MRC. Please introduce yourself, your role within ASCII and specifically your involvement in any MSX activities or projects at ASCII.

TM. I was a MSX programmer at ASCII Corporation those days. For the sake of accuracy, I will write in chronological order.

  • Summer 1989:

    I was offered a job at ASCII Corporation. I was a university student those days. At that time in Japan, it was customary to find a job in the summer.

  • April 1990:

    I joined ASCII Corporation, where I was trained for about three months with only new employees. Around July, the new employees were temporarily assigned to a department that handled MSX.

  • October 1990:

    Formally assigned to a department that handled MSX systems. The MSX turbo R FS-A1ST was already released. In other words, I had nothing to do with the development of the MSX turbo R. I was checking the operation of the MIDI interface for the next MSX turbo R and adding MIDI code to the MSX-BASIC. The FS-A1GT, released in November 1991, was the last MSX, I became "last person to add code to the MSX".

  • November 1992:

    I moved to the publishing department. After this, I didn't do any programming as a job.

  • 1999:

    I left ASCII Corporation.


MRC. Were you a MSX user before joining ASCII? If so, what did this mean for you?

TM. Of course, I was a user. When the details of MSX were announced in 1983 in the magazine Monthly ASCII, I was very surprised, and I thought that a unified standard was going to become popular in the future. As you know, microcomputers at that time had different standards and no binary (program) compatibility. So MSX was revolutionary, and I felt it was fair that it wasn't manufacturer driven.

By this time I was writing programs in Basic. I didn't have a microcomputer of my own, so I used a microcomputer from the school's club. But I remember that it was also another computer that one of the members brought in. It was also around this time that I learned Z80 assembler from a book.

The first MSX I bought (and my parents bought me) was a Yamaha YIS 503II. This was in April 1986, when I entered college. I also bought the SFG-05, the YRM-52 and the YRM-55. I wanted to play computer music but I didn't know how to compose music, and I enjoyed typing some music by Bach and others.

The next year, 1986, I bought a Yamaha YIS 805/128 with the money I earned from my part-time job. I wanted to buy the YIS 805/256, but they were no longer in stock.

In those days, Japanese universities were still using large computers. Magnetic tape still existed! In class I learned FORTRAN, COBOL, Pascal and Lisp. In the world, "the time of C was coming!". I used to write programs for Mandelbrot sets and life games, and it took a lot of time to do the math on MSX, and I could see the limitations of 8-bit computers.

And since this was the case, it was no surprise that I considered ASCII as a place of employment. Computer manufacturers seemed big and old-fashioned. ASCII looked like a company that could do something new and exciting.

At that time, Japanese manufacturers were large global corporations, so many people must have wondered what it would be like to join a small company.

MRC. What was the atmosphere in the MSX related teams at ASCII at that time?

TM.Everyone was programming. :-) No, not everyone because not all of us were programmers. Well, it was a department of programmers that still exists. There is no doubt that the atmosphere was quite liberal for a Japanese company at that time.

I remember the first time I called my boss Mr. Suzuki... he immediately told me to call him "Jay". :-)
Most people in Japan are obsessed with titles, so this was a surprise to me. But of course I obeyed him. Because that's what everyone around me called him. (Mr. Suzuki was about 30 years old then. Most of the development team members were in their 20's.)

Suzuki Hitoshi is the developer of MSX-BASIC and is the person who knows the MSX system software best. He is a top-notch programmer. You can find more information here.

The name "Jay" was his nickname when he was at Microsoft in Seattle. Japanese names were hard to pronounce, so they were called by nicknames. All the members in Seattle at that time had nicknames. He later founded Ubiquitous Corporation in 2001 (now Ubiquitous AI Corporation). More relevant information [1, 2].

MRC. Was the working environment open to suggestions/feedback on the whole platform, or was it more like a top-down decision structure?

TM. I was a newcomer, so I didn't have much of an opinion on the subject. They were all MSX veterans, so I didn't have anything to say. But they gave the BASIC coding to a rookie, so it looked "top-down" to me, but I guess that's because I was a rookie. The details were left up to each of us. Of course there was coordination across the board.

MRC. What was the ASCII organization like at that time? We know about ASCII hardware, software and books/magazines. How was the collaboration regarding MSX between these departments?

TM. We were also focused on computer communications. The Internet was not generally known at that time.

I'm not sure about the cooperation between the departments either. I don't think most people from the publishing department came to my department, but maybe I just didn't see it.

One time I went to the editorial office of MSX Magazine to deliver some material or something. However, there was almost no one there. When I asked them about it, they told me that it was almost empty in the morning and that they would come in the afternoon. That was the common practice in the editorial department. :-) So the development team was working seriously from the morning, wasn't it?

MRC. What was the roadmap for the MSX turbo R and beyond if the ST and GT would have been successful enough? Were there any future types of MSX turbo R in mind and what would have been their specs? Has the MSX3
generation concept ever existed at ASCII?

TM. I'm not sure. Times were moving from 8-bit to 16-bit and I'm sure MSX would have needed to go 16-bit. However, it seems that there was no plan for a specific product. At that time, NEC's PC-9800 series was overwhelmingly strong in Japan, and it was thought to be quite difficult to break its power.

MRC. Were there specific sinergies between ASCII and the main MSX manufacturer companies such as Sony, Panasonic, Sanyo, etc., during a new MSX computer or hardware development?

TM. I'm not sure. I was still a student those days. (I can hardly answer any of the following questions about development.)

MRC. What collaborations were there with ASCII and other software makers. Were some ASCII titles made in collaboration with other software developers?

TM. When I was on the development team, it was around the time we launched the MSX-View. This was an improved version of HALNOTE. However, when I learned about it, development was completed. I was given the task of creating a sample file. See attached file. Images are from this video.

I painted "Guilin". Why Guilin? It's a black and white image with few pixels. The tonal expression is also limited. If I had to draw under these unfavorable conditions, I thought that an ink painting would be better. That's what I thought when I drew it. Guilin, Karst Landscapes, in Japanese, it is pronounced "Keirin".

MRC. What was the driving factor for developing the MSX turbo R in a market that already started to be dominated by 16 bit and 32 bit CPUs? What was the intended target market for this machine?

TM. It was supposed to be for hobby users, but at that time there were already Super Nintendo and Game Boy, so it was clear that we would have a hard time. For most people, playing games was enough, and only a few people bothered to study programming.

MRC. ASCII also developed software for other platforms. How was collaboration with those teams and MSX teams or was there a friendly rivalry?

TM. After MSX, the team developed the system software for Omron's handheld computer, Massif. It was based on DR-DOS. We put Japanese language software on it so that it could be used in Japanese. I designed the ASCII font. I guess an original design was necessary because of copyright issues.


MRC. What kind of specific hardware and software was used for a MSX development? What kind of cross platform tools were used, if any?

TM. There was a MSX (test machine, product that was released) to check the operation. I mainly used the NEC PC-9800 series. I used it for programming and e-mail. I also used it as a UNIX terminal. ASCII was able to connect to JUNET (a computer network for research in Japan). We used EPROMs (programmable ROMs that erase data with ultraviolet light) to check the operation of the MSX BIOS/BASIC. We wrote binary data to the EPROMs and plugged them into the test machine.

MRC. Were all developments made in assembly at the time or were some higher-level languages already used?

TM. Z80 assembly language. High-level languages are inefficient in the limited memory space of an 8-bit computer. Using the most efficient assembly language is the obvious choice.

MRC. The initial specs of the MSX3 mentioned the Zilog Z280 as the CPU. Later, ASCII decided to switch to the R800. But the Z280 went into production in 1987, so 3 years before the R800. Why did they chose to use to develop the R800, instead of focusing all resources on the V9978?

TM. I'm not sure. However, it was considered very important to show that Japan could develop its own CPUs.
We knew at the time that CPUs would have a huge impact on the personal computer landscape. Intel's influence was already significant at that time. It made a lot of sense for a small company like ASCII to develop a CPU.

MRC. On the R800 the Z80 is still there. Do you know why ASCII didn't switch to a more powerful CPU like the Motorola 68000 and let the z80 act like a co-processor or for compatibility?

TM. I'm not sure. However, compatible chips often don't fully replicate the original. My guess is that they had that kind of unresolvable problem at this time as well. The best way to get full compatibility is to put the Z80 itself on it.

MRC. The MSX turbo R BIOS looks kinda like a hack, almost like a patch. Had they lost the original source code at that point?

TM. I'm not sure. The source code must have been there. If the BIOS in this question refers to the MSX-BASIC MIDI extension, then of course the source code was there. I just added code to it, so it's not surprising that it looks like a patch.


MRC. Did you know about the Aucnet NIA-2001 computer when you worked for ASCII? It was a hidden MSX turbo R computer without the PCM hardware built-in especially developed for online car auctions in Japan by Aucnet company.

TM. Takaoka Seisakusho. This is a machine developed by Takaoka Seisakusho. And I remember doing some small work on it, but I don't remember what it was. What was it? The company's current name is Takaoka Toko Co., Ltd.

MRC. There being several and more powerful platforms coexisting with the MSX in Japan, in your opinion what were the pros and cons of the MSX computers over its competitors?

TM. The pros of the MSX are that it was compatible with different manufacturers and it was cheap. The cons were slow processing speed and less business software. The image display on TVs was not as fine as it could be.

MRC. In retrospect, what is your opinion about the MSX system in general?

TM. I think it was a great achievement to come up with a "unified standard". It has reduced the burden of entry for both hardware and software. Not all PC manufacturers were happy, but software and peripheral development benefited greatly from lower costs. The same thing was happening in the United States at that time, with the proliferation of IBM PC compatibles.

MRC. Related to market sales, do you know what position ranked the MSX normally over other computer platforms?

TM. I'm not sure. I was'nt really interested in that kind of thing. But the market share of the PC-9800 was quite high, and everyone thought it would be difficult to break that market share. The IBM PC was not yet widespread in Japan. The last manufacturer to produce the MSX was only Matsushita Electric Industrial, and this meant that the market could not be expected to expand. After Windows 95, the momentum of the PC-9800 waned.

MRC. Do you have any good backstage story that would like to share?


  • MIDI Cartridge:

    This was during the development of the Panasonic FS-A1GT. One day I received a prototype of an external MIDI interface cartridge, the same as μ-PACK. There was no case and the board was bare. I was assigned to test its operation. However, for some reason it did not work at all. It's not that my test program was wrong. I went to the hardware development department of the company (it was right next door). I showed them the schematic, and they immediately pointed out to me that I should cut this line.

    I cut the line on the board, and it worked fine. I suppose they had to wire it separately to check for hardware fabrication, but that information was not passed on to us, which caused me a lot of trouble.

    During this operation test, I also used an oscilloscope to verify that the waveform of the signal output from the MIDI OUT jack was correct. Of course, the waveform was correct, so I use an oscilloscope for a short period of time.

    I think we used the Roland CM-64 sound module for operational testing. We also had a Yamaha DX7. I also remember that I made a standard MIDI file player that ran on MSX-DOS.

  • Z80 Test:

    One day, Jay gave me three Z80 machine language questions. It was to perform an action in a specified number of bytes. I don't remember what it was, but it might have been like a puzzle. I could only solve one of them. The other programmers on the development team seemed to be able to solve all of them.

    I was a good programmer, but the members of the development team were even better. It occurred to me after many years that it was not necessarily good to work with people who were too good at what they did.

  • A bug in the play statement:

    Few people seemed to notice it, but there was a bug in BASIC play statements in MSX1-2+. It caused an extra space for one count when a play command ended. In other words, there was a short pause at the moment of the transition from one play sentence to the next. The tempo shifts for a moment, but many of you may not have noticed it. I was noticing this for quite some time now, and I was frustrated for a long time, not knowing if this was a spec or a bug. (One count is the minimum note length of a play sentence, which is a 96th note).

    When I added the MIDI function to BASIC, I carefully read the music performance part of the BASIC source, and I confirmed that the above phenomenon was indeed happening. I knew it was a bug in BASIC. I fixed this bug at this time. So, in the Panasonic FS-A1GT (MSX-BASIC Ver4.1?) the behavior of the play sentence is slightly different from the previous model.

    I told my bosses (Jay-san and others) about this and the bug, but they said "Oh, really? So it must have been a really unnoticed bug, because the response was something like, "I'm not sure".

MRC. Finally, would you like to send a message to MSX users worldwide?

TM. It's important to keep records. If someone doesn't make a conscious effort to record it, it won't go down in history. The work of the MSX Resource Center is a wonderful record of history. I hope that the people involved in MSX in Japan will also keep a record. I encourage all MSX users around the world to write down their own MSX history. I am also interested in where you learned about MSX, where you bought MSX, what software you used, etc.

Well, I have a job that has nothing to do with computers at the moment. I am also an amateur researcher of tanuki (Racoon dog) and Abura-koumori (Japanese house bats). Out of all the development teams at the time, I'd say I've gone the farthest from MSX. Today, computers and the Internet have become commonplace in our daily lives, and in the age of MSX, it was a dream come true, and in my opinion, MSX was ahead of its time.

There is only a website in Japanese about the study of tanuki and Abura-koumori, but if you are interested, please visit it.

Tokyo Tanuki Tankentai! (Tokyo Tanuki Explorers!)
Tokyo Koumori Tankentai! (Tokyo Bats Explorers!)

Translation: kimoan

Comments (10)

By TheKid

Paragon (1238)

TheKid's picture

06-03-2021, 08:02

Congrats,very interesting interview. Thank you mrc team and thank you Takumi Miyamoto. And t is true, when people record their history, certainly when they are passionate about it, it”s mostly very intersting to read. So I would encourage mrc to do mor interviews like this. It could be a reaccuring topic. Mrc announces who they are going to interview, members can send in their questions and we get an informing interview.

By Bengalack

Paladin (877)

Bengalack's picture

06-03-2021, 09:50

This is really nice. Thank you for bringing us these stories.

By warmize

Expert (74)

warmize's picture

06-03-2021, 10:59

Congrats for the Interview!
An interview with Jay Suzuki would be there any already made? Anyone has contact information?

By sd_snatcher

Prophet (3684)

sd_snatcher's picture

06-03-2021, 12:14

Thank you MRC team for the interview, and Miyamoto-san for dedicating his time to answer all these questions. Smile

By Argon

Paragon (1126)

Argon's picture

06-03-2021, 13:44

Didn't you ask what "MSX" stands for? :-D

By Takamichi

Paladin (667)

Takamichi's picture

06-03-2021, 14:59

If every PLAY statement really takes the length of a 96th note = 1/48 second,
FOR I=0 to 47:PLAY "":NEXT
should at least take a second with the default tempo 120, but it doesn't. I thought it might meant "1/96 second" and tried
FOR I=0 to 95:PLAY "":NEXT
The result is almost but not as long as 1 second and it doesn't vary between FS-A1ST and GT (emulators). Puzzlingly, the time increases by attempting repeatedly like 1st try: instantaneous 2nd: 1/3 second 3rd: 1/2 second.

By tfh

Prophet (3487)

tfh's picture

06-03-2021, 15:33

The for/next loop will also take time.

By Parn

Paladin (864)

Parn's picture

06-03-2021, 16:23

I loved the interview. Very interesting and thought-provoking. Thanks to all involved.

About the FM-BIOS:

Takamichi wrote:

The result is almost but not as long as 1 second and it doesn't vary between FS-A1ST and GT (emulators). Puzzlingly, the time increases by attempting repeatedly like 1st try: instantaneous 2nd: 1/3 second 3rd: 1/2 second.

Try doing that after CALL MUSIC, maybe he only optimized the FM-BIOS routine. I also wonder whether any of this would make any difference when there is an empty string in the PLAY command since nothing would be sent to the music buffer. Perhaps you should try it with a sequence of notes that would surely fill the buffer. Or send a few notes to each channel in order to quickly fill the queues.

By Gloriou

Master (204)

Gloriou's picture

07-03-2021, 18:37

Interesting Story,
Thanks to Mr Miyamoto & the MRC team for this nice interview!

This makes me suggest having MSX Stories section in this forum, would me awesome history sharing to know more history about our beloved machine Smile Big smile

By Pac

Scribe (7173)

Pac's picture

07-03-2021, 20:00

Don't forget Kimoan, his help was crucial! Wink