Jump to content

Photo

TMS99110 - nice!


7 replies to this topic

#1 Willsy ONLINE  

Willsy

    Stargunner

  • 1,874 posts
  • Location:Uzbekistan (no, really!)

Posted Thu May 10, 2012 2:24 AM

Just reading about the TMS99110 processor here. Wow, what a smart chip!
  • 32 bit math instructions
  • 32 bit logic instructions
  • Programmable macro store (create new machine code instructions!)
  • Stack
Yummy! I want one!

TMS 99110 microprocessor is a member of TMS99000 family - the third generation of Texas Instruments' 16-bit microprocessors. Like two previous generations, TMS9900 and TMS9995, the TMS99110 family is based on memory-to-memory architecture. The main advantage of this architecture is that the set of CPU registers (called "Workspace") can be located anywhere in memory. This architecture makes saving and restoring of the contents of all CPU registers as simple as switching the base address of the workspace. The disadvantage of this architecture is that the processor speed is highly dependent on memory speed. The 99110 CPUs have external clock frequency 24 MHz, which is divided by four internally. It takes three or more machine cycles for the CPU to execute any instruction when no wait states are required to access the memory. This translates to maximum execution speed 2,000,000 instructions per second or less. Execution speed drops significantly when the processor is used with slow memory - instructions may execute up to two times slower when memory access requires 1 wait-state, or up to 3 times slower when 2 wait-states are required.
The TMS99110 family supports all instructions found in TMS9900 and TMS9995 families, and includes new instructions - 32-bit arithmetic and logic, bit testing, signed multiply and divide, and stack operations. In addition to new instructions, any unused processor opcode on the 99000 can be redefined as a new instruction. This can be done with the help of an attached processor (co-processor), or with the help of a Macrostore feature. The Macrostore allows system designers to map unused microprocessor opcodes to custom functions located in a separate memory address space. The TMS99110 even comes with 15 such "pseudo" instructions (12 of them are Floating Point instructions). Because the Macrostore memory is separate from main memory, this allows the CPU to access up to 128 KB simultaneously (64 KB of main memory + 64 KB of Macrostore memory). It is possible to increase maximum addressable memory size to 192 KB by splitting main memory into two memory segments - 64 KB instruction segment and 64 KB data segment.
The TMS99110 is almost identical to TMS99105 16-bit processor, except that the TMS99105 doesn't have on-chip memory with 15 pre-programmed instructions.



#2 Stuart OFFLINE  

Stuart

    Chopper Commander

  • 177 posts
  • Location:Southampton, UK

Posted Thu May 10, 2012 5:27 AM

This bit ...

[[ It is possible to increase maximum addressable memory size to 192 KB by splitting main memory into two memory segments - 64 KB instruction segment and 64 KB data segment. ]]

The 99000 family has a set of bus activity output lines which indicate exactly what the bus is doing - getting an instruction, getting a first operand, getting a second operand, writing a result, and so on. So by decoding this along with the memory address, the bus activity can be directed to different, independent banks of memory. Which must lead to some interesting things with assemblers/compilers as an instruction operand could in theory have the same address as the instruction itself - it's just retrieved from a different bank of memory.

#3 Willsy ONLINE  

Willsy

    Stargunner

  • Topic Starter
  • 1,874 posts
  • Location:Uzbekistan (no, really!)

Posted Thu May 10, 2012 6:12 AM

It really is a shame that the 9900 family never "caught on" (though it was quite successful in other areas).

#4 sometimes99er OFFLINE  

sometimes99er

    River Patroller

  • 2,706 posts
  • Location:Denmark

Posted Fri May 11, 2012 11:30 PM

Nice.

I went thru Z80A (Timex/Sinclair), 9900 (TI-99/4A), 6510 (C64), 68000 (Amiga) and some limited x86 (PC). Boy was the 6510 taking a step back, but capabilities and numbers, made it need to do. x86 also felt like a step back.

I still do use pencil and paper a lot, and the Amiga finally made it possible to have stacks of notes and drawings handy in electronic form (well, it was cassette based prior for me).

#5 matthew180 OFFLINE  

matthew180

    Stargunner

  • 1,806 posts
  • Location:Ceresco, Michigan

Posted Fri May 18, 2012 2:56 PM

I've been looking into a lot of different CPUs lately, watching videos of presentations by pioneers like Alan Kay, etc., and this thread was just another "ironic" event that lead me to actually dig into the 99000. For once, I think TI was really on to something with some of the ideas on the 99000. Once you start looking at the hardware from the Mainframe and Mini era, or anything pre-1980's, you see a lot of exploration and experimentation going on in the hardware architecture of machines. But, when "home computers" took over, computer architecture and advancement stopped. It was interesting to hear Alan Kay say in one of this lectures that "we may never recover from the home computer era". When you look at things from a hardware perspective, it is obvious what he means. I'm so glad FPGAs exist.

#6 Willsy ONLINE  

Willsy

    Stargunner

  • Topic Starter
  • 1,874 posts
  • Location:Uzbekistan (no, really!)

Posted Fri May 18, 2012 3:25 PM

Can you post links to the Alan Kay stuff? Who is he? Never heard of him. Might be interesting.

#7 matthew180 OFFLINE  

matthew180

    Stargunner

  • 1,806 posts
  • Location:Ceresco, Michigan

Posted Fri May 18, 2012 3:53 PM

Alan Kay was one of the people who was at Xerox PARC (Palo Alto Research Center) and worked on the Star computer, i.e. the one in the 1970's that had a GUI, Mouse, Laser Printer, and Network. Not sure how computers went from that to 80x25 text. And in the 1960's Engelbart invented Skype. If you have not watched the Engelbart videos, you are required by law as a geek to watch them. Know you history, know where you came from, and sadly know how computers were seriously set back in the 80's. Just like going to the moon in a decade, Mars should have been next by 1980, and "Space 1999" should have been reality. Humans are easily distracted by shiny things. "oh shiny!" Anyway, I digress.

Just go to youtube and search on these topics and be prepared to spend a lot of hours:

* Douglas Engelbart (The Mother of All Demos - yes you must watch all NINE parts!)

* Alan Kay (I recommend the "Personal Computing: Historic Beginnings" and "Ivan Sutherland's Sketchpad" videos in particular, as well as the others on education.)

* bbc computer history

* computer history

* xerox computer

Edited by matthew180, Sat May 19, 2012 1:04 AM.


#8 sometimes99er OFFLINE  

sometimes99er

    River Patroller

  • 2,706 posts
  • Location:Denmark

Posted Fri May 18, 2012 10:47 PM

That stuff is interesting reading to me too. There’s a few different opinions in there, apparently not too easy to predict the future at some levels. When “Space 1999” was running, it seemed like in a distant future, and with Apollo under your belt, then certainly why not ? Recently I have begun to watch “Space 1999” - if you can abstract from present day (like with the TI-99/4A) it's just wonderful. - When I look back I think of my 1987 Amiga 2000 as about the only real quantum leap.






0 user(s) are browsing this forum

0 members, 0 guests, 0 anonymous users