TMS9918 designer

By karloch

Prophet (2059)

karloch's picture

19-05-2018, 02:30

I found myself by chance in a interesting 8 months old Reddit thread of one of the 6 engineers that worked designing the TMS9918 architecture. Many mentions to the MSX and even some to V9938, so I though it would be interesting to share it here.

Quote:

I was one of the original designers of the TMS9918 used in Colecovision, the TI home computer and the MSX Computers. Thanks to the use in the MSX, both the NES and Sega systems used a register compatible upgraded clones ("Z80 Like) of the 9918. The 9918 was the first consumer chip to directly interface to DRAMs.

The 9918 was my first design at Texas Instruments and I was one of only 6 Engineers that designed the whole 9918. My main logic work was on the Sprite design and control. Wally Rhines mentioned me in his recent IEEE article on the 9900 Family and the home Computer. I commented on the article in my blog (the blog mostly discussing display devices, particularly those used in AR and VR) http://www.kguttag.com/tag/9918/

Back a number of years ago, I gave a bunch of documents related to the 9918xx to someone that archived them: http://spatula-city.org/~im14u2c/vdp-99xx/

Quote:

The 9918 had a generic I/O interface and could be connected to any CPU (Z80 and 6502 were both used with the 9918). Ascii Microsoft came out with the V9938 what was developed by Yamaha and my understanding is a slightly customized version of this chip when into the NES game systems.

What I meant by "Z80 like" was that the Nintendo PPU was a "register compatible superset" of the 9918 (like the Z80 was a register compatible superset of the Intel 8080). The 9918 was designed in 1977 and by 1981/82 you could put a lot more on a chip. The main thing they improved was the background modes and the number of Sprites. After all in 4+ years they could put a lot more on the chips and the DRAMs had gotten faster.

Compare the V9938 (http://bitsavers.informatik.uni-stuttgart.de/pdf/yamaha/Yama... ) to the TMS9918 (http://map.grauw.nl/resources/video/texasinstruments_tms9918...))

https://www.reddit.com/r/retrogaming/comments/6wosd7/tms9918_designer/

Login or register to post comments

By Thom

Hero (576)

Thom's picture

19-05-2018, 13:14

Thanks! Very interesting read, especially this quote:

Quote:

BTW, just about every limitation in the background graphics and Sprites could be traced back to DRAM cycle times. We could not afford the complexity and buffering of page mode in 1977, so we have very few cycles to work with and just about every cycle but a few were used (the few left were given to CPU access).

By mohai

Paladin (835)

mohai's picture

21-05-2018, 14:33

karloch]

[quote

wrote:

The 9918 had a generic I/O interface and could be connected to any CPU (Z80 and 6502 were both used with the 9918). Ascii Microsoft came out with the V9938 what was developed by Yamaha and my understanding is a slightly customized version of this chip when into the NES game systems.

Very interesting readings. It is always interesting to know how and why VDP chip was designed.
This paragraph is not right, as NES was marketed on 1983 and V9938 was not available then. I suppose that Karl meant a TMS9918 customized version.
It is a curious coincidence that both Nintendo NES and Sega Master System both have a customised TMS9918 VDP on them. Difference is that NES, being designed a couple of years before, had some slightly lesser capabilities.