POLL: AVDP from texas

페이지 2/4
1 | | 3 | 4

By PingPong

Prophet (3529)

PingPong의 아바타

04-09-2019, 10:17

First one should find more docs. And i'm not sure it's an easy task.

By Parn

Hero (565)

Parn의 아바타

04-09-2019, 22:11

Grauw wrote:

Interesting document nevertheless Smile. I wonder when this AVDP was being designed. Before or after they got wind of Yamaha developing the V9938 and the SMS VDP.

I would guess earlier than Yamaha's VDPs, since the 9918 was already available in 1979, and Yamaha's VDPs were probably ready in 1984, since both MSX2 and Master System were out in 1985. Also, it only addresses 64kB, against V9938's 128kB. But it's just a guess, of course.

jltursan wrote:

And thinking out loud, ¿can it be recreated with a FPGA?, ¿or maybe software emulated?

It could be fun to try to fill the gaps "recreating" a non existent hardware Big smile

I was talking about exactly this to a friend yesterday. It would be fascinating to get our hands on more complete documentation to see what they had in mind. I think we would be better off plugging certain frustrating limitations of current VDPs, like text modes with independent color attributes for each character, more sprites per line (and maybe supporting different logic operations for extra flexibility!), sprite flipping, a 512-pixel horizontal resolution pattern-based screen mode (which this TMS9928 would have) and maybe some pattern-based modes with less local color restrictions.

By Grauw

Ascended (9170)

Grauw의 아바타

04-09-2019, 22:30

You mean the V9990 Big smile

By jltursan

Prophet (2285)

jltursan의 아바타

05-09-2019, 08:48

If V9990 had been compatible with the old TMS9918, it would have been THE VDP

Maybe implementing it with a FPGA and adding the old modes? Smile. Mmmm, all MSX VDPs into one seems a huge beast, no one has tried it before...

By Grauw

Ascended (9170)

Grauw의 아바타

05-09-2019, 09:39

Just super impose it on the V9958. You got perfect compatibility. If you ask me that was the plan all along, maybe later integrating them on one chip. Not encumbering the V9990 with a legacy design like the V9938 was encumbered by the TMS9918.

By Parn

Hero (565)

Parn의 아바타

05-09-2019, 16:10

Now THAT would have been awesome.

By hit9918

Prophet (2895)

hit9918의 아바타

05-09-2019, 16:39

Quote:

(What a waste! MY GOD! i can hear Karl Guttag cry hard on this waste)

why do you rumble at the 9918 all day but never at the 9938. because under the hood of your rumbeling is this story that "the 9938 quirks are all the fault of 9918". and this myth is BUSTED in one second by pointing at the sega.

By PingPong

Prophet (3529)

PingPong의 아바타

05-09-2019, 19:54

Simple: the 9938 is what it is mainly because of the TMS9918. For example: look at the stupid sprite color by scanline. It also got the EARLY CLOCK per scanline, a NON SENSE!
Why it is done in this way? Because otherwise the V9938 had to redesign the entire sprite system. Maybe they have enough space on chip to accomodate a good sprite engine but probably not enough to accomodate the legacy one.
Look at the stupid magic Y coordinate. It's an invention of Yamaha or an invention of TMS?
Other systems like C64 have separate bit in register to enable single sprites (perhaps not needed). a clean solution
the dirty man (Karl Guttag) invented the magic Y value, again to save a register. With a dirty solution
look at the horrible sprite cloning effect ( that is somewhat related to the chip temperature) all those crappyness are from TMS / Karl Guttag.

When yamaha can design without the legacy TMS it does things right like on V9990. Clean and speedy .

And you ask me why i rumble with TMS? i think should be clear.
not to mention the stupid screen1 with 1 byte specifying 8 chars colors. Would be so inaceptable to have a 1 byte per charater?

TMS=CRAPPY

By PingPong

Prophet (3529)

PingPong의 아바타

05-09-2019, 20:16

sega has it's quirks too. But they have sacrificed some good v9938 features (command engine, larger vram, bitmap modes, etc) for a decent sprite engine (and this is mainly the difference) they also dropped the super stupid screen 2 way of interpreting vram data to get some color. using a more smart way to code pixels color. Just to remember you the advantages of screen 2 ;-) :

With 12 KB+768 bytes of VRAM (relatively a huge amount of VRAM) you get:

TWO COLORS per 8 px lines, pratically the same colour clash you see in spectrum that have a attribute table less than 1/8 bigger.

Take an example, you have a 8x1 line of entirely white color: you can code this in a lot of way in screen 2:

PATTERN BYTE COLOR BYTE
255 Fxh (16 different combinations)
0 xFh (other 16 different combinations)

No answer the question: Is there in the world another more stupid way to waste bytes getting the same visual layout and making increbibly complicaìted to simply turn on a pixels on screen with a given color?

I expect that sega SCREEN 4 was a lot different from other modes.

But the problem is the same: starting from a crappy chip you need to make compromises:

1) Good graphics, bitmap modes (one byte configuration IS one pixel color, no ambiguity), command engine.
2) Forget (1) and do a better sprite engine reusing it also for tiles. But you sacrifice other things.

With the amount of 12 K VRAM TMS could get a bitmap mode like amstrad CPC mode 0. Instead got a colour clash mode like spectrum with the ADVANTAGE of having 8 times BIGGER the vram attribute table .

the TMS ran @ 10 Mhz! at this speed they could have done more. Look at the VIC-II: only 8 bMhz but it can put 8 24x21 hw sprites on screen. That's a difference between a crappy design only focused to save costs and a good design that squeze every resource from what you have.

By hit9918

Prophet (2895)

hit9918의 아바타

05-09-2019, 20:17

Quote:

Simple: the 9938 is what it is mainly because of the TMS9918. For example: look at the stupid sprite color by scanline

all this is BUSTED by the sega. as a living proof of how it can be done without the quirk.

페이지 2/4
1 | | 3 | 4