Loading ...
Sorry, an error occurred while loading the content.

Using a 16 Bit CPU with a 8 Bit Bus

Expand Messages
  • Eric
    Sometimes I get really frustrated by the actions of the head people at TEXAS INSTRUMENTS at how much they purposely crippled the TI-99/4A! With Win994a the
    Message 1 of 22 , May 4, 2013
    • 0 Attachment
      Sometimes I get really frustrated by the actions of the head people at TEXAS INSTRUMENTS at how much they purposely crippled the TI-99/4A!

      With Win994a the user has the choice of using an 8 Bit bus to run the computer like TI supplied to the consumer or using a 16 Bit bus to match the CPU's actual architecture.

      A very PRIME example is the running of the conversion of the Extended Basic graphics and music program "Axel F" by Robert Gagle over to assembly by Mr. Jones.

      With the standard TI-99/4A P.E.B. 8 Bit bus, the graphics appear on the screen at the 'normal' refreshing rate and the music runs at the intended tempo.

      With a 16 Bit bus provided for the CPU's data to be funneled through, the graphics are now refreshed at a much quicker rate and the music now is played a a much, much faster tempo that makes it almost distorted in its presentation.

      QUESTION: Why did Texas Instruments cripple the TI-99/4A so much in its hardware architecture?

      With leadership like that it is no wonder the corporation was doomed to failure in the Home Computer Market!
    • Dano
      One phrase comes to mind immediately.... You can lead a horse to water but, you can t make him drink. It is truley sad that some thick skulls thwarted the
      Message 2 of 22 , May 4, 2013
      • 0 Attachment
        One phrase comes to mind immediately....

        "You can lead a horse to water but, you can't make him drink."

        It is truley sad that some thick skulls thwarted the TI...  Unfortunately, I see it all the time in the IT world and this is why I chose to exit IT, and IF I return, it will be under my terms not some thick skulled employer.

        -Dano

        "Long Live the TI-99/4A!"

        On Sat, May 4, 2013 at 11:29 AM, Eric <eric-bray@...> wrote:
         

        Sometimes I get really frustrated by the actions of the head people at TEXAS INSTRUMENTS at how much they purposely crippled the TI-99/4A!

        With Win994a the user has the choice of using an 8 Bit bus to run the computer like TI supplied to the consumer or using a 16 Bit bus to match the CPU's actual architecture.

        A very PRIME example is the running of the conversion of the Extended Basic graphics and music program "Axel F" by Robert Gagle over to assembly by Mr. Jones.

        With the standard TI-99/4A P.E.B. 8 Bit bus, the graphics appear on the screen at the 'normal' refreshing rate and the music runs at the intended tempo.

        With a 16 Bit bus provided for the CPU's data to be funneled through, the graphics are now refreshed at a much quicker rate and the music now is played a a much, much faster tempo that makes it almost distorted in its presentation.

        QUESTION: Why did Texas Instruments cripple the TI-99/4A so much in its hardware architecture?

        With leadership like that it is no wonder the corporation was doomed to failure in the Home Computer Market!


      • Owen Brand
        I d like to think that, while the developers didn t take 100% FULL advantage of the available technology, we are all still here... passionately pursuing the TI
        Message 3 of 22 , May 4, 2013
        • 0 Attachment
          I'd like to think that, while the developers didn't take 100% FULL advantage of the available technology, we are all still here... passionately pursuing the TI hardware and software modifications and improvements... 

          Lots of new games and utilities are being written, Matthew created the F18A, Jon's cart boards, Fred's HDX boards, Marc's SID card.... 

          It's a passionate group... There's a reason for that


          Owen


          From: Dano <ti994a@...>
          To: ti99-4a@yahoogroups.com
          Sent: Saturday, May 4, 2013 12:50 PM
          Subject: Re: [TI-99/4A] Using a 16 Bit CPU with a 8 Bit Bus

           
          One phrase comes to mind immediately....

          "You can lead a horse to water but, you can't make him drink."

          It is truley sad that some thick skulls thwarted the TI...  Unfortunately, I see it all the time in the IT world and this is why I chose to exit IT, and IF I return, it will be under my terms not some thick skulled employer.

          -Dano

          "Long Live the TI-99/4A!"

          On Sat, May 4, 2013 at 11:29 AM, Eric <eric-bray@...> wrote:
           
          Sometimes I get really frustrated by the actions of the head people at TEXAS INSTRUMENTS at how much they purposely crippled the TI-99/4A!

          With Win994a the user has the choice of using an 8 Bit bus to run the computer like TI supplied to the consumer or using a 16 Bit bus to match the CPU's actual architecture.

          A very PRIME example is the running of the conversion of the Extended Basic graphics and music program "Axel F" by Robert Gagle over to assembly by Mr. Jones.

          With the standard TI-99/4A P.E.B. 8 Bit bus, the graphics appear on the screen at the 'normal' refreshing rate and the music runs at the intended tempo.

          With a 16 Bit bus provided for the CPU's data to be funneled through, the graphics are now refreshed at a much quicker rate and the music now is played a a much, much faster tempo that makes it almost distorted in its presentation.

          QUESTION: Why did Texas Instruments cripple the TI-99/4A so much in its hardware architecture?

          With leadership like that it is no wonder the corporation was doomed to failure in the Home Computer Market!




        • Michael Zapf
          As far as I know the design of the TI system was originally centered around 8-bit architectures, or 16-bit systems based on the 9980. The TMS 9980 is a 16-bit
          Message 4 of 22 , May 4, 2013
          • 0 Attachment
            As far as I know the design of the TI system was originally centered around 8-bit architectures, or 16-bit systems based on the 9980. The TMS 9980 is a 16-bit CPU with 8-bit data bus, but with a smaller address space (16 KiB). There must have been some last-minute change in the design of the TI-99/4 which led to the redesign for the TMS 9900.

            There is another problem which comes from the instruction set. The TMS99xx family has byte-oriented commands which raise some problems with the memory interaction. Interestingly, as you can see with the TMS9995 which only has an 8-bit data bus, an 8 bit data bus may turn out as an advantage in the presence of such byte commands. The 9995 is tremendously faster than the 9900 with the same cycle times.

            Example. When you do a "MOVB @>A001,@>A002", the 16-bit CPU must write a 16-bit value to A002 (and A003), but it has no information about A003 yet. Thus, it reads the word at A000, shifts left the word to get the low byte, then it reads the word at A002, moves the stored byte into the internal output register, and finally writes a word to A002 (including the previous value of A003). This is called "read-before-write", and in the TMS9900, it occurs on every write operation (even when the complete word will be written).

            When you have a closer look at the 9995 you see it does not apply read-before-write because it only uses an 8-bit bus. In turn, it has a complete address bus (A0-A15, 64K of 8-bit words) where the 9900 only has address lines A0-A14 (32K of 16-bit words). For a MOVB as above, the Geneve indeed does not fetch the word at the destination; it just writes the involved byte.

            Michael
          • Tursi
            It makes sense when you think about it from a project management point of view: 1) board is designed and laid-out for 8-bit CPU, being designed by another team
            Message 5 of 22 , May 4, 2013
            • 0 Attachment
              It makes sense when you think about it from a project management point of
              view:

              1) board is designed and laid-out for 8-bit CPU, being designed by another
              team
              2) 8-bit CPU project is cancelled. Closest match is 9900 16-bit CPU
              3) Engineers say, "we have to re-layout the board"
              4) Project Management says "we don't have time on the schedule for that.
              Just glue it in."
              5) With much grumbling, that is what happens.

              Sometimes the engineers are encouraged to meet the schedule with the lie
              "we'll fix it in revision 2".

              Of course, it's an exercise for the viewer whether the "project management
              point of view" /ever/ makes sense when designing a new, hopefully long-life
              product.... ;)

              -----Original Message-----
              From: Dano
              Sent: Saturday, May 04, 2013 11:50 AM
              To: ti99-4a@yahoogroups.com
              Subject: Re: [TI-99/4A] Using a 16 Bit CPU with a 8 Bit Bus



              One phrase comes to mind immediately....

              "You can lead a horse to water but, you can't make him drink."

              It is truley sad that some thick skulls thwarted the TI... Unfortunately, I
              see it all the time in the IT world and this is why I chose to exit IT, and
              IF I return, it will be under my terms not some thick skulled employer.

              -Dano

              "Long Live the TI-99/4A!"


              On Sat, May 4, 2013 at 11:29 AM, Eric <eric-bray@...>
              wrote:



              Sometimes I get really frustrated by the actions of the head people at TEXAS
              INSTRUMENTS at how much they purposely crippled the TI-99/4A!

              With Win994a the user has the choice of using an 8 Bit bus to run the
              computer like TI supplied to the consumer or using a 16 Bit bus to match the
              CPU's actual architecture.

              A very PRIME example is the running of the conversion of the Extended Basic
              graphics and music program "Axel F" by Robert Gagle over to assembly by Mr.
              Jones.

              With the standard TI-99/4A P.E.B. 8 Bit bus, the graphics appear on the
              screen at the 'normal' refreshing rate and the music runs at the intended
              tempo.

              With a 16 Bit bus provided for the CPU's data to be funneled through, the
              graphics are now refreshed at a much quicker rate and the music now is
              played a a much, much faster tempo that makes it almost distorted in its
              presentation.

              QUESTION: Why did Texas Instruments cripple the TI-99/4A so much in its
              hardware architecture?

              With leadership like that it is no wonder the corporation was doomed to
              failure in the Home Computer Market!
            • Gregg Eshelman
              ... Because of cost. RAM for a 16 bit bus was extremely expensive then. There would need to be more traces on the circuit board. Larger connectors for many
              Message 6 of 22 , May 4, 2013
              • 0 Attachment
                --- On Sat, 5/4/13, Eric <eric-bray@...> wrote:

                > QUESTION: Why did Texas Instruments cripple the TI-99/4A so
                > much in its hardware architecture?

                Because of cost. RAM for a 16 bit bus was extremely expensive then. There would need to be more traces on the circuit board. Larger connectors for many things too. The entire computer would've had to increase in cost and complexity.

                Just a few years later, the 16 bit IBM AT started at over $4,000.

                The other part of the reason was TI's attempt at an 8 bit CPU was a failure, so they took the 9900 minicomputer CPU and shoehorned it into their 8 bit design that was getting behind schedule.

                The really rotten and completely avoidable hobbling was double interpreting of console BASIC from BASIC code to GPL to 9900 code which made it extremely slow, slower than the console BASIC of most other microcomputers, despite the TI's faster CPU.
              • Matias Ruiz
                To understand why the TI engineers took the decisions they finally took I recommend you a book based on real facts:     The Soul of a New Machine by Tracy
                Message 7 of 22 , May 4, 2013
                • 0 Attachment
                  To understand why the TI engineers took the decisions they finally took I recommend you a book based on real facts:
                   
                   
                  The Soul of a New Machine by Tracy kidder (winner of a pulitzer prize)
                   
                   
                  "recounts the feverish efforts of a team of Data General researchers to create a new 32-bit superminicomputer. "
                   
                   
                  --- El sáb 4-may-13, Tursi <yahoogroup@...> escribió:

                  De: Tursi <yahoogroup@...>
                  Asunto: Re: [TI-99/4A] Using a 16 Bit CPU with a 8 Bit Bus
                  Para: ti99-4a@yahoogroups.com
                  Fecha: sábado, 4 de mayo de 2013, 18:59

                  It makes sense when you think about it from a project management point of
                  view:

                  1) board is designed and laid-out for 8-bit CPU, being designed by another
                  team
                  2) 8-bit CPU project is cancelled. Closest match is 9900 16-bit CPU
                  3) Engineers say, "we have to re-layout the board"
                  4) Project Management says "we don't have time on the schedule for that.
                  Just glue it in."
                  5) With much grumbling, that is what happens.

                  Sometimes the engineers are encouraged to meet the schedule with the lie
                  "we'll fix it in revision 2".

                  Of course, it's an exercise for the viewer whether the "project management
                  point of view" /ever/ makes sense when designing a new, hopefully long-life
                  product.... ;)

                  -----Original Message-----
                  From: Dano
                  Sent: Saturday, May 04, 2013 11:50 AM
                  To: ti99-4a@yahoogroups.com
                  Subject: Re: [TI-99/4A] Using a 16 Bit CPU with a 8 Bit Bus



                  One phrase comes to mind immediately....

                  "You can lead a horse to water but, you can't make him drink."

                  It is truley sad that some thick skulls thwarted the TI...  Unfortunately, I
                  see it all the time in the IT world and this is why I chose to exit IT, and
                  IF I return, it will be under my terms not some thick skulled employer.

                  -Dano

                  "Long Live the TI-99/4A!"


                  On Sat, May 4, 2013 at 11:29 AM, Eric <eric-bray@...>
                  wrote:



                  Sometimes I get really frustrated by the actions of the head people at TEXAS
                  INSTRUMENTS at how much they purposely crippled the TI-99/4A!

                  With Win994a the user has the choice of using an 8 Bit bus to run the
                  computer like TI supplied to the consumer or using a 16 Bit bus to match the
                  CPU's actual architecture.

                  A very PRIME example is the running of the conversion of the Extended Basic
                  graphics and music program "Axel F" by Robert Gagle over to assembly by Mr.
                  Jones.

                  With the standard TI-99/4A P.E.B. 8 Bit bus, the graphics appear on the
                  screen at the 'normal' refreshing rate and the music runs at the intended
                  tempo.

                  With a 16 Bit bus provided for the CPU's data to be funneled through, the
                  graphics are now refreshed at a much quicker rate and the music now is
                  played a a much, much faster tempo that makes it almost distorted in its
                  presentation.

                  QUESTION: Why did Texas Instruments cripple the TI-99/4A so much in its
                  hardware architecture?

                  With leadership like that it is no wonder the corporation was doomed to
                  failure in the Home Computer Market!













                  ------------------------------------

                  For users/owners of TI-99/4A Geneve 9640 computers everywhere!
                  Visit the TI99'ers Hall of Fame at http://www.ti99hof.org/.
                  Check out the TI99ers On-Line User Group at http://www.ti99ers.org/.
                  Send abuse reports to abuse@...! Groups Links

                  <*> To visit your group on the web, go to:
                      http://groups.yahoo.com/group/ti99-4a/

                  <*> Your email settings:
                      Individual Email | Traditional

                  <*> To change settings online go to:
                      http://groups.yahoo.com/group/ti99-4a/join
                      (Yahoo! ID required)

                  <*> To change settings via email:
                      ti99-4a-digest@yahoogroups.com
                      ti99-4a-fullfeatured@yahoogroups.com

                  <*> To unsubscribe from this group, send an email to:
                      ti99-4a-unsubscribe@yahoogroups.com

                  <*> Your use of Yahoo! Groups is subject to:
                      http://docs.yahoo.com/info/terms/

                • Thierry
                  ... Actually, there are a few cases when this is not true. For instance, the LI (load immediate) instruction does not perform read-before-write. Nor do BL and
                  Message 8 of 22 , May 5, 2013
                  • 0 Attachment
                    >This is called "read-before-write", and in the TMS9900, it occurs on every write operation (even when the complete word will be written).

                    Actually, there are a few cases when this is not true. For instance, the LI (load immediate) instruction does not perform read-before-write.
                    Nor do BL and BLWP when saving values in dedicated registers (R11, R13-R15): these registers don't get read before writing.

                    I guess it was just too complicated to write dedicated microcode for operations like MOV, CLR and SETO, which don't require knowing the current value, from ABS, NEG, INC, DEC, A, S, etc, which must know the current value to alter it.

                    This is good to know if you are attempting to program an EEPROM that has an unlocking code. You can't use MOV or MOVB because the read-before-write may disturb the unlocking sequence. But if you place the workspace at the right address, you *could* use a series of LI to write the unlocking code. Not that I've ever tried, but it should work in theory...

                    Just my 2 cents,

                    Thierry
                  • Michael Zapf
                    Hi Thierry, ... Right, there are some, but in the majority of cases (and this comprises all these arithmetic and move operations) you have this
                    Message 9 of 22 , May 5, 2013
                    • 0 Attachment
                      Hi Thierry,

                      > Actually, there are a few cases when this is not true. For instance,
                      > the LI (load immediate) instruction does not perform read-before-
                      > write.
                      > Nor do BL and BLWP when saving values in dedicated registers (R11,
                      > R13-R15): these registers don't get read before writing.

                      Right, there are some, but in the majority of cases (and this comprises all these arithmetic and move operations) you have this Read-before-write, which largely contributes to the bad performance, which was my point in the last message.

                      There are groups of operations that share the same instruction sequences, so it is likely that they also share the same microprograms, parametrized by the respective command. One of the largest groups is the following:

                      (from the file "9900-FamilySystemsDesign-04-Hardware Design.pdf" (page 4-92) on whtech)

                      A, AB, C, CB, SOC, SOCB, SZC, SZCB, MOV, MOVB, COC, CZC, XOR

                      Cycle 1: Memory read (get instruction)
                      Cycle 2: ALU
                      Cycle 3..N: Data derivation sequence (source operand)
                      Cycle N+1: ALU
                      Cycle N+2..M: Data derivation sequence (destination operand)
                      Cycle M+1: ALU
                      Cycle M+2: Memory write (result)

                      The DDS in the 9900 is the same micro-subprogram for source and destination operand; in both cases, the operand address is determined from the command and the addressing mode (probably requiring to load another word from memory) and then the contents are loaded.

                      The Geneve's 9995 differs from that procedure by using an "address derivation sequence" which does not load the word from memory but only calculates its address. The 9980, however, also uses the 9900 DDS and therefore gets the same performance penalty.

                      As you correctly said, the read-before-write is only applicable for those commands that make use of the DDS, which the immediate operations (AI, ANDI, ORI, LI, CI) and all shifts, STWP, STST do not use, including the branch operations when saving values to registers.

                      Another funny thing revealed by the document is that B @XXX and BL @XXX actually load the contents from XXX, but discard them, because they just need the address where to branch to. This is another consequence of using the common data derivation sequence.

                      Michael
                    • Jeff White
                      ... The ill-fated 9985 was the processor meant for the 99/4. There is Bryan Roppolo s TI 99/4 Timeline page: http://www.ti994.com/timeline/ In fact, if you
                      Message 10 of 22 , May 5, 2013
                      • 0 Attachment
                        On 04 May 2013, at 16:07, "Michael Zapf" <forums12@...> wrote:
                         

                        As far as I know the design of the TI system was originally centered around 8-bit architectures, or 16-bit systems based on the 9980. The TMS 9980 is a 16-bit CPU with 8-bit data bus, but with a smaller address space (16 KiB). There must have been some last-minute change in the design of the TI-99/4 which led to the redesign for the TMS 9900.

                        The ill-fated 9985 was the processor meant for the 99/4.  There is Bryan Roppolo's TI 99/4 Timeline page:

                        http://www.ti994.com/timeline/

                        In fact, if you search the U.S. Patent and Trademark Office website for first term "Texas Instruments" and second term "9985" you will locate five patents all related to the 99/4.  Here is the link:

                        http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=0&f=S&l=50&TERM1=texas+instruments&FIELD1=&co1=AND&TERM2=9985&FIELD2=&d=PTXT

                        which results in the following (which make or may not format correctly):

                      • Jeff White
                        ... According to http://www.ti994.com/timeline/ the TI 99/4 was prototyped with the 9900 emulating the 9985. When the 9985 failed, the prototype design became
                        Message 11 of 22 , May 5, 2013
                        • 0 Attachment
                          On 04 May 2013, at 17:59, "Tursi" <yahoogroup@...> wrote:
                          > It makes sense when you think about it from a project management point of
                          > view:
                          >
                          > 1) board is designed and laid-out for 8-bit CPU, being designed by another
                          > team
                          > 2) 8-bit CPU project is cancelled. Closest match is 9900 16-bit CPU
                          > 3) Engineers say, "we have to re-layout the board"
                          > 4) Project Management says "we don't have time on the schedule for that.
                          > Just glue it in."
                          > 5) With much grumbling, that is what happens.
                          >
                          > Sometimes the engineers are encouraged to meet the schedule with the lie
                          > "we'll fix it in revision 2".

                          According to http://www.ti994.com/timeline/ the TI 99/4 was prototyped with the 9900 emulating the 9985. When the 9985 failed, the prototype design became part of the released version.

                          The 9995 was the fix for the 9985, but not in time for the 99/4A.

                          Budgeting and Marketing probably need to take the lead on blame.

                          Jeff White
                        • Jeff White
                          On 05 May 2013, at 05:42, Thierry wrote: ... This reminds me of the No-Slot Clock that I purchased several years ago for my
                          Message 12 of 22 , May 5, 2013
                          • 0 Attachment
                            On 05 May 2013, at 05:42, "Thierry" <nouspikel@...> wrote:
                            <snip>
                             

                            This is good to know if you are attempting to program an EEPROM that has an unlocking code. You can't use MOV or MOVB because the read-before-write may disturb the unlocking sequence. But if you place the workspace at the right address, you *could* use a series of LI to write the unlocking code. Not that I've ever tried, but it should work in theory...

                            This reminds me of the No-Slot Clock that I purchased several years ago for my Percom standalone disk controller and drive.  It fit underneath the 28-pin DSR EPROM using only 8 pins for operation and only CE did not pass-through.  It has 64-bit unlock sequence that happens on A0 and A2, or A15 and A13 of the TI bus.  You can read about it here:

                            ftp://ftp.apple.asimov.net/pub/apple_II//documentation/hardware/clocks/SMT%20No-Slot%20Clock%20Users%20Manual.pdf

                            Read-before-write is only one challenge in using the No-Slot Clock with a 99/4A.  Since the serial bit stream happens on A15 (TI nomenclature), the least significant address line, how do you get A15 to follow a defined "random" sequence of 1's and 0's?

                            Before giving an answer, I will let the reader think about it.

                            Jeff White

                          • urbite
                            ... The original product migration plan was to have a processor that executed GPL natively. This was early development of the 99/4, in the late 1970 s.
                            Message 13 of 22 , May 13, 2013
                            • 0 Attachment
                              --- In ti99-4a@yahoogroups.com, Gregg Eshelman <g_alan_e@...> wrote:
                              >
                              > --- On Sat, 5/4/13, Eric <eric-bray@...> wrote:
                              >
                              > > QUESTION: Why did Texas Instruments cripple the TI-99/4A so
                              > > much in its hardware architecture?
                              >
                              > Because of cost. RAM for a 16 bit bus was extremely expensive then. There would need to be more traces on the circuit board. Larger connectors for many things too. The entire computer would've had to increase in cost and complexity.
                              >
                              > Just a few years later, the 16 bit IBM AT started at over $4,000.
                              >
                              > The other part of the reason was TI's attempt at an 8 bit CPU was a failure, so they took the 9900 minicomputer CPU and shoehorned it into their 8 bit design that was getting behind schedule.
                              >
                              > The really rotten and completely avoidable hobbling was double interpreting of console BASIC from BASIC code to GPL to 9900 code which made it extremely slow, slower than the console BASIC of most other microcomputers, despite the TI's faster CPU.
                              >

                              The original product migration plan was to have a processor that executed GPL natively. This was early development of the 99/4, in the late 1970's.
                            • urbite
                              ... I don t know if the following story is true, but this is what I was told about the ill-fated 9985 when I was working in the Home Computer Group. The 9985
                              Message 14 of 22 , May 13, 2013
                              • 0 Attachment
                                --- In ti99-4a@yahoogroups.com, "Eric" <eric-bray@...> wrote:
                                >
                                > Sometimes I get really frustrated by the actions of the head people at TEXAS INSTRUMENTS at how much they purposely crippled the TI-99/4A!
                                >
                                > With Win994a the user has the choice of using an 8 Bit bus to run the computer like TI supplied to the consumer or using a 16 Bit bus to match the CPU's actual architecture.
                                >
                                > A very PRIME example is the running of the conversion of the Extended Basic graphics and music program "Axel F" by Robert Gagle over to assembly by Mr. Jones.
                                >
                                > With the standard TI-99/4A P.E.B. 8 Bit bus, the graphics appear on the screen at the 'normal' refreshing rate and the music runs at the intended tempo.
                                >
                                > With a 16 Bit bus provided for the CPU's data to be funneled through, the graphics are now refreshed at a much quicker rate and the music now is played a a much, much faster tempo that makes it almost distorted in its presentation.
                                >
                                > QUESTION: Why did Texas Instruments cripple the TI-99/4A so much in its hardware architecture?
                                >
                                > With leadership like that it is no wonder the corporation was doomed to failure in the Home Computer Market!
                                >

                                I don't know if the following story is true, but this is what I was told about the ill-fated 9985 when I was working in the Home Computer Group.
                                The 9985 was being designed in the processor design group in Houston by an engineer who was a savant of sorts. As with many savants, this designer was a bit 'different' - not your typical corporate employee. Apparently his manager tried to make him conform to the corporate norms, so he quit before he finished the 9985. When other engineers tried to pick up his work and finish it they couldn't make heads nor tails of his work, because of his unconventional design approach.
                                It wouldn't surprise me a bit if this is what actually happened.
                              • RedskullDC
                                Hi Urbite, et al. ... Could you elaborate a little on the Migration plan involving a Native GPL processor? Cheers, Leslie
                                Message 15 of 22 , May 21, 2013
                                • 0 Attachment
                                  Hi Urbite, et al.

                                  --- In ti99-4a@yahoogroups.com, "urbite" <urb@...> wrote:

                                  > The original product migration plan was to have a processor that executed GPL natively. This was early development of the 99/4, in the late 1970's.
                                  >

                                  Could you elaborate a little on the Migration plan involving a Native GPL processor?

                                  Cheers,
                                  Leslie
                                • Dano
                                  I agree, that sounds intriguing. Is there more to the story? And Urbite; did you have anything to do with parsec? -Dano Long Live the TI-99/4A!
                                  Message 16 of 22 , May 22, 2013
                                  • 0 Attachment
                                    I agree, that sounds intriguing.  Is there more to the story?

                                    And Urbite; did you have anything to do with parsec?

                                    -Dano

                                    "Long Live the TI-99/4A!"


                                    On Tue, May 21, 2013 at 6:06 AM, RedskullDC <layling@...> wrote:
                                     

                                    Hi Urbite, et al.

                                    --- In ti99-4a@yahoogroups.com, "urbite" <urb@...> wrote:

                                    > The original product migration plan was to have a processor that executed GPL natively. This was early development of the 99/4, in the late 1970's.
                                    >

                                    Could you elaborate a little on the Migration plan involving a Native GPL processor?

                                    Cheers,
                                    Leslie


                                  • Hank Mishkoff
                                    I don t know how often Paul Urbanus ( Urbite ) checks in here, so I ll answer for him: Paul was the creator of Parsec.
                                    Message 17 of 22 , May 22, 2013
                                    • 0 Attachment
                                      I don't know how often Paul Urbanus ("Urbite") checks in here, so I'll answer for him: Paul was the creator of Parsec.
                                    • urbite
                                      As I cooperative education student I worked for Dr. Granville Ott, who was one of the architects of the 99/4. I asked him why TI BASIC was written in an
                                      Message 18 of 22 , May 28, 2013
                                      • 0 Attachment
                                        As I cooperative education student I worked for Dr. Granville Ott, who was one of the architects of the 99/4. I asked him why TI BASIC was written in an interpreted language (GPL) instead of assembly, with a clear deleterious affect on performance. He told me that the plan was to have a custom processor where GPL was the native assembly language. IIRC, this migration plan was very early in the evolution of the 99/4, in the late 1970's. At the time I worked for him, in 1982, that plan had been long abandoned.

                                        I wish I had more to tell, be that's all I can recollect.

                                        Urbite

                                        --- In ti99-4a@yahoogroups.com, "RedskullDC" <layling@...> wrote:
                                        >
                                        > Hi Urbite, et al.
                                        >
                                        > --- In ti99-4a@yahoogroups.com, "urbite" <urb@> wrote:
                                        >
                                        > > The original product migration plan was to have a processor that executed GPL natively. This was early development of the 99/4, in the late 1970's.
                                        > >
                                        >
                                        > Could you elaborate a little on the Migration plan involving a Native GPL processor?
                                        >
                                        > Cheers,
                                        > Leslie
                                        >
                                      • urbite
                                        A MAJOR correction to Hank s comment, which is only 49% true :-) (no offense, Hank) Parsec was a joint collaboration between Jim Dramis and myself. Of course,
                                        Message 19 of 22 , May 28, 2013
                                        • 0 Attachment
                                          A MAJOR correction to Hank's comment, which is only 49% true :-) (no offense, Hank)
                                          Parsec was a joint collaboration between Jim Dramis and myself. Of course, prior to Parsec Jim had created both Car Wars and Munch Man.

                                          There are three named attack characters in Parsec.
                                          Dramites - named after Jim Dramis
                                          Urbites - you've got that one figured out :-)
                                          Bynites - named after Don Bynum, the head of TI Home Computer at the time, who often conducted our department meetings with the enthusiasm of a revivalist preacher.

                                          Jim and I reasoned that if we named a character after the boss it would increase our chances of keeping our namesakes in the games. He'd either have to remove all the named characters, or leave ours in with his. We should have doubled down...

                                          That was one of the most fun 'jobs' I've ever had. Getting paid to write video games. Heck, we even got reimbursed (under the pretext of 'competitive research') for the quarters we spent on the coin arcades during our long Friday lunches.

                                          Ah, the memories....

                                          Urbite

                                          --- In ti99-4a@yahoogroups.com, Hank Mishkoff <hank@...> wrote:
                                          >
                                          > I don't know how often Paul Urbanus ("Urbite") checks in here, so I'll
                                          > answer for him: Paul was the creator of Parsec.
                                          >
                                        • Jeff White
                                          Hi, Paul. Granville Ott has his name on the 99/4(A) patents, and he seems to be living in Austin, TX. I wonder if he would grant an interview. At 77, I
                                          Message 20 of 22 , May 28, 2013
                                          • 0 Attachment
                                            Hi, Paul.

                                            Granville Ott has his name on the 99/4(A) patents, and he seems to be living in Austin, TX.  I wonder if he would grant an interview.  At 77, I suppose he is retired.  

                                            I suppose he might welcome contact from one of the 99/4A demi-gods -- you, that is, Paul Urbanus.  His coop student from 30-some years ago.

                                            While I have your attention, I understand the Urbites and Dramites were named for you and Jim Dramis.  Were the Bynites really named after Don Bynum, as videogamehouse and wikipedia suggest?  Seems plausible and nearly obvious, but there seems to be some lingering doubt you could clear up.

                                            Jeff White

                                            On 28 May 2013, at 22:48, "urbite" <urb@...> wrote:
                                             

                                            As I cooperative education student I worked for Dr. Granville Ott, who was one of the architects of the 99/4. I asked him why TI BASIC was written in an interpreted language (GPL) instead of assembly, with a clear deleterious affect on performance. He told me that the plan was to have a custom processor where GPL was the native assembly language. IIRC, this migration plan was very early in the evolution of the 99/4, in the late 1970's. At the time I worked for ndethim, in 1982, that plan had been long abandoned.

                                            I wish I had more to tell, be that's all I can recollect.

                                            Urbite._,___

                                          • Jeff White
                                            There, you answered it in a crosspost. Now http://en.wikipedia.org/wiki/Parsec_(video_game) can be corrected. Jeff White
                                            Message 21 of 22 , May 28, 2013
                                            • 0 Attachment
                                              There, you answered it in a crosspost.  Now http://en.wikipedia.org/wiki/Parsec_(video_game) can be corrected.

                                              Jeff White

                                              On 28 May 2013, at 23:02, "urbite" <urb@...> wrote:
                                               

                                              A MAJOR correction to Hank's comment, which is only 49% true :-) (no offense, Hank)
                                              Parsec was a joint collaboration between Jim Dramis and myself. Of course, prior to Parsec Jim had created both Car Wars and Munch Man.

                                              There are three named attack characters in Parsec.
                                              Dramites - named after Jim Dramis
                                              Urbites - you've got that one figured out :-)
                                              Bynites - named after Don Bynum, the head of TI Home Computer at the time, who often conducted our department meetings with the enthusiasm of a revivalist preacher.

                                              Jim and I reasoned that if we named a character after the boss it would increase our chances of keeping our namesakes in the games. He'd either have to remove all the named characters, or leave ours in with his. We should have doubled down...

                                              That was one of the most fun 'jobs' I've ever had. Getting paid to write video games. Heck, we even got reimbursed (under the pretext of 'competitive research') for the quarters we spent on the coin arcades during our long Friday lunches.

                                              Ah, the memories....

                                              Urbite

                                            • Bryan Roppolo Boulder
                                              I talked with Granville back in 2001. He still at the time had bits and pieces in his garage as well as documentation. I’m sure there’s some good stuff
                                              Message 22 of 22 , May 28, 2013
                                              • 0 Attachment

                                                I talked with Granville back in 2001. He still at the time had bits and pieces in his garage as well as documentation. I’m sure there’s some good stuff there.

                                                 

                                                From: ti99-4a@yahoogroups.com [mailto:ti99-4a@yahoogroups.com] On Behalf Of Jeff White
                                                Sent: Tuesday, May 28, 2013 11:29 PM
                                                To: ti99-4a@yahoogroups.com
                                                Subject: Re: [TI-99/4A] Re: Natvive GPL Processor? was: Using a 16 Bit CPU with a 8 Bit Bus

                                                 




                                                Hi, Paul.

                                                 

                                                Granville Ott has his name on the 99/4(A) patents, and he seems to be living in Austin, TX.  I wonder if he would grant an interview.  At 77, I suppose he is retired.  

                                                 

                                                I suppose he might welcome contact from one of the 99/4A demi-gods -- you, that is, Paul Urbanus.  His coop student from 30-some years ago.

                                                 

                                                While I have your attention, I understand the Urbites and Dramites were named for you and Jim Dramis.  Were the Bynites really named after Don Bynum, as videogamehouse and wikipedia suggest?  Seems plausible and nearly obvious, but there seems to be some lingering doubt you could clear up.

                                                 

                                                Jeff White


                                                On 28 May 2013, at 22:48, "urbite" <urb@...> wrote:

                                                 

                                                As I cooperative education student I worked for Dr. Granville Ott, who was one of the architects of the 99/4. I asked him why TI BASIC was written in an interpreted language (GPL) instead of assembly, with a clear deleterious affect on performance. He told me that the plan was to have a custom processor where GPL was the native assembly language. IIRC, this migration plan was very early in the evolution of the 99/4, in the late 1970's. At the time I worked for ndethim, in 1982, that plan had been long abandoned.

                                                I wish I had more to tell, be that's all I can recollect.

                                                Urbite._,___




                                              Your message has been successfully submitted and would be delivered to recipients shortly.