmop · Unread post by **tmop** » Tue Aug 24, 2021 9:45 am

@Flandango

I've done some other tests, expecially on disk, since one guy on AA reported that is not working. I was able to run fine some of the previosly tested disks (Argroids, Bandit, Diablo, Old Dark Caves) and all are working fine. Then I've done some tests with carts (changing the cart type on the fly, etc.) and reloading again XB to autoboot a game and it was blocked on the boot screen. After resetting the core all was working again.
Then I've tested some new disks and it seems that all is fine (except for one disk, but I need to check on the emulator to see if it's working fine).

The use of the led is really an interesting solution for providing a feedback on disk activity.

Regarding the PAL/NTSC I've checked with some music (taken from this new thread on AA: https://atariage.com/forums/topic/32418 ... ted-music/). These songs are taken from NES version of the Smurf game and converted using VGMComp2 with a 60HZ conversion, so on a PAL machine they should be slower. I've verified on Classic99 and the difference is audible, so I've checked on the MiSTer core and, yes, it's also audible also on it! I've done a similar check also on Freddy game, that has a background music, and I can perceive the speed difference. I'm not able to perceive the difference from a video point of view, but this could be related to the specific games I've played.
Probably the PALTEST is failing for some minor delay in XB (also the Turbo mode has problems with XB). I'll try to write some other code for this test.

In attachment the cart with some of the music used for the test (to load set Cart type to Uber GROM, then Load Mega Cart). Note that the 256K cart posted on AA (https://atariage.com/forums/topic/32418 ... nt=4892129) is not working on MiSTer core while is working fine on Classic99 (but I've not tested on the real iron).

tmop · Unread post by **tmop** » Tue Aug 24, 2021 6:35 pm

@Flandango,

I've done some tests for the PAL/NTSC. The fact that the difference of the setting is visible with music, Megademo, etc. and not with the PALTEST XB program has lead me to think that the loop used in it is no long enough for helping with the debugging.

I've created a new XB program (it loads automatically in XB from the disk) and run it on Classic 99, Mister, MAME and real PAL TI99s (I've no NTSC machines).
The program creates a simple sprite that moves at constant speed in vertical, there is a empty FOR/NEXT loop to add some delay (the sprites continues to move) and at the end the sprite is stopped and its X position (upper left corner value) is displayed. If the sprite reaces the bottom of the screen can appear again at top.

Results (file PAL_NTSC_Test_XB_Autoload.dsk, autoload with XB V110):
Classic 99 (V399.050) PAL: 146
Classic 99 (V399.050) NTSC: 175 or 176
Classic 99 (V399.031) PAL: 141
Classic 99 (V399.031) NTSC: 171
Mister PAL: 128
Mister NTSC: 154
MAME PAL (Version MESS UI 0219): 151 (internal 32K), 146 (with PEB Memory), 151 (without 32K)
MAME NTSC (Version MESS UI 0219): 173 (internal 32K), 175 (with PEB Memory), 182 (without 32K)
Real TI99 #1 PAL on CRT monitor: 18 (no 32K)
Real TI99 #2 PAL with F18A on LCD monitor (PEB 32K): 89
Real TI99 #3 PAL on CRT monitor + side 32K (from '80s): 146
Real TI99 #4 PAL on LCD monitor + side 32K (NanoPEB): 146

Classic 99 has the 32K by default. The results with MESS/MAME seem to be the same using various versions. The test on TI99 #3 was done with a real XB cart and the FinalGROM version with the same results.

I've then compiled the XB code, changing the FOR/NEXT loop from 2500 to 25000 to add enough delay due to extra speed from compilation).

Results (file TEST1_8.bin and TEST1.rpk for MAME):
Classic 99 (V399.050) PAL: 98
Classic 99 (V399.050) NTSC: 118
Classic 99 (V399.031) PAL: 96
Classic 99 (V399.031) NTSC: 116
Mister PAL: 86
Mister NTSC: 103
MAME PAL (Version MESS UI 0219): 68 (internal 32K), 97 (with PEB Memory)
MAME NTSC (Version MESS UI 0219): 82 (internal 32K), 117 (with PEB Memory)
Real TI99 #1 PAL on CRT monitor +32K (recent 32K expansion): 97
Real TI99 #2 PAL with F18A on LCD monitor +32K (PEB 32K): 116
Real TI99 #3 PAL on CRT monitor + side 32K (from '80s): 97
Real TI99 #4 PAL on LCD monitor + side 32K (NanoPEB): 97

The compiled code needs a 32K expansion to run.

Finally, I've modified the compiled version removing the empty loop and using a CALL LINK("DELAY",10000) instead of the FOR loop from 1 to 25000 (it should be the same value of the empty loop, in theory).

Results (file TEST2_8.bin and TEST2.rpk for MAME):
Classic 99 (V399.050) PAL: 188
Classic 99 (V399.050) NTSC: 188
Classic 99 (V399.031) PAL: 188
Classic 99 (V399.031) NTSC: 188
Mister PAL: 188
Mister NTSC: 188
MAME PAL (Version MESS UI 0219): 188 (internal 32K), 188 (with PEB Memory)
MAME NTSC (Version MESS UI 0219): 188 (internal 32K), 188 (with PEB Memory)
Real TI99 #1 PAL on CRT monitor +32K (recent 32K expansion): 188
Real TI99 #2 PAL with F18A on LCD monitor +32K (PEB 32K): 188
Real TI99 #3 PAL on CRT monitor + side 32K (from '80s): 188
Real TI99 #4 PAL on LCD monitor + side 32K (NanoPEB):188

Yes, it's the same results in all cases!

Some considerations: in TEST1 the 3 PAL TI without the F18A mod have all the same result, that is the same also for MAME (with PEB memory) and Classic 99. MiSTer TI99 core is close (approx. 11%). And the same for the NTSC version (if someone, maybe @rhester72, could confirm with a NTSC machine it would be great).

For the first test in XB there is less uniformity, due to different speed of memory for parsing the XB code. In particular if we look at the TI 99 #1, we can see the extra delay introduced by internal 16K VDP memory.

I hope that these test could help in the final tuning of timings. I think we are really closer.

Moreover, during a test on a disk (066-b-4 XB Games (19xx).dsk, in attachment) I've found a XB loader that has I/O errors with the loader if run on NTSC, but not on PAL. It's working fine in MAME (both PAL and NTSC).

Flandango · Unread post by **Flandango** » Wed Aug 25, 2021 11:27 am

WOW thanks for all the work Tmop! This should help me a lot. Yes XB is very sensitive to timing. Fix one thing, another thing breaks.

Flandango · Unread post by **Flandango** » Wed Aug 25, 2021 5:13 pm

Well, I did some tweaking and hopefully this bring it closer to the values Tmop posted.
It also seems to have fixed the 006-b-4 XB Games disk posted above....at least it doesn't crash immediately on startup.

tmop · Unread post by **tmop** » Thu Aug 26, 2021 2:55 pm

Flandango wrote: ↑Wed Aug 25, 2021 5:13 pm Well, I did some tweaking and hopefully this bring it closer to the values Tmop posted.
It also seems to have fixed the 006-b-4 XB Games disk posted above....at least it doesn't crash immediately on startup.

@Flandango,

excellent work!

The PAL_NTSC_Test_XB_Autoload.dsk is now reporting on my MiSTer (with a 32MB memory bank):
Mister PAL: 147
Mister NTSC: 176

and TEST1_8.bin:
Mister PAL: 98
Mister NTSC: 118

The previous PALNTSC XB test is now returning the correct results for PAL and NTSC and the Indiana game is only working fine if the machine is set to PAL!

I've run some tests on DISK subsystem and all seems to work fine. I've tested some of the disks that had problems in the past (Old Dark Caves 2, Display Master) and other disks with various loaders and everything was loaded correctly. I've also successfully formatted a disk with Disk Manager 1000.

I suggest to start checking the process for having this version published in the official repository.

Really an excellent work! I hope that you are still having fun with the TI99 and would like to add some other features... some suggestions are the F18A (to play some excellent homebrew conversion like Mario Bros, Zaxxon, etc.), tape support (maybe with .wav files like in MAME), GRAM Kracker.

Flandango · Unread post by **Flandango** » Thu Sep 02, 2021 12:48 am

Something I am trying out....

tmop · Unread post by **tmop** » Thu Sep 02, 2021 7:47 am

Flandango wrote: ↑Thu Sep 02, 2021 12:48 am Something I am trying out....

Yes!!! Great news!!!

I do not want to spoil the new feature in progress, but it's really cool!

Just tested on the fly. The Scandobler have to be removed (I had it set to CRT 25%) to have it booting correctly, but it's a great start for an alpha version.

Waiting impatiently for new updates!

Thanks for it, you are the man!

Flandango · Unread post by **Flandango** » Thu Sep 02, 2021 12:08 pm

tmop wrote: ↑Thu Sep 02, 2021 7:47 am The Scandobler have to be removed (I had it set to CRT 25%) to have it booting correctly, but it's a great start for an alpha version.

Yes, it has to go. The F18A has it's own scanline setting which I didn't setup yet. So for the time being it should be set to Normal.

This is a lot tougher to implement mostly because it requires a 100mhz crystal/clock and the 3 clocks on the DE10-Nano are being used up so I had to use the clock that is used for the TI system clock and derive those clocks from this one, which are then off by a bit, enough that can issues.
But I'll keep playing with it and see what happens. So far I only had success running a few F18A demos/apps/games. 80 column apps seem to work ok.

tmop · Unread post by **tmop** » Sun Sep 05, 2021 10:07 am

In attachment a disk of demos for the F18A if someone is curious to test this hardware. There is a demo of a space game that is really interesting.
Some of the demos are not working/crashing but they can provide an idea of this VDP chip replacement.

The F18A implementation could also be used on other systems based on the TMS9918 (MSX1, Colecovision, etc.) as for the real hw.

breiztiger · Unread post by **breiztiger** » Sun Sep 05, 2021 11:06 am

sorry but how to run this demos disk

i don't know anythings from ti99 but want to learn

edit : i have found, i must load extended ti basic and after disk autoload

Flandango · Unread post by **Flandango** » Sun Sep 05, 2021 11:23 am

breiztiger wrote: ↑Sun Sep 05, 2021 11:06 am sorry but how to run this demos disk

i don't know anythings from ti99 but want to learn

First, this is to demo the F18A capabilities and most of it will crash on you since F18A in this core is not working fully.
But if you still want to try it out, first thing you are going to need is a full rom image with Extended Basic and disk support.
Once you have that and loaded up with the Load Full option or you have named it boot.rom in your games folder to auto boot, then select the disk image that Tmop linked above with the Drive 1 *.DSK option, and on the boot screen select option #2 For TI Extended Basic, disk will autoload and then select which demo you want to try.

Moondandy · Unread post by **Moondandy** » Mon Sep 06, 2021 5:51 pm

Has anyone ever looked into supporting the Tomy Tutor in this core, or branching off and making one? I was just reading about it now and has similar architecture it seems.

The Tomy Tutor, originally sold in Japan as the Pyūta (ぴゅう太) and in the UK as the Grandstand Tutor,[2] is a home computer produced by the Japanese toymaker Tomy. It was architecturally similar, but not identical, to the Texas Instruments TI-99/4A, and used a similar Texas Instruments 16-bit CPU.[3] The computer was launched in Japan in 1982, and in the UK and the United States in the next year.

Technical specifications
CPU: Texas Instruments TMS9995
Video: Texas Instruments TMS9918 VDP
Resulution: 256 x 192 pixels
16 colors, up to 2 colors out of 8 horizontal dots
Up to 32 monochrome sprites of 8 x 8 pixels, max 4 per horizontal line
No hardware scrolling function
Memory: 20 KB ROM built-in, 16 KB RAM (with 256 bytes CPU RAM built-in)
Keyboard: 56 keys, JIS compliant
Sounds: 4 onomatopoeia and 3 chords
Weight: 1.7 kg (3.75 pounds)
Media: ROM cartridge, cassette
Expansion: data recorder (optional), game adapter

https://en.wikipedia.org/wiki/Tomy_Tutor

EeDee · Unread post by **EeDee** » Mon Sep 06, 2021 9:04 pm

Moondandy wrote: ↑Mon Sep 06, 2021 5:51 pm Has anyone ever looked into supporting the Tom Tutor in this core, or branching off and making one? I was just reading about it now and has similar architecture it seems.

Just looked it up and rediscovered this:
https://m.youtube.com/watch?v=iIJyNpNFMvY

tmop · Unread post by **tmop** » Tue Sep 21, 2021 3:48 pm

The updated MegaPack V3.1 for the TI99 core is now available for download from the TI99IUC site:

https://www.ti99iuc.it/web/index.php?pa ... Un8RexxeUk!

it includes disks and carts to test the new features (eg. MiniMemory games, etc.) added in the last update of the core.

There is a readme file in the archive with some instructions on how to load the various types of programs for those that are not familiar with the TI99. Of course, in case of doubts/clarifications, just ask...

cathrynmataga · Tue Sep 21, 2021 5:04 pm

Nice, this core looks pretty good what I've seen so far. And yeah, I never had TI99, to me it is all a little quirky. I remember hearing TI99 programmers complaining and taking a long time and wondering myself what was so hard about it back in the day.

rhester72 · Unread post by **rhester72** » Wed Sep 22, 2021 12:37 am

The 99 had...personality. *laughs* But it's those interesting quirks that distinguished it and keep it worth exploring to this day.

Then again, it was also my first computer, and in that era *ANY* computer was like a combination of black magic and Disneyland. It's something easily recaptured but hard to imagine for someone new to the machine in the 21st century.

MiSTer FPGA Forum

TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To

Re: TI-994a How-To