Topic: IQ's PGM WIP

[JacKc]

Ok iq_132,

I will made some tests about this new PGM Build version

[blueonhell]

thx

[iq_132]

Ok iq_132,

I will made some tests about this new PGM Build version

Thank you.

I've modified the way the PGM priority works for Mame 0.130u1 to hopefully fix the bugs shown here in MAME (there is a bit of a speed hit as a result right now, but I haven't tired optimizing it)  I'd appreciate it if people could run through the (working) games after the next release.

Great! I'm looking forward to seeing that.

As for XingXings kov2p hacks, sorry, they're not acceptable in MAME.  The real ARM rom should be dumped instead, hacking another one isn't a valid solution from our point of view.

Not unexpected.  However, you should take a look at kov2p205 it seems to be a valid dump (as best I can tell). ^^

[JacKc]

Ok...some results about FBA PGM4 build.

- Sounds in kov2 and generally PGM games need to be improved
- Set olds100a works but not fully works (in fact it crashes in the end of stage 3 when fighting the Boss - bug also in FBA Shuffle - See snap 1)
- Graphical bug in Ranking Screen (bug already known by iq_132 - See snap 2)

[Leaf]

stage 3 crash has been fixed, also fix the controller problem. (by Creamymami)

[iq_132]

Ah, I was wondering what that code was in xingxing's source for the ram patch (0xffff-i). ^^
Still doesn't fix the problem with the intro though...

[pmc2]

Mame PLus 0.130 r4111 released:

PGM Update:

- PGM fix [katana]
- fixed gfx priority in kov2p [XingXing]
- fixed freeze problem in olds100a, added region dips for several PGM games [Creamymami]
- updated PGM driver, kov2p and olds100a is working [XingXing, Creamymami]

interesting for fba ?

[iq_132]

Mame PLus 0.130 r4111 released:

PGM Update:

- PGM fix [katana]

Not sure what that entails...

- fixed gfx priority in kov2p [XingXing]

It's still not right.  The characters are still missing shadows in level 2.
I'm going to wait and see what Haze has come up with.

- fixed freeze problem in olds100a

Already added.

added region dips for several PGM games [Creamymami]

Based on my sources

- updated PGM driver, kov2p and olds100a is working [XingXing, Creamymami]

Added a few days ago.

[JacKc]

A off topic post but found a interesting kovsh set...It's Knights of Valour Superheroes \ Sangoku Senki Superheroes (v106) (surely a hack but need to be confirmed)

Unicode

Code: [Select]

L"Knights of Valour Superheroes\0\u4E09\u56FD\u6218\u7EAA\u98CE\u4E91\u518D\u8D77 (V106)\0"

- Some real hardware screenshots for Knights of Valour Super Heroes (not V106)



kovsh106 ROMset :

p0600.322     crc c48df7de size 4 194 304
t0600.320     crc 164b3c94 size 4 194 304
a0604.v200   crc 26b59fd3 size 4 194 304
b0602.v200   crc 9df77934 size 1 048 576
kovsharm.bin crc 9d8ecc0c size     16 384

some snaps from Nebula (not works on Mame and FBA - See Attached Files)

Note : To enable Hidden Character XuZhu, at character selection screen :
* B, B, B, Left, C, C, C, Up, D, D, D, Right, B, C, D, Right, D, C, B.

[narcissus]

- PGM fix [katana]
only the Chinese lst... not src

Knights of Valour Superheroes \ Sangoku Senki Superheroes (v106)
it's a Chinese hack, and there're about 4,5 versions...

[Haze]

Not sure what that entails...
It's still not right.  The characters are still missing shadows in level 2.
I'm going to wait and see what Haze has come up with.
Already added.
Based on my sources and no credit given
Added a few days ago.

What I've done for PGM in MAME is convert the rendering of the sprites to be a single pass (rather than the kludgy finding start and end marker stuff), marking the priority bit into a priority buffer (actually a high bit in the temp render buffer) for each pixel drawn.

I then copy the bitmap once before the tilemap, only copying low priority pixels, and then once after the tilemap, only copying high priority pixels.

This means that sprite-sprite priority remains intact (as the original render to buffer is done in one pass) and the sprite-tilemap priority is correct (because it determines if a pixel is shown above / below the tilemap based on the priority value in the render buffer)

As for misplaced credit, it happens, and anybody who is doing emulation stuff purely for credit should probably just find something else to do instead.  Credit is unimportant, achieving progress and knowing you've achieved something is much more important, I don't care if I get credited for things or not.

[iq_132]

What I've done for PGM in MAME is convert the rendering of the sprites to be a single pass (rather than the kludgy finding start and end marker stuff), marking the priority bit into a priority buffer (actually a high bit in the temp render buffer) for each pixel drawn.

I then copy the bitmap once before the tilemap, only copying low priority pixels, and then once after the tilemap, only copying high priority pixels.

This means that sprite-sprite priority remains intact (as the original render to buffer is done in one pass) and the sprite-tilemap priority is correct (because it determines if a pixel is shown above / below the tilemap based on the priority value in the render buffer)

Sweet! I've implemented this in FBA.  Thanks for the explanation Haze.

[JacKc]

Wow...Now thoses Shadows in stage 2 for kov2 have been fixed

Thanks Haze and iq_132 for all your works !!!

[iq_132]

I got bored and started looking at olds/olds100/olds100a and managed to get them all to run. ^^

Olds100 was easily fixable, since it is exactly the same data as olds100a, just with a different rom set. The problem is that it was missing a rom. ^^
If you look at olds, it has 5x512kb program roms, and olds100 only has 4. Given that the data from olds100a is the same, just merged into a single 4mb rom, it was easy enough to figure out what the data should be.

   { "v100-u1.040",     0x080000, 0x37ea4e75, 1 | BRF_ESS | BRF_PRG },

You can get this rom by splitting olds100a's p0500.v10 into 512kb chunks -- it should be chunk #5.

Olds only needed the ram_dump added and the protection hooked up. ^^ Nothing terribly difficult there.

[JacKc]

WOW 

Excellent works iq_132 (as always  )