V9978... what could it have been?

Page 2/4
1 | | 3 | 4

By PingPong

Prophet (3436)

PingPong's picture

19-05-2019, 16:39

would be nice to know the opinion of someone with FPGA expertise about this...

By iamweasel2

Hero (524)

iamweasel2's picture

19-05-2019, 18:27

Wouldn't the easiest solution be a v9958 in paralell with V9990 (each one in his port) with a superimpose between the two to have a single output in the machine? This way you as a programmer could have the option to use only V9958, only V9990 or both at the same time.

That way you wouldn't have to worry about retrocompatibility in the new V9978, and I always thought why ASCII didn't go to this solution when designing the Turbo-R. Ok, you would have the cost of the extra VRAMS (besides having both V9990 and V9958 in the machine), but it would had allowed ASCII to search for the best solution for the new VDP among all the options available in the market, that could have been in the end cheaper and could provide the best result.

By Manuel

Ascended (15694)

Manuel's picture

19-05-2019, 23:43

So, what I learned so far from your reactions:
1. V9978 wouldn't have been as powerful as the V9990, because of all the trouble of keeping it backwards compatible. Perhaps it was even the reason why it got delayed/canceled?
2. Ideally, one could imagine the V9978 now as simply a V9958 and a V9990 in one chip, with some register to select which layer (V9958's or V9990's) is visible (perhaps also a register to tell which layer is up front).

Of course we could only guess what the specs/details would have been of the original V9978... or is there still someone around in contact with the Yamaha or ASCII people who were involved? Smile

By enribar

Paladin (996)

enribar's picture

20-05-2019, 01:03

All of us, we have a game-centered point of view of the MSX.
But consoles are excellent game machines, and MSX is an excellent computer platform capable of great games.
Scroll and other features must be related to "office" or "multimedia" purposes, and only to a second step viewed as useful to game programmers.
Nishi explained his project with the v9978 into MSX Magazine 2 (revival) showing a handmade schema: the MSX3 with the v9978 on board was capable of telecommunications (videocalls) and multimedia (cd-rom support). In this vision, all v9958 screens and scrolls are needed, but more advanced sprites and loosing backward compatibility are not interesting matters.
So check that magazine, the v9978 was intended for a more powerful computer or more obiquitous computing rather than to make a game machine.

By RetroTechie

Paragon (1563)

RetroTechie's picture

20-05-2019, 01:30

Superimpose requires that both VDP's are synchronized. That's near impossible with totally different chip, that each have their own screen resolutions, sync rates etc. Like a VGA compatible mode on the V9990, how would you 'align' that with a PAL or NTSC mode from V9958? Forget it.

I could imagine some 'extended' V9958. But what's the point? If a V9958 is what you need, then using an uber-V9958 would shrink your target audience to <1%. If a V9990 is what you need, you want the real thing & all of it. Only exception is extra's that aren't necessary for things to work. Like remove # sprites / scanline limit to reduce sprite flicker, speed up blitter etc. Trying to combine the two is like trying to roll an ARM and x86 cpu into one. Evil Impossible, or the franken-result woudn't be compatible with either of them.

So switchable VDP makes most sense. Perhaps share VRAM or video output. Or side-by-side so you can do a dual screen setup.

Manuel wrote:

Of course we could only guess what the specs/details would have been of the original V9978...

Most likely "V9978" was a working title of sorts. And then while drawing up specs, @ some point compatibility with V9958 was dropped (see previous comments), and name changed to V9990 to avoid the suggestion of 9938->9958->9978 compatibility. Imagining a V9978 as an actual chip that was on the design table once, is giving that title "V9978" more weight than it deserves imho. Whatever it was or wasn't, on paper or in the heads of designers, it never made it to silicon. End of story. And the V9990 which did, is a totally different beast.

Btw Wikipedia says Yamaha re-targeted the V9990 with some success towards VGA card market. Anyone ever seen a VGA card with this chip? Question

By gdx

Prophet (2979)

gdx's picture

20-05-2019, 02:19

PingPong wrote:
Quote:

The first problem is they conflict with the ports 64h and 65h of Panasonic MSX2+ with the turbo mode.

Interesting, what do you expect will happen if one try to use V9990 with panasonic MSX2+ ?

In fact this is the ports 40h and 41h that are used by Panasonic MSX2+ for the turbo mode. So no conflict. Smile

By NYYRIKKI

Enlighted (5366)

NYYRIKKI's picture

20-05-2019, 12:14

Manuel wrote:

So, what I learned so far from your reactions:
2. Ideally, one could imagine the V9978 now as simply a V9958 and a V9990 in one chip, with some register to select which layer (V9958's or V9990's) is visible (perhaps also a register to tell which layer is up front).

Actually this is quite likely guess... If we look what V9990 became it seems that they were not trying very hard to make it strictly V9958 compatible. Also I think the project was already quite far when they decided to do this emergency change to scrap V9958 compatibility. If it would have been hard to rip off V9958 functionality this kind of decision would have not made sense. If they were already pretty much separated then this was probably quite easy decision to make to reduce manufacturing costs. It makes sense also if you think about why to do another VDP that tries to be V9958 compatible, if you can just copy V9958 there pretty much as is. I think the part that was most incomplete when they did run out of time was the DMA functionality that would have been likely pretty important to realize these planned video calls and stuff.

By PingPong

Prophet (3436)

PingPong's picture

20-05-2019, 13:19

Quote:

So, what I learned so far from your reactions:
1. V9978 wouldn't have been as powerful as the V9990, because of all the trouble of keeping it backwards compatible. Perhaps it was even the reason why it got delayed/canceled?

I agree, expecially in those times, it was hard. Part of the power of V9990 comes from DUAL PORT RAM. This is not so compatibile with the old tricks and 8 bit bus access used on V9958. Of course teoretically they could have created a memory access layer almost transparent to upper layers, but is is hard to keep compatibility with timing issues.

About the reason of being cancelled: if the solution was simply two chips on one chips i cannot see the reason of drop, otherwise the number of transistors used and die size. they already knew how to make a v9958 or V9990...
To me the reason was to make a "merge" between the two to keep costs down and ensure compatibility......

Quote:

2. Ideally, one could imagine the V9978 now as simply a V9958 and a V9990 in one chip, with some register to select which layer (V9958's or V9990's) is visible (perhaps also a register to tell which layer is up front).
.....
or is there still someone around in contact with the Yamaha or ASCII people who were involved? Smile

i always wondering if there is some one like karl guttag to ask for V9938... if anyone know should be great. Maybe asking Nishi?

By DanySoft

Champion (432)

DanySoft's picture

22-05-2019, 00:05

Perhaps there is only one problem:
v9978 is identical to v9958, 32 sprite, max 8 on line.
So they discarded this project to switch to the new VDP, obviously v9990!
It's true ? Or I was wrong.
I thought it was the VDU project, as a "successor" of v9958 ?? VDU project <=

No comment.
Bye bye
DanySoft

By Grauw

Ascended (8395)

Grauw's picture

22-05-2019, 01:00

Relevant, and I don’t think linked here yet:

http://map.grauw.nl/resources/video/yamaha_v9978.pdf

Specs are clearly the V9990’s, as is the name E-VDP-III.

RetroTechie wrote:

Superimpose requires that both VDP's are synchronized. That's near impossible with totally different chip, that each have their own screen resolutions, sync rates etc. Like a VGA compatible mode on the V9990, how would you 'align' that with a PAL or NTSC mode from V9958? Forget it.

The VGA modes wouldn’t genlock, but those modes also don’t work with a regular TV (requires a dual sync monitor like iirc the X68000 does). The other modes can superimpose with the V9958 fine, the sync rates are the same (15 kHz horizontal, 50/60 Hz vertical), thus the vertical resolution is too, and the horizontal resolution (dot clock) doesn’t matter.

I too think both V9958 and V9990 combined on a single chip would’ve been what they were going for. The V9990 is clearly not designed to be V9958-compatible (just similar in spirit), so the logical backwards compatible solution is to include the V9958 verbatim and overlay them through the superimpose mechanism. This approach then allowed Yamaha to design the new VDP capabilities unrestrained by legacy considerations (unlike the V9938 which was encumbered by the TMS9918 design legacy).

Page 2/4
1 | | 3 | 4