V9990 FPGA recreation.

Page 1/2
| 2

Par BeroBBS

Supporter (1)

Portrait de BeroBBS

06-02-2021, 21:01

Howdy, if there are tms9918's-v9958's on FPGA then where is the V9990? It would greatly improve the count of the chips that are rare nowadays.

!login ou Inscrivez-vous pour poster

Par syn

Prophet (2061)

Portrait de syn

06-02-2021, 21:44

These kinda things need to be made by someone(s) who

- cares (cares enough about msx and v9990) to invest time in it
- has the knowledge to create something like this
- actually has spare time and resources to do it

I think the amount of people who can meet these 3 requirements can be counted on one hand, otherwise it would have been made already.
Take note that its hard to do it "perfect". From what I can tell the existing FPGA msx implementations (1ChipMSX, Zemmix Neo, and the Zemmix Neo Clones) are not not perfect. Hell even the SID from the C64 is not perfectly replicated yet from what I can tell and they got like 3 or so projects dedicated to it.

Par Ivan

Ascended (9296)

Portrait de Ivan

07-02-2021, 10:10

A bit off-topic but related to the V9990, I do wonder what is the status of the KrakenGraph development. Supposedly it is going to be the cheapest MSX cartridge with built-in V9990 chip.

Par AxelStone

Prophet (3026)

Portrait de AxelStone

09-02-2021, 11:56

I don't think we can see a 9990 implementation at short term. The fast way should be translate emulation to FPGA core, sometimes it's done in this way if schematics are not available. Let's wait to see how HRA project goes.

Par Grauw

Ascended (10016)

Portrait de Grauw

09-02-2021, 12:54

And even if we did get one I don’t trust the implementation to be accurate. There is not enough software to confirm its correctness. OpenMSX’s implementation has uncovered new bugs with pretty much every new piece of V9990 software.

Par AxelStone

Prophet (3026)

Portrait de AxelStone

09-02-2021, 13:09

I agree, V9990 has all the ingredients to never be implemented:

  • Very low market impact (in fact never commercialized officially)
  • Very low software available, not only for testing purposes but also to consider if it is worth implementing.

Just to understand how much impact it should have, CPC core developer said that it's not interested in CPC+ at all because of the lack of software and market impact.

Par iamweasel2

Paladin (683)

Portrait de iamweasel2

09-02-2021, 17:55

Grauw wrote:

And even if we did get one I don’t trust the implementation to be accurate. There is not enough software to confirm its correctness. OpenMSX’s implementation has uncovered new bugs with pretty much every new piece of V9990 software.

well, if we follow the software test philosophy, we only need it to be accurate enough to run the softwares that we have now. We don't need it to be accurate with VDP functions / behaviour that aren't (up until now) used.

Par Grauw

Ascended (10016)

Portrait de Grauw

09-02-2021, 18:00

As a developer I run into emulation issues quite often, and I’m not a fan Smile. I’d rather have good accurate emulation.

Par PingPong

Prophet (3756)

Portrait de PingPong

09-02-2021, 19:26

would be nice to have instead a v9990 implementation. however i think it will not be a reality.
Maybe a v9958 replaceable core with improved functions like a really fast blitter, decent sprites, like F18 for the TMS?

Par iamweasel2

Paladin (683)

Portrait de iamweasel2

09-02-2021, 19:42

Grauw wrote:

As a developer I run into emulation issues quite often, and I’m not a fan Smile. I’d rather have good accurate emulation.

As a developer, I also want it to be as accurate as it can be, but I know this is something that takes time. Bugs will only be discovered when new software is developed, and you need to start somewhere.

If a initial V9990 FPGA implementation is capable of running what we have now, I don't care if it is not 100% accurate, I know it will get better in the future, when new software appears.

Par iamweasel2

Paladin (683)

Portrait de iamweasel2

09-02-2021, 19:43

PingPong wrote:

would be nice to have instead a v9990 implementation. however i think it will not be a reality.
Maybe a v9958 replaceable core with improved functions like a really fast blitter, decent sprites, like F18 for the TMS?

What good that would do, if we have no software that could use it?

Page 1/2
| 2