new GFX card

페이지 7/20
1 | 2 | 3 | 4 | 5 | 6 | | 8 | 9 | 10 | 11 | 12

By MagicBox

Master (198)

MagicBox의 아바타

13-09-2008, 11:04

What you describe is merely a 1-n layer engine without pattern support. While that would work for desktop types of screens, it's not suitable for (fast) games on a 3.5MHz Z80. Pattern system is essential.

As for memory I had in mind 64KBx32bit 100MHz SRAM (256Kb total). This allows for realtime interleaving VDP/Z80 access time. Z80 reads being buffered so SRAM access cycles don't need to be prolonged to that of the Z80 cycle time.

As for V9990, I think it's been explained already why it has limited support. The programming interface is not easy and it is still limited in some areas. As described in previous posts the new VDP doesn't need to be backward compatible as the old VDP can just privde that.

Regardless, I'll continue to design accoring to my proposed specs with other community feedback and get the thing to work first. Then we can see what will happen Wink

By wolf_

Ambassador_ (9855)

wolf_의 아바타

13-09-2008, 11:15

As for ease of use of the G9k: *click*

By PingPong

Prophet (3680)

PingPong의 아바타

13-09-2008, 12:20

@MagicBox: the v9990 is not compatible with any other VDP.
The designers removed all the garbage that has its roots on the TMS. For example to set the VRAM ptr you do not have to set bit 6 to write or reset if you want to read, just out the address. Plus you have two vram ptrs, for read and write operations.
Only an example, of course.

However, I go for a new card mainly for these reason:

- More performances allowed by recent hw
- No problems about limited chip availibility (how many v9990 there are?)
- FPGA allow hw changes more easily
- simple , bit powerfull compatibility. Just plug the old video into the card and choose if you want VDP or VDPX video out.

By MagicBox

Master (198)

MagicBox의 아바타

13-09-2008, 13:39

The question arises.. is it worth for me putting effort in (besides personal worths like the challenge)?

As for defining a new standard, writing software for it and eventual volume production, it does seem appealing. But for it to take off, ofcourse it would need all the support from the community it can get.

And to avoid having tons of carts dangling from your MSX, incorporating OPL4 would be a feasible option.

If the design succeeds and the cost is manageable, would there be other MSX devs out there who'd want to code something for it, be it demos or games? I think it can take off if indeed some sort of organization would be formed around it. The VDPX Consortium.. has a nice ring to it Wink People that provide feedback, offer ideas, write code and what not more.

By MagicBox

Master (198)

MagicBox의 아바타

13-09-2008, 13:43

As a sidenote on compatibility.. once the VDPX design would be final, it's logical presentation to the CPU should never change again, any design upgrade should never render software written for it useless. Never. Hence a good deal of R&D is needed for such project before it is finalized. Upgrades would be limited to performance upgrades only rather than architectural upgrades.

By wolf_

Ambassador_ (9855)

wolf_의 아바타

13-09-2008, 14:00

If the design succeeds and the cost is manageable, would there be other MSX devs out there who'd want to code something for it, be it demos or games?

This is indeed the most important issues of all:
- acceptance
- content

As I've said earlier, if I'd have to make a Pac Man or a Rise Out for that chip just to have 'something' we could call content, then count me out.. I'm not going to make an 1985-cassette-style game anno today. If others think the same then there's your problem. Large games take serious development time and you actually need to be skillful in the relevant areas (gfx, code, music, game design) as well in order to do so. (I don't see an average/beginner team just make an RPG the scale of Xak, just to mention something)

Content for such a chip is a bit like searching for an extra-terrestrial planet on which humans can live. I'm sure there is one, but there are a lot of conditions to meet first.

By Leo

Paragon (1236)

Leo의 아바타

13-09-2008, 14:09

i remember doing demos for pc in asm, a fast cpu like 486dx66 in these times accessing the videoram directly ( was on segment $a000) was enough to do everything that console could on the 320x200 x256 color + palette , the 320x200x32kcolor was very fun but starting to be slow...

why dont we keep msx standard unchanged but give it vram access direct from Z80, many vhdl design for fast Z80 are available for fpga on opencores.org for instance...

By PingPong

Prophet (3680)

PingPong의 아바타

13-09-2008, 14:51

why dont we keep msx standard unchanged but give it vram access direct from Z80, many vhdl design for fast Z80 are available for fpga on opencores.org for instance...

AFAIK allowing vram access from z80 is not a thing that could be done only by putting a new cartridge in a slot, instead requires hw modifications. Not acceptable IMHO.
Plus, if the rest of HW remain unchanged what can you gain from a faster CPU if it has to wait to access VRAM because of VDP concurrency?
For example in screen 2 the vdp have a 1 cycle every 16 dedicated for cpu access. AFAIK the memory cycle is about 300 ns. Assuming 1 T-STATE = 1 access, because of the limitation of 1 access every 16 is still here, the z80 must wait about 5us for any access. Even a PIII could not perform more faster if compared to a old z80 in a vram filling routine.

I think we need a new hw. I also think that if this is simple but powerfull, a good number of peoples start to enjoy with it.
Plus, having a gfx hw that work on every msx family is good. A sw developer does not have to think about the alternatives:

- a crappy limited game but for all (msx 1 and upper compatible)
- a little more better but only for msx2+

By ivke2006

Resident (53)

ivke2006의 아바타

13-09-2008, 14:59

The question arises.. is it worth for me putting effort in (besides personal worths like the challenge)?

And to avoid having tons of carts dangling from your MSX, incorporating OPL4 would be a feasible option.

If the design succeeds and the cost is manageable, would there be other MSX devs out there who'd want to code something for it, be it demos or games? I think it can take off if indeed some sort of organization would be formed around it. The VDPX Consortium.. has a nice ring to it People that provide feedback, offer ideas, write code and what not more.

Is it worth the effort? I'm not sure wat you mean, but there are always people who want to stay at the orginal msx 1 specs and there are also people who have a expensive V9990 card, and fear that no new software will be made for their v9990 because there is a newer, better and cheaper gfx card available.

Incorporating OPL4 ? You started with the idea of a cheap (<$50), easy to program but powerfull GFX card. What will it do to the costs and complexity? and remember that the OPL4 chip is also outdated. I would advise to make the gfx card first and then, if you still like the idea, a 'deluxe' version which incorparates a sound chip.

The VDPX organisation Well, I think that's a good idea, please define what you need! Something like the msx DEV competition could also increase popularity of your GFX card.

By Manuel

Ascended (17777)

Manuel의 아바타

13-09-2008, 16:28

AFAIK allowing vram access from z80 is not a thing that could be done only by putting a new cartridge in a slot, instead requires hw modifications. Not acceptable IMHO.

A GFX9000 is already a cartridge, it could be modified to have the VRAM mapped into Z80 address space like any RAM containing cartridge. Would be cool for any graphics-cartridge, of course, including VSU!

페이지 7/20
1 | 2 | 3 | 4 | 5 | 6 | | 8 | 9 | 10 | 11 | 12