Samurai Shodown unplayble demo on MSX+V9990

Page 5/6
1 | 2 | 3 | 4 | | 6

Par erpirao

Paragon (1334)

Portrait de erpirao

28-11-2022, 20:27

thanks, your videos confirm my suspicions, my Vram is broken.

Par albs_br

Champion (499)

Portrait de albs_br

29-11-2022, 01:55

I may output debug info on the standard video output for each step, so it will put some light on what's happening.
But without a real hardware on my side here for testing the round-trip for debugging will be a pain.

Par ToriHino

Paladin (925)

Portrait de ToriHino

29-11-2022, 16:48

In the mean time I also tested on my Sony MSX2+, which gave similar results for both roms. I'm also experimentnig with my own game for V9990 (based on a UBox MSX1 library conversion for V9990 which I made), but also there I do see some artefacts (and especially some problems with the sprites, it often ends up with a lot of duplicate sprites all over the screen oO ) which don't show on OpenMSX. Looks like there are some timing or initialization issues.

Par albs_br

Champion (499)

Portrait de albs_br

29-11-2022, 19:53

Just commited a new version (0.15.1) with debug info spit to standard VDP output.

https://github.com/albs-br/msx-samurai

Thanks for helping

Par ToriHino

Paladin (925)

Portrait de ToriHino

29-11-2022, 20:31

I did a quick test, the second monitor doesn't like 60hz/NTSC but still I can see the following (here the slow down function of the Sony 2+ finally is usefull Tongue ):

  • Up until 'Loading SPR Earthquake' nothing is visible (only the VRAM gets cleared)
  • At 'Loading SPR Haohmaru' the right sprite becomes visible
  • At Enable layer B only the background becomes visible (so it looks like layer B is on top now, or the layers are swapped. I do see in your code that you put the sprite priority at 0 so would expect to be visible even in front of layer A)

After a reset of the MSX, before clearing VRAM again both sprites are visible so it looks these are loaded fine.

Par aoineko

Paragon (1138)

Portrait de aoineko

29-11-2022, 22:58

I tested my MSXgl V9990 sample on my Powergraph Lite and I confirm the glitches.
It was working not so long ago so I should be able to find out what the problem is.
With some luck, it might also fix the albs_br problem.

Par albs_br

Champion (499)

Portrait de albs_br

29-11-2022, 23:21

Anyway the emulators should behave accordingly. I hope it helps them to improve.

Par aoineko

Paragon (1138)

Portrait de aoineko

29-11-2022, 23:40

Yes, I hope that it will help to improve the emulation of the V9990.
The problem with the MSXgl sample seems to be limited to the color management (so probably not the same problem you have).
I'll dig to see if it comes from the setting of the colors in the palette registers, or the setting of the palettes in the tile layers.

Par albs_br

Champion (499)

Portrait de albs_br

29-11-2022, 23:50

It looks like it has something to do with R#22, I observed before the following strange behavior:

This doc (http://msxbanzai.tni.nl/v9990/manual.html) says:

SDA	: Set to "1" to disable layer "A" and sprites.
SDB	: Set to "1" to disable layer "B" and sprites.

But on the emulators it doesn't disable the sprites, only the layer, so I trust this to disable only layer A.

To emulate hardware well known is hard, hardware most obscure is even harder.

Par aoineko

Paragon (1138)

Portrait de aoineko

30-11-2022, 00:47

Problem solved on my side. It was the function to choose the palette of the two layers of the P1 mode which was writing only in the 4 LSB of R#13. I thought the others were not used in the P1 mode and if this is the case in the emulators (OpenMSX and WebMSX), it is not the case with a real V9990. Setting the 4 MSB to 0 solves the problem.

I will put a message on the general V9990 thread.
I hope you will find a solution to your problem!

https://webmsx.org/?ROM=https://github.com/aoineko-fr/MSXgl/...

Page 5/6
1 | 2 | 3 | 4 | | 6