V9958 replacement

Enlighted (4138)

05-11-2021, 21:23

here there is a C64's VIC-II replacement outputting hdmi
would be nice to have a v9958 replacement that does the same maybe allowing a msx2 2+ chip replacement.
As F18 this chip could have also some enhanced mode to unblock allowing to remove some boring / unwanted features of the v9958 like:
stupid magic Y coordinate for sprites.
faster blitter
......
no scaline limit for sprites

link: https://www.youtube.com/watch?v=-4J3ENHCOHo

By Metalion

Paragon (1625)

05-11-2021, 21:46

Very nice ... It would be indeed a nice idea to have the same hardware for MSX.

By Manuel

Ascended (19469)

05-11-2021, 23:09

Maxim Vlasov from Meteor-M was working on something like that: the Procyon. Unfortunately I haven't heard from him since 2017, but his demo on the MSX fair Nijmegen 2016 was impressive!

By PingPong

Enlighted (4138)

05-11-2021, 23:14

Effectively i think the v9958 was not a so bad chip, it is well balanced for an 8 bit computer in terms of capabilities.
Its major drawbacks are due mainly to the heritage with TMS VDP (the real pain in the ass chip) and in minor measure to brand new mistakes in V9938 .
Removing some stupid choices would made this chip very interesting. let's imagine what you can do if some stupid 'features' could be removed by setting an enhanced bit in some register, say:
- no sprite/scanline limits
- no horridus x scanline early clock (WHY!!!!!)
- no magic stupid Y magic sprite coordinate (was already stupid on TMS it's even more stupid on v9938 that can scroll vertically via hw)
- ability to hook the SCT address not to the plane but to the sprite pattern number!
- command terminated interrupt.
- faster blitter, no difference between byte move and logical pixel move, ability to perform logical operations always enabled, maybe enabling a special optimization if you use the logical IMP operation avoiding read/logicalop/write for the target operation.
NOTE: this could be been done internally on v9938, by "inferring the state" of the (now internal) CM2 bit of command register using the logical operation code
(all block operations have CM2=0 for logical CM2=1 for high speed, so:
CM2 = CM3 == 0 || LogOp == 0 if i'm not mistaken ...)

By PingPong

Enlighted (4138)

05-11-2021, 23:21

yeah, it was a dead hope.

just now i've got the curiosity of checking the vdp commands and args register values.
looking at the command table the CM2 bit (n. 2) is meaningful only for command block operations and switch between byte move or logical move.
this make me think that this bit is somekind of useless because it's value could be "synthetized" by a combination of bit 4 (CM3 that switch between block or non block operation and the value of logical operation in argument register.
Basically you can have CM2 = 0 only if CM3 is 1 AND logop=0 otherwise is always 1.
So pratically the vdp engineers could have removed this bit because it's implied by CM3 and logop.
Or i'm wrong?

By PingPong

Enlighted (4138)

06-11-2021, 15:22

Manuel wrote:

Maxim Vlasov from Meteor-M was working on something like that: the Procyon. Unfortunately I haven't heard from him since 2017, but his demo on the MSX fair Nijmegen 2016 was impressive!

I've checked out, of this project there is no mention on the official site.
It is a dead project. Maybe instead of focusing on a more complex project that do not reach the end, a small V9958 improved replacement is more feasible. What about asking to the F18 developer?

By jltursan

Prophet (2619)

06-11-2021, 18:49

Quote:

What about asking to the F18 developer?

He's struggling to keep alive it's new MK2 project and after three years, the parts shortage is not helping.
The last report of the main features of the second F18A version is:

* Same size as the 40-pin footprint of the original IC. Removes the physical problems of installing in different systems. (done)
* Digital video output, and now also supports VGA output (requires a firmware change). (done)
* 512K of VRAM, with support for placing VDP tables anywhere in the VRAM. (physical SRAM is done, updating HDL to support external SRAM is WIP)
* Built-in firmware updating via USB to a PC. (proven, WIP)
* Audio capability via 12-DAC, also includes 12-bit ADC to sample the host-system audio which allows mixing and control (requires external connection of host audio signal). (done)
* Compatible with original F18A. (WIP, ongoing as HDL is written)

Not exactly a V9958 but a great replacement anyway (for the TMS9918 family of course, although the V99x8 replacement is in its roadmap...).

By PingPong

Enlighted (4138)

06-11-2021, 20:38

jltursan wrote:

Not exactly a V9958 but a great replacement anyway (for the TMS9918 family of course, although the V99x8 replacement is in its roadmap...).

Where is stated? AFAIK the maker did not plan to do a v9958 replacement, only to be compatible with 80 column mode, that is another story...

By jltursan

Prophet (2619)

06-11-2021, 21:26

In the first post of the F18A MK2 thread.

Check where he says "* With simple low-cost adapter boards to re-arrange the wiring, plus a firmware change, the MK2 can be used to replace other VDPs like those in the NES, the Master System, MSX2 (i.e. 9938/59 system), etc."

...but this was three years ago. Hope he can finally launch a first batch, then with a bit of luck more upgrades will come.

By PingPong

Enlighted (4138)

06-11-2021, 23:38

It is a very long thread... but i've only seen some discussions. not a real commitment

By erpirao

Paragon (1315)

07-11-2021, 00:09

Wouldn't it be feasible to use the 1chipMSX core?