V9990 write mask

Page 3/7
1 | 2 | | 4 | 5 | 6 | 7

By PingPong

Prophet (3501)

PingPong's picture

07-01-2020, 19:01

Juan Luis wrote:

Ok, Manel. I have already sent you an e-mail and I have already received your answer with the files.

The good new is Flip X is possible in real V9990 with just one BMXL command. The bad new, for me, is that I can't understand why PNTB or PGTB is dirty. At least the original Rygar patterns, the flipped pattern and Sprite is shown correctly.

I continue studying the problem.

Thanks for your help.

so bmlx works on real v9990?

By erpirao

Paladin (1018)

erpirao's picture

07-01-2020, 19:38

So the V9990 can flip bitmap / sprites without the CPU having to intervene?

By Manel46

Hero (517)

Manel46's picture

07-01-2020, 19:44

PingPong wrote:

so bmlx works on real v9990?

Yes. But in the emulator you must have the latest version of Openmsx:
https://openmsx.fixato.net/
With version 0.15.0, no.

By GhostwriterP

Hero (527)

GhostwriterP's picture

07-01-2020, 19:53

Juan Luis wrote:

The good new is Flip X is possible in real V9990 with just one BMXL command. The bad new, for me, is that I can't understand why PNTB or PGTB is dirty. At least the original Rygar patterns, the flipped pattern and Sprite is shown correctly.

Interesting... what is the trick? I have not tried it on real thing myself so I remain a little sceptic... Wink

Anyway, I could not help quickly tracing the test code and noticed that the write mask for reg 46 and 47 are both 0ffh. Normally one should prohibiting writing to screen "B" (as they call it) with setting reg 47 to zero. This could explain the "corruption" on PGTB (or also even PNTB?).

So, does it still work when reg 47 is zero (00h)?

Also interesting to know is whether both PNTB and PGTB affected!? you see, I would expect it that only PGTB is affected and only locally (at the same destination but then "projected" on screen "B"). If this is indeed the case possibly there is a use for bmxl afterall, at least knowing PGTB gets "corrupted" and simply not use that part of vram Wink.

By Manel46

Hero (517)

Manel46's picture

07-01-2020, 20:29

erpirao wrote:

So the V9990 can flip bitmap / sprites without the CPU having to intervene?

Yes. With BMLX as Juan Luis explains. In the rom it looks. In bitmap modes without problem. In modp P1, it is more complicated, but also.

By Juan Luis

Expert (100)

Juan Luis's picture

07-01-2020, 22:22

GhostwriterP wrote:
Juan Luis wrote:

The good new is Flip X is possible in real V9990 with just one BMXL command. The bad new, for me, is that I can't understand why PNTB or PGTB is dirty. At least the original Rygar patterns, the flipped pattern and Sprite is shown correctly.

Interesting... what is the trick? I have not tried it on real thing myself so I remain a little sceptic... Wink

Anyway, I could not help quickly tracing the test code and noticed that the write mask for reg 46 and 47 are both 0ffh. Normally one should prohibiting writing to screen "B" (as they call it) with setting reg 47 to zero. This could explain the "corruption" on PGTB (or also even PNTB?).

So, does it still work when reg 47 is zero (00h)?

Also interesting to know is whether both PNTB and PGTB affected!? you see, I would expect it that only PGTB is affected and only locally (at the same destination but then "projected" on screen "B"). If this is indeed the case possibly there is a use for bmxl afterall, at least knowing PGTB gets "corrupted" and simply not use that part of vram Wink.

Yes GhostwriterP. There are still good reasons to remain sceptic. I have to find out the origin of dirty pixels on plane B. I have uploaded to my Google drive another version called sprite-test2.rom. Now we have Flip X, Flip Y and Flip XY. Window Mask High is set to zero. Perhaps this avoid the problem, but I can't test it.

https://drive.google.com/drive/folders/1a_CY1ZD8t5vYIuwDE3NH...

I cross my fingers B-)

By Manel46

Hero (517)

Manel46's picture

08-01-2020, 01:15

The problem persists.
As I said, you should clean the pattern tables first
I sent you the routine, Juan Luis.

By Manuel

Ascended (16140)

Manuel's picture

08-01-2020, 09:17

It would be nice if in the end someone could summarize what is wrong in the emulation and provide a test case for it Smile

By alexito

Hero (613)

alexito's picture

08-01-2020, 16:43

@Juan Luis: this is my screen shot from sprite-test.rom it was tested on Turbo R ST 512KB RAM + Tecnobytes V9990 Power Graph + romload ver1.99u (via Playsoniq 512KB SCC mode)

Picture

By alexito

Hero (613)

alexito's picture

08-01-2020, 16:50

@Juan Luis: This is my screenshotfrom sprite-test2.rom --> it was tested on Turbo R ST 512KB Ram + Tecnobytes V9990 Power Graph + ROMLOAD Ver1.99u (Via playsoniq 512KB SCC Mode)

Picture

Page 3/7
1 | 2 | | 4 | 5 | 6 | 7