Jump to content

Photo

Texas Instruments TI-99/8 and Other Unreleased


16 replies to this topic

#1 ColecoFan1981 OFFLINE  

ColecoFan1981

    Moonsweeper

  • 353 posts
  • Location:Milwaukie (Oak Grove), OR

Posted Tue Oct 20, 2009 6:34 PM

Hi,

What does everyone know about the Texas Instruments TI-99/8? While it never was released officially, a few prototype units do exist and are in the hands of a few collectors. There also existed the instruction book to this model.

The TI-99/8 was to be TI's answer to the Commodore 64. I also believe Atari's 1400XL, which was to replace the short-lived 1200XL, among other things featured a built-in voice synthesizer, was to be that company's answer to the C64. The 1400XL was never released because at the time Atari executives felt the company should be committed to doing video games. The TI-99/2, another no-release, was to be a low-cost alternative to the 99/4 line, with monochrome graphics and no sound.

Among other things, while it was backwards-compatible with the 99/4(A) library, for some reason the 99/8 was not compatible with the Extended BASIC which supported the older model.

And had TI not went through all their millions in the home computer industry, we also could've seen their competitor to the Commodore Amiga, the 99/16. And heck, there was this TI computer, the Professional, that was a TI in name only, which was an IBM-compatible computer and featured an Intel chip. This could've sold better if it were to have been based on Atari's 1600XL, which, once again, is on the list of never-released computers. The 1600XL was conceived to have both the usual MOS 6502C chip as well as the Intel 80186 chip for greater DOS compatibility. By that I mean the TI Pro could've been made by TI itself and featured not only their TMS9995 chip but also the Intel 80186 chip.

Other TI computers that could've been released if the company's home computer division survived:
*TI-99/GC (TI-99 Game Console; remember Atari's XE-GS computer/console and the Tomy Pyuuta, Jr.?)
*TI-99/M (TI-99 Music; this would feature an 8-voice sound chip to rival Commodore's SID sound chip)
*TI-99/P (TI-99 Portable; with a built-in monochrome monitor)

~Ben

Edited by ColecoFan1981, Tue Oct 20, 2009 6:50 PM.


#2 Tempest OFFLINE  

Tempest

    Fallen Moderator

  • 27,124 posts
  • Location:Dark Places

Posted Tue Oct 20, 2009 6:45 PM

I know I want one. :)

Tempest

#3 Wickeycolumbus OFFLINE  

Wickeycolumbus

    Red Sea Crosser

  • 5,157 posts
  • Location:Michigan

Posted Tue Oct 20, 2009 7:50 PM

I remember seeing one of these on eBay a few years back when I was into the TI 99 (as well as several prototype add-ons). I must say, it is pretty cool :) Wish I owned one!

#4 retroclouds OFFLINE  

retroclouds

    Stargunner

  • 1,627 posts
  • Location:Germany

Posted Wed Oct 21, 2009 11:36 AM

You can find some details about the TI-99/8 prototype below:
http://www.99er.net/998.html
http://www.ninerpedi...dex.php/TI-99/8

Here is a picture of the box:
http://www.facebook....&oid=2209774843

If I'm not mistaking MESS already (partly) emulates the TI-99/8.

retroclouds

#5 Tursi OFFLINE  

Tursi

    Quadrunner

  • 5,231 posts
  • HarmlessLion
  • Location:BUR

Posted Wed Oct 21, 2009 11:43 AM

The TI Professional computer wasn't related to the 9900 line, and though it was Intel based, it wasn't compatible with the PC, which prevented it from arising as a serious clone.

The 99/8 is emulated in MESS. Reportedly the ROMs were assembled from printed listings rather than dumped from any machine - I don't know what state any of the prototypes are in so far as being operational. The 99/8 came with a full 64K of RAM (expandable to megs), ran on the 12MHz 9995 (4 times faster than the 4A - the boot menu let you slow it down for compatibility), and as well as the speech synth, had both BASIC and Pascal languages in ROM. I believe that the built-in BASIC was Extended BASIC II, which would make the incompatibility with regular XB less of a problem. However, it had the same graphics chip (9918A) and same sound chip, so there was little enhancement on that side of things. This means still no 80-column text, and no improved graphics/sound for games. I don't know if that was the final plan or not. So far as I know, all of the known prototypes are in different states of completion. I also don't know if any run. ;)

I've never heard of the 99/16, the 99/GC or the 99/M. The 99/P I believe I heard a note about.. but TI was reportedly working on as many as eight machines when they pulled the plug. ;)

Though it's worth noting that the 99/4 started life as a game machine and this was changed to a computer relatively late in production. The original 99/4 design had a volume slider and headphone jack on the front, and a port on the top for an IR receiver (intended for the wireless controllers). I'm told some consoles have the headphone jack and volume slider (others fill the hole and insert a "solid state software" badge to cover up the slider slot). The IR receiver mount is hidden underneath the aluminum shell but is still there in the plastic if you take it apart. I took a bunch of photos of the inside of my 99/4 last night, since there isn't a lot on the web about it, and will add them to Ninerpedia.org in the next few days.

At any rate, the 99/8 was looking like it would have been a great machine if it had made it to market. I wouldn't mind one either, if it works. ;)

(edit: Kroy's links correct me, the box says 80k RAM (never saw the box before, nice!), and the Wiki page lists a new video chip but with similar features.., Wiki also claims 10Mhz rather than 12Mhz.)

Edited by Tursi, Wed Oct 21, 2009 11:46 AM.


#6 Tempest OFFLINE  

Tempest

    Fallen Moderator

  • 27,124 posts
  • Location:Dark Places

Posted Wed Oct 21, 2009 11:59 AM

There are at least some operational prototypes. I think one had the Pascal built in while the other didn't.

Tempest

#7 ColecoFan1981 OFFLINE  

ColecoFan1981

    Moonsweeper

  • Topic Starter
  • 353 posts
  • Location:Milwaukie (Oak Grove), OR

Posted Wed Oct 21, 2009 12:08 PM

You can find some details about the TI-99/8 prototype below:
http://www.99er.net/998.html
http://www.ninerpedi...dex.php/TI-99/8

Here is a picture of the box:
http://www.facebook....&oid=2209774843

If I'm not mistaking MESS already (partly) emulates the TI-99/8.

retroclouds


I also have to say that the 99/8 never took off due to problems correlating to 3rd-party cartridges: they must contain GROM. Again, TI shouldn't have (wrongfully) gone through their millions over the 99/4A line. The Extended BASIC I problem on this prototype correlates to the fact the computer was already pre-loaded with the Extended BASIC II. Another problem lies with the Terminal Emulator program.

~Ben

#8 Kurt_Woloch OFFLINE  

Kurt_Woloch

    Stargunner

  • 1,369 posts

Posted Mon Oct 26, 2009 2:08 PM

(edit: Kroy's links correct me, the box says 80k RAM (never saw the box before, nice!), and the Wiki page lists a new video chip but with similar features.., Wiki also claims 10Mhz rather than 12Mhz.)


OK, then I suppose the 80K would have been 16K of video RAM (the same amount the TI-99/4A has - it's the maximum the video chip can address) and 64K of CPU RAM (while the TI-99 has only 256 bytes here).

#9 Casey OFFLINE  

Casey

    Moonsweeper

  • 290 posts

Posted Tue Oct 5, 2010 5:12 PM

I know that I am late to this discussion, but I just saw this topic via a Google search and I wanted to add my experiences.

At one time, for a few years, I owned a TI 99/8. The machine I had had a functioning p-System in it, accessible from the master selection list. Where the MESS emulator gives you option A for Extended BASIC II and B for Set Speed, my unit would offer the p-System as option B and C was used for the Set Speed choice. Once in the p-System, there was nothing more I could try, as my unit did not have any ability to use a disk drive, nor did I have any of the p-System disks.

Extended BASIC II ran many times faster than Extended BASIC for the TI 99/4A, partially because the CPU was faster, and partially because Extended BASIC II was not written in GPL, but native assembly language.

My 99/8 did not have a functional peripheral expansion port. The cutout for the port was there, but the system board had no connection there, meaning I could only use cassette or Hex Bus for expansion.

Cassette based programs were interchangeable between the 99/4A and the 99/8, though BASIC programs, as I said before, ran much faster. The set speed option had no effect on BASIC - it always ran at full speed. My recollection is that cartridge based games typically ran at the same speed as on the 99/4A, regardless of the set speed chosen.

Extended BASIC II could use CALL SAY to access built-in speech, but as stated by others, OPEN #1:"SPEECH",OUTPUT did not work.

It definitely would have been a nice machine to upgrade from a 99/4A to - it's too bad it didn't get produced!

Casey

#10 retroclouds OFFLINE  

retroclouds

    Stargunner

  • 1,627 posts
  • Location:Germany

Posted Wed Oct 6, 2010 11:01 AM

I know that I am late to this discussion, but I just saw this topic via a Google search and I wanted to add my experiences.

At one time, for a few years, I owned a TI 99/8. The machine I had had a functioning p-System in it, accessible from the master selection list. Where the MESS emulator gives you option A for Extended BASIC II and B for Set Speed, my unit would offer the p-System as option B and C was used for the Set Speed choice. Once in the p-System, there was nothing more I could try, as my unit did not have any ability to use a disk drive, nor did I have any of the p-System disks.

Extended BASIC II ran many times faster than Extended BASIC for the TI 99/4A, partially because the CPU was faster, and partially because Extended BASIC II was not written in GPL, but native assembly language.

My 99/8 did not have a functional peripheral expansion port. The cutout for the port was there, but the system board had no connection there, meaning I could only use cassette or Hex Bus for expansion.

Cassette based programs were interchangeable between the 99/4A and the 99/8, though BASIC programs, as I said before, ran much faster. The set speed option had no effect on BASIC - it always ran at full speed. My recollection is that cartridge based games typically ran at the same speed as on the 99/4A, regardless of the set speed chosen.

Extended BASIC II could use CALL SAY to access built-in speech, but as stated by others, OPEN #1:"SPEECH",OUTPUT did not work.

It definitely would have been a nice machine to upgrade from a 99/4A to - it's too bad it didn't get produced!

Casey



Do you mind sharing how you obtained the machine ? Did you sell it ?

#11 OldAtarian OFFLINE  

OldAtarian

    Stargunner

  • 1,599 posts

Posted Thu Oct 7, 2010 6:25 PM

I read that most of the prototypes for the 99/8 exist in motherboard form only. There's probably less than a handful that are complete, and maybe not even that many.

#12 Ksarul OFFLINE  

Ksarul

    River Patroller

  • 4,741 posts

Posted Sun Jul 10, 2011 6:38 PM

Actually, quite a few of them made it as far as being full machines, although that is a relative term. I have read that the total was in the range of 150 fully assembled machines, and my two fit into that range--I have serial number 80 and serial number 135. Both have the pSystem installed, although there is a problem with the copy installed on S/N 80 that I haven't tracked down yet. I also have the pSystem disks for the 99/8. It used V4.11, not V4.0 as in the /4A. I've scanned most of the documentation for it and put it up on WHTech, I'm still missing a few of the internal engineering documents, but not many. Several interesting 99/8 only PEB cards exist as well: 128K memory cards, 512K memory cards (I have one of each), and an RS232 card with a modified DSR. It was a nice machine, but it never made it into full production before they pulled the plug. I've also got a couple of partially populated motherboards with serial numbers in the 250-350 range, though these were never installed in cases, and most of the daughter boards are missing. They appear to have been used as parts mines over the years. . .as several 99/8 chips exist only on those boards.

Edited by Ksarul, Sun Jul 10, 2011 6:39 PM.

  • jhd likes this

#13 JamesD OFFLINE  

JamesD

    Quadrunner

  • 8,209 posts
  • Location:Flyover State

Posted Mon Jul 11, 2011 2:55 AM

Didn't the 99/8 also have the RAM on a full 16 bit buss instead of 8 which also made it faster than the 99/4?

#14 Ksarul OFFLINE  

Ksarul

    River Patroller

  • 4,741 posts

Posted Mon Jul 11, 2011 4:50 AM

That has less of an impact than you might think here--the 99/4A had a 16-bit data bus at teh CPU and polled memory twice to fill the LSB/MSB bytes because the memory was on an 8-bit bus. The 99/8 basically had to do the same thing because the 16-bit bus was internal to the processor--the external bus on the processor was 8-bit. The real difference was in processor speed--the 99/8 was clocked at 10 MHz. It was also possible to increase speed on a /4A by putting 32K on the processor's 16-bit bus in the console, as the bottleneck was in the I/O port. That gave a zero to 25% increase in speed, depending on the program being executed.

#15 apersson850 OFFLINE  

apersson850

    Dragonstomper

  • 500 posts

Posted Mon Jan 26, 2015 7:18 AM

The impact of having fast memory in the 99/4A is larger than stated above. A normal memory cycle for a TMS 9900 requires two CPU cycles. Due to the multiplexing logic for the 16 <-> 8 bit conversion for external memory (like the 32 K RAM expansion in the PEB), four cycles are required in the 99/4A. On top of that, one wait state per byte access is added, so a one memory cycle expands to six CPU cycles.

 

Normally, when you write assembly language programs on the 99/4A, you try to keep the workspace somewhere inside the 256 bytes of 16-bit wide RAM that exists in the console. But if you don't, but for some reason want both workspace, code and all other operands in the memory expansion, without using GROM or VDP RAM, then the speed gain is actually close to 110%. Programs will run in less than half the time. As I have a 99/4A console with 80 K memory (64 K CPU RAM + 16 K VDP RAM) myself, I know this very well.

But programs like TI BASIC, running totally outside the 32 K RAM expansion, is not affected at all.

 

The TMS 9995 is often said to run at a higher clock frequency than the TMS 9900. That's a modification of the truth. If you look at technical data for the 9995, you'll find that at the same clock frequency (3 MHz), its instructions will complete in about one third of the clock cycles required by the 9900. Thus it's not really the clock frequency that changes, but the efficiency of the CPU. Part of this is due to a one-stage pipeline, where the next instruction is pre-fetched before the current one is completed. Thus this cycle, which depicts an add instruction (parts requiring memory access marked with [M]) in the 9900 can be compacted so that the 9995 is reading/writing to memory at every stage.

 

Instruction fetch [M],

instruction decode,

first operand fetch [M],

second operand fetch [M],

add them together,

write result [M].


Edited by apersson850, Mon Jan 26, 2015 7:19 AM.


#16 mizapf OFFLINE  

mizapf

    River Patroller

  • 3,350 posts
  • Location:Germany

Posted Mon Jan 26, 2015 10:30 AM

Resurrecting a 3.5-year-old thread? ;)

 

One more thing is significant: The TMS9900 uses 2 clock cycles per machine cycle; accordingly, the A operation (add word) takes 14 clock cycles, while the A operation on the 9995 only takes 6, with 2 cycles hidden in command prefetch/decode and 1 ALU cycle less than 9900.

 

The higher clock frequency could be effective inside the CPU when executing the micro-instructions. I do not know the exact implementation, but I can imagine that each of the fetch/alu/store micro-instructions may take up to four input clock cycles (@ 12 MHz), while the visible cycle time is only 333ns.



#17 jens-eike OFFLINE  

jens-eike

    Star Raider

  • 82 posts

Posted Tue Jan 27, 2015 6:26 AM

Actually, both CPUs run on 12MHz.

 

The 9995 has an internal clock generator deriving it's microcycles from the crystal, externally, all we see is the memory cycle and the -phi clock output.

 

The 9900 gets four clock phases from it's TIM9904(A), derived from a 12MHz crystal (older TIM9904 use a 48MHz crystal, most 99/4 have these). Peripheral chips like the TMS9902 use only one phase of the clock signal, so a 3MHz signal -phi3 controls most of the timing.

 

Note: -phi3 is one fourth of the clock signals, so it's duty cycle is about 80ns active (low) and ~250ns inactive (high) (+/- rise and fall times). Early 9901s had a problem with the 50% duty cycle of the 9995, so TI produced chips marked TMS9901-95 and TMS9902-95 for compatibility with the 9995.






0 user(s) are browsing this forum

0 members, 0 guests, 0 anonymous users