UME 0.154ex2

UME is the complete/combined version of the MAME / MESS project.

MAME大老 Haze 發布了由他編譯的UME新版本，加入了雷電2與NMK等基板最新驅動，原文如下:

0.154ex2 is built from SVN revision 32127
I haven’t had time to look at refactoring the MAME rom filenames to be built from separate label and location fields, but feel it’s about time to release an update anyway as things like the NMK004 sound emulation and Raiden 2 progress represent interesting and significant progress.
A number of sites have been posting ‘0.155’ builds, they are fake, there is no scheduled release date for MAME 0.155 yet, none of the final preparation that happens before a new release has been initiated yet.
UME 0.154ex2 Windows binaries – 32-bit, 64-bit and all tools
UME 0.154ex2 sources
Here is the 0.154ex1 to 0.154ex2 SVN log
Other Binaries (if you don’t know what these are you don’t need them)
MAME/MESS split 0.154ex2 Windows binaries – 32-bit, 64-bit and all tools
Points of Interest
I feel the most significant news relating to this update is the progress with the NMK004 sound emulation, bringing the sound from imperfectly simulated using HLE code to practically perfect. Others might argue the recent progress in Raiden 2 / DX is more significant. That is also big news, furthering the understanding of the protection systems used on that game greatly, and it is, for all intents now playable, but without further testing the confidence level isn’t that high; several of the changes that were subsequently ported to the 68k based games (Legionnaire, Heated Barrel etc.) actually caused them to regress slightly, and Zero Team (V30 based) still has significant issues too, all clear indications that there is still a lot to understand with the COP protection.
I’ve seen the question asked in a number of places ‘does this also improve the Raiden 2 / DX NEW version’ and the answer to that is no, that game (it’s a 2-in-1) uses an evolution of the protection, and handles a lot of the basic protection setup in a different way. Rom banking still isn’t handled there, so they continue to crash like they always have. Even if we get them working they’re actually inferior versions because the sound system is replaced with bootleg quality sound. As a side note, if anybody has a ‘Raiden 2 2000′ or ‘Raiden DX 2000′ board it would be nice if you could send it for dumping, they’re probably the same thing as the ‘New’ versions but with a different title (so same poor quality sound etc.) but so far they’ve eluded us.
(I also added CHD entries for the Playstation Raiden Project and Raiden DX to the Playstation list, for reference. If you’re wondering why they weren’t already there it’s because many of the lists for CD / Tape / Disk based systems are just built on an ‘as-needed’ basis, or from obscure sourced material that isn’t already listed elsewhere – they’ll mature with time)
Speaking of Raiden II, while it’s true that Seibu games have an awful lot of code revisions, and it’s not always obvious what the difference is between them unless they’re clearly designated ‘OLD’ and ‘NEW’ versions there are two very obviously different versions of Raiden 2 that have been documented for a while. First of all there is the regular Raiden II game, this has a sepia high score table in the attract mode.


In addition to this version there is one dubbed ‘Easy version’ by many fans of the series. It can be identified by a coloured fighter on a black background behind the high score table instead of the usual picture. It’s not a simple difficulty rebalance either, some enemies have been moved about, as you can see from the 2nd screenshot comparing with the one above, the pictured enemy simply doesn’t appear in the first level with the regular version of the game.


Some of the 68k based COP games have improved tho, for example I fixed the scrolling on the Godzilla game, although the reason the glitching was so obvious is because of the absurd way in which the game tilemaps are programmed. Instead of using the full range of the scroll register the game only uses 4-bits, allowing for a scroll range of 0 to 15- pixels, after that point they re-upload the entire tilemap shifted by a tile, rather than just adding a new column to one side and using bigger scroll values, it’s an enormously inefficient way of doing things! The problem was (due to missing private buffer) DMA the sync between the scroll values and the re-uploaded tilemaps was wrong.
system11 dumped a Visco ‘Go-Stop’ board, this runs on Seta ST-0016 based hardware, it’s currently non-working because it needs some attention, but it looks like it will reveal a few more secrets of the ST-0016 to us, the backgrounds for example clearly need a bigger tile setting enabled.


http://mamedev.emulab.it/haze/