Jump to content
IGNORED

Are there accelerator cards for the TI-99/4A?


Boschloo

Recommended Posts

ChatGPT answered this: "Yes, there were accelerator cards developed for the Texas Instruments TI-99/4A, although they were not as common as those for some other platforms like the Apple II or the Commodore 64. These accelerator cards were designed to increase the processing speed of the TI-99/4A, which originally came with a 3 MHz TMS9900 CPU but was internally slowed down to approximately 300 kHz due to its architecture."

WOW! 300kHz??? So it was crippled to a tenth???

What accelerator cards are there for the TI?

  • Haha 4
Link to comment
Share on other sites

Now more than ever, humans need to learn to check the sources of any information they receive. 

 

I have heard that there was a 99105 attempt at a CPU replacement accelerator, but that it never reached a functional state. I cannot back that up with any sources except our chatter. 

 

Overclocking to 4Mhz was a thing. 

 

Installing zero wait state 32k memory expansion inside the console was a thing that is well documented.

 

The F18A VDP replacement has compute power that is accessible, like a coprocessor with its own memory.

 

The strangecart accelerates TI BASIC to crazy speed. 

 

The Myarc 9640 Geneve replaces the console & flex cable with a card that runs most software accelerated with a 9995 processor.

  • Like 3
Link to comment
Share on other sites

Reminder: ChatGPT is not a search engine, nor is it a reliable source of information.

 

 

To my knowledge, there are no "accelerator cards" for the 99/4a.

And I'm reasonably sure the "300 kilohertz" thing is equally bullcrap(though there is truth in the 4a's architecture restricting performance)

  • Thanks 1
Link to comment
Share on other sites

26 minutes ago, roots.genoa said:

Meme Reaction GIF by Robert E Blackmon

 

I'm tired of people using ChatGPT as Google. It's a conversational AI. It just says what is "statistically more probable" to be said, but in terms of words, not meaning.

Believe it or not we are now exploring its use in the medical field. Already it has exceeded the performance of residents on the board exams and it can generate differential diagnoses with uncanny precision. However, its performance is proportional to the number and quality of the starting information, and that is where humans come in as we are better able to extrapolate based on limited data and non-specific clinical findings. Don't toss your physician just yet! 😄

  • Like 1
Link to comment
Share on other sites

9 hours ago, jedimatt42 said:

Now more than ever, humans need to learn to check the sources of any information they receive. 

 

I have heard that there was a 99105 attempt at a CPU replacement accelerator, but that it never reached a functional state. I cannot back that up with any sources except our chatter. 

 

Overclocking to 4Mhz was a thing. 

 

Installing zero wait state 32k memory expansion inside the console was a thing that is well documented.

 

The F18A VDP replacement has compute power that is accessible, like a coprocessor with its own memory.

 

The strangecart accelerates TI BASIC to crazy speed. 

 

The Myarc 9640 Geneve replaces the console & flex cable with a card that runs most software accelerated with a 9995 processor.

On the 99105 accelerator, I think that @Gary from OPA may be able to shed some light on the subject. ISTR that the major issues were related to bus timing, especially as related to the other peripherals in the system.

 

The only other accelerator that I've seen mention of from BITD was a math accelerator card using the Motorola 68881. That one never made it past the prototype stage (I actually have that prototype and the rights to further develop it).

  • Like 2
Link to comment
Share on other sites

On 11/25/2023 at 7:25 AM, Vorticon said:

Don't toss your physician just yet! 😄

Never.

On 11/25/2023 at 7:49 AM, GDMike said:

It might be good for legalease and sales I would think. 

Two New York attorneys were recently sanctioned for filing a brief generated by ChatGPT which included  citations of six fictitious cases.  When they asked ChatGPT if those cases actually existed, it said they did.  In other tests, ChatGPT "admits" that it can and will lie to reach a desired outcome.  But, at least it does not turn into a lovelorn stalker like Bing.

On 11/25/2023 at 9:40 AM, jedimatt42 said:

I forgot, the TI-99 MiSTer core also supports a Turbo mode for accelerated CPU speed.

I am not sure if ChatGPT would consider MiSTer or Classic99 as accelerator cards, unless someone out there has written that those are acceleration.

On 11/25/2023 at 12:05 PM, Ksarul said:

The only other accelerator that I've seen mention of from BITD was a math accelerator card using the Motorola 68881. That one never made it past the prototype stage (I actually have that prototype and the rights to further develop it).

The logic behind converting a 9900 buss cycle to a 68020 buss cycle for the 68881 makes my head hurt.  I can imagine, though, that having a math co-processor on the 9900 would speed things up considerably over GPL.  I recall another post here recently which said holding off the 9900 while allowing another CPU to take over the system would be possible, but I think it would need to be on the 9900's side of the expansion.  My first Amiga accelerator, and MTec 68020i with 68881, sat in the CPU socket.  Later I got (and still have) a Derringer 68030 with 68882, before upgrading to the GVP A530+ which connected to the expansion port  I have some thoughts on local CPU RAM versus Zorro II RAM on these, but my biggest thought related to the TI is the directionality of control through the multiplexer.

 

I can see a math co-pro in the PEB, but not a replacement system CPU as it could not get data to the VDP, PSG, &c.  Although, an auto-start DSR on an external CPU card could set up the 9900 to be a service device with a shared communication space between the CPUs, with the external CPU doing all the computational work.  Something similar to the Turbo XT but with way, way more intrusion into the system.

Link to comment
Share on other sites

9 hours ago, Vorticon said:

Believe it or not we are now exploring its use in the medical field.

AI has even been used for quite some time in the medical field, but not ChatGPT, which is once again just for chatting. A typical behavior of ChatGPT is when they're mistaken and you tell them, they'll apologize politely (as any person would)... then repeat the same mistake, or make another one. 🤦‍♂️ Personally, I may use it to write a first draft of a cover letter or maybe dialogues for a screenplay/video game for instance, but not much else, especially if it's important. It's said to be useful to program basic stuff if you're too lazy to make it yourself, but I'm pretty sure you could obtain the same result by googling the algorithm you're looking for.

Link to comment
Share on other sites

20 hours ago, jedimatt42 said:

Now more than ever, humans need to learn to check the sources of any information they receive. 

 

I have heard that there was a 99105 attempt at a CPU replacement accelerator, but that it never reached a functional state. I cannot back that up with any sources except our chatter. 

 

Overclocking to 4Mhz was a thing. 

 

Installing zero wait state 32k memory expansion inside the console was a thing that is well documented.

 

The F18A VDP replacement has compute power that is accessible, like a coprocessor with its own memory.

 

The strangecart accelerates TI BASIC to crazy speed. 

 

The Myarc 9640 Geneve replaces the console & flex cable with a card that runs most software accelerated with a 9995 processor.

Yep, and out of all of those, for the /4A itself, I think in the 80s and 90s, the most prevalent was the internal 32K mod, followed by the overclocking to 4MHz.  A lot of adjustments to DSR ROMs and such had to follow those because of timing issues for some critical things  (RS232 comes to mind, but there's cobwebs going back that far.)  

 

And I agree, people really need to double check facts that ChatGPT presents before posting.  When I utilize it as a reference, that's the first thing that I do.  The second is check how it structures sentences, because it loves coming up with some odd subject/object agreement syntax.  If I am going to quote it, I'm going to make sure it is using proper grammar and has the facts correct.

 

 

Link to comment
Share on other sites

23 hours ago, Boschloo said:

These accelerator cards were designed to increase the processing speed of the TI-99/4A, which originally came with a 3 MHz TMS9900 CPU but was internally slowed down to approximately 300 kHz due to its architecture."

WOW! 300kHz??? So it was crippled to a tenth???

I think ChatGPT is trying to say this: The TMS9900 was clocked at 3MHz but the internal architecture of the processor requires about 10 clock cycles to process each instruction. (I think the actual number is somewhat higher)

Of course, this was not done intentionally. It is just the nature of the 9900 and so it would not be accurate to say it was crippled.

As Jedimatt and Ksarul discuss in some posts above, there was some talk about a 99105 accelerator. I seem to remember seeing ads for it in Micropendium.

(edit) Although the 9900 does less mips than the 8 bit processors, remember that the instructions act on 16 bits and so that would help compensate.

Edited by senior_falcon
Link to comment
Share on other sites

7 hours ago, roots.genoa said:

AI has even been used for quite some time in the medical field, but not ChatGPT, which is once again just for chatting. A typical behavior of ChatGPT is when they're mistaken and you tell them, they'll apologize politely (as any person would)... then repeat the same mistake, or make another one. 🤦‍♂️ Personally, I may use it to write a first draft of a cover letter or maybe dialogues for a screenplay/video game for instance, but not much else, especially if it's important. It's said to be useful to program basic stuff if you're too lazy to make it yourself, but I'm pretty sure you could obtain the same result by googling the algorithm you're looking for.

Actually as a physician I can assure you that ChatGpt is being tested out in the medical field. As a statistical knowledge model, it's ideal for drafting differential diagnoses based on likelihood probabilities. 

  • Like 5
Link to comment
Share on other sites

Trying to understand why ChatGPT said 300khz is missing a fundamental point about it -- it has no idea what it is saying. It will state fact and fiction with equal conviction and the only scoring is really how closely the words appeared to each other in the training data.

 

It didn't mean ANYTHING when it said that. It saw some words about running slow being correlated with the TI architecture. It saw 300khz - and it might have had nothing to do with the TI where it saw that, but only slow, or architecture.

 

Don't waste brain cells trying to understand an AI hallucination. ;) It is perhaps the most literal case of "it didn't mean anything by it" we've ever had.

 

  • Like 6
Link to comment
Share on other sites

On 11/25/2023 at 9:40 AM, jedimatt42 said:

I forgot, the TI-99 MiSTer core also supports a Turbo mode for accelerated CPU speed.

I was thinking more in line with like the UltraWarp card that ReactiveMicro sells for the Apple IIe. It accelerates the CPU to 13Mhz. But I imagine it would be a hard thing to stick something similar into the TI expansion port. And, if an accelerator were created for the PEB, I guess that would be like an entire other computer as a card?

 

 

Link to comment
Share on other sites

On 11/25/2023 at 3:39 PM, OLD CS1 said:

I can see a math co-pro in the PEB, but not a replacement system CPU as it could not get data to the VDP, PSG, &c.  Although, an auto-start DSR on an external CPU card could set up the 9900 to be a service device with a shared communication space between the CPUs, with the external CPU doing all the computational work.  Something similar to the Turbo XT but with way, way more intrusion into the system.

Just throwing this out there, but if someone expanded the messaging system for the TIPI, you may be able to offload those calculations to the PI.  I doubt this is in Matt's interest, but if someone approached him with the skills to write the necessary code, he might be inclined to add it.

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...