David Haywood’s MAME(tm) WIP

(29th Jan 2005)

David Haywood fixed the missing net in Super Volleyball and the glitches in Rabio Lepus lifebars in the Rabbit Punch driver.

David Haywood also fixed some corrupt backgrounds in Level 2 of Zero Zone.

Bryan McPhail submitted a fix for a large number of fixes for non-GCC compilers to aid with the building of Mame under Visual C++ (www.vcmame.net)

Bryan also started adding proper V20/V30 Disassembly support to the i386 disassembler which is now being used to diassemble code for these CPUs

Phil Stroffolino submitted a Work in Progress driver for Monza GP due to a lack of time, in the interest that somebody may be able to help out. It currently shows nothing but a black screen.

Quench and Nicola Salmoria discussed the TNZS hanging bug.

Nicola Salmoria noted a possible problem with the Charlie Circus title screen which could do with verifying against a working board. (opaque squares masking stars)

Aaron Giles fixed a bug in the Memory Optimizations added in 0.91

Aaron Giles submitted a driver for the Atari / Midway Vegas system, Gauntlet Legends and Tenth Degree are the only working games. (sorry, no screenshots)

Lawrence Gold submitted a fix for some compiler warnings / errors.

Nicola Salmoria submitted an update to the Qix driver which fixes the hang in Electric YoYo attract mode, decrypts the encrypted Kram set, and incorporates some sound updates from Derrick.

David Haywood made some observations regarding the problem with Solomon’s key blanking out levels at certain scores.

Lawrence Gold sent yet more compiler fixes.

Some discussion took place regarding the cleaning up of the remaining warnings so that -werror can once again be enabled in the makefile.

(30th Jan 2005)

R.Belmont submitted a fix which stops Street Fighter the Movie from crashing on AMD64 systems.

R.Belmont also sent an update to Lethal Enforcers fixing a few issues.

(external) Dave Didion sent a fix for the mirrored scenes in Terminator 2, however the way in which it was implemented is questionable. (additional #define when the existing one should suffice)

R.Belmont and Nicola Salmoria improved the colours in Lethal Enforcers

Gerald and David Haywood added a Japanese version of Karate Blazers to the driver.

Nathan Woods and David Haywood cleaned up the remaining compile warnings.

Angelo Salese sent an ST-V update which amongst other things allows Astra Superstars to boot, however it still has many problems.

Nicola Salmoria sent an update to the TNZS driver, fixing the hang / crash in the game, fixing sprite banking which as a result fixes various graphic problems in Insector-X and Kabuki-Z, making the latter playable.

Nicola Salmoria and R.Belmont further improved the palette handling in Lethal Enforcers.

(31st Jan 2005)

Nothing major to report.

– sorry, no screenshots yet, I may add a few later.

Ok, sorry for the quick update but I feel its necessary in this case for the reasons previously outlined.

This is pretty big for a little update anyway, there are quite a few changes. Download it from the link at the side, whatsnew.txt is here.

The u1 update will probably be out rather quickly this time, mainly due to the Memory System bug Aaron introduced which breaks a number of games. In the case of core bugs I tend to prefer having them fixed in an official update as quickly as possible.

I thought it would be kinder on those compiling various builds to let you know now, rather than after you’ve all posted .91 builds etc.

The update will of course include the submissions since .91 like the vegas driver, which might not work properly but is bound to interest some.

I’ve spent the last hour looking at this with Dox based on some inps and save states I was sent.

For a long time (since it was first emulated) the game has has had a problem where it blanks out the sprites and playfield when a certain score is reached.

Given how simple the general hardware is this seemed rather odd and I was starting to wonder if it was either a bug in this revision of the game, or a bad dump.

Careful investigation of the code revealed something rather different, the game actually appears to be protected. When you reach certain scores the game reads from one of the unmapped ports, makes a comparison with a single bit of a value stored in RAM, and if it doesn’t like the result it blanks out the sprites and tilemaps.

I’m currently attempting to find a fix for this which works in all cases, but I do wonder if there will be more to it than meets the eye.

Sometimes I consider simple fix requests from people, in quite a lot of cases the bugs are hard to fix, but this one seemed fairly easy.

The problem in Super Volleyball was that the net was missing, apparently its been like that for a while.

After looking in the GFX with the tile viewer it was clear the the net was Sprite 0.

Taking a look into vidhrdw/rpunch.c (the driver on which svolley runs) I found the following line of code in the sprite drawing function:

if (code < 0x600 && code != 0)

very strange, this was telling Mame to only draw sprites if their tile code was above 0, and below 0×600, since the net is sprite 0 it was being ignored.

After Removing this line and recompiling I found that the net was back, however there were now more glitches in both svolly and rpunch. It was clear now why there was also a check to make sure the code was below 0×600, for some reason the programmers draw sprites with codes above 0×600 when they simply want blank tiles.

The kludge in the sprite drawing however didn’t seem like a good idea, masking a sprite code to under 0×600 isn’t a very natural solution to me and is unlikely to be how the hardware worked, instead its more likely that the hardware masks the code to 0×7ff and since no roms are present for tiles 0×600-0×7ff those tiles are simply drawn as blank.

changing all the REGION_GFX3 (sprite region for this game) parts in the rom loading to create space for 0×800 tiles is one way of approaching this problem

ROM_REGION( 0×80000, REGION_GFX3, ROMREGION_ERASEFF )

(the ROMREGION_ERASEFF is neccessary in this case as the highest sprite pen is the transparent one, without it black squares would be present instead of garbage)

Indeed, changing that for all the games in the driver appears to have fixed the games.

I haven’t included screenshots of the glitches in Rabbit Punch, but if you try the various steps yourself they are clear to see in the intro after you start a game prior to the last fix.