The next version of RetroCopy has finally been released! Available in 32bit and 64bit versions, grab it from the download page.

The biggest changes in this release are massive improvements to Genesis/Megadrive accuracy, across the board sound improvements and save states.


Here is what's new :-

v0.720 - March 17th 2010
-[EMU] Improved speed by average 20% in all cores
-[EMU] Fixed issues with shadow/hilight mode, vertical scrolling, DMA bugs, sprite collision, aspect ratio and timing in the Genesis/Megadrive VDP
-[EMU] Cleaned up Sega Genesis / Megadrive state, reduced size of it by 600% (allowing more rewind per megabyte)
-[EMU] Fixed bug in 6button controller that stopped it working in some Genesis/Megadrive games [Sasha]
-[EMU] Fixed a few Genesis/Megadrive memory map issues [Eke]
-[EMU] Improved YM2612/Z80/VDP timing in Genesis/Megadrive (along with other fixes 98+% of all Genesis games work very good now)
-[EMU] Fixed slowdown bug when games read input ports too much (Legend of Galahad, Kawasaki, Aleste) [Sasha]
-[EMU] Fixed numerous bugs in YM2612 core (fixes Gynoug, Beavis and Butthead, Streets of Rage and a few other games with strange glitches) [Eke,Sasha]
-[EMU] Cleaned up Sega Master System, Game Gear, Sega-E and Nintendo states
-[EMU] Added "full display" option to Game Gear. Shows the full Game Gear frame which often has valid screen data.
-[EMU] Added native sampling to YM2413 FM chip (SMS)
-[EMU] Improved efficiency, accuracy and general cleanliness of SMSVDP (SMS, GG, SEGAE)
-[EMU] Improved NES sound filtering, now sample internally at over 240000Hz and then down sample for increased quality.
-[EMU] Fixed bug in Z80 LD R, A instruction
-[EMU] Implemented Z80 IM2 support
-[EMU] Added some config options to set frequency cutoffs and mixing levels of certain audio devices

-[SOUND] Added new high quality resampling engine with different settings for speed. Every sound chip now uses it.
-[SOUND] Added 3 band equalizer (using Linkwitz Riley 4th order) , low (bass boost), medium and high frequencies can now be amplified separately
-[SOUND] Added filtering option which is designed to simulate capacitors/resistors/etc that sound on real machines was filtered through

-[CORE] Added reset button which forces hard reset
-[CORE] Added automatic save states to all systems. Simply leave a game and resume it at any time at same point you left. Can also do manual saves
-[CORE] Added SSE2 optimizations throughout (now a requirement)
-[CORE] Added memory alignment functions for better multithreading performance of video/audio buffers
-[CORE] Upgraded to boost 1.42
-[CORE] Fixed bug that didn't recognize .SMD files for Genesis/Megadrive conversion [lolcano118]
-[GUI] Fixed bug that wasn't translating some keys correctly like the underscore [Pixel_Crusher, Entrax]
-[GUI] Added new slider control for volume/bass/etc adjustment

Congratulations Ralf for this amazing emulator !!!

Ralph, i already tested games, and found very big list of delicious glitches

http://www.retrocopy.com/forumthread/112-3/glitches-and-emulation-errors-in-mega-drive-system.aspx

I love search glitches

You are quick Sasha, I'll have to add you as a beta tester one of these days...

I think you'll find there are much less issues than last build, but there are still some things to improve in Genesis certainly.

Yeah i found much less issues than last build type: beta tester, for me big honor

Thanx ralph for this wonderful emulator...will try it even more after work tonight

Hello Ralph,
I gave it a quick try with the few problematic/sensitive games I knew and it seems you fixed most of the VDP bugs/inaccuracies, good work !
You might want to check the following games though, they still have graphical glitches due to inaccurate VDP timings: Skitchin, Striker, Zero the Kamikaze, Mickey Mania (3D level), Outrunners...

You might also want to test those "homebrew" programs that do some odd tricks with VDP color palette, they are not accurately emulated in Retrocopy:
http://www.mediafire.com/download.php?oizzn3knmyz

Other games that you might want to fix::
- Sesame Street (need a delay between VINT activation in VDP register and INT triggering)
- Super Street Fighter 2 (need ROM banking hardware emulation)


Regarding sound emulation, I would say I am a little disappointed (well, after all the stuff you said you added) because, at least on my laptop, it does not sound as "crystal clear" as Kega imo, I don't really know how to describe it but it's a little bit too "noisy" and less nice to listen. I tried modifying the various sound options but did not notice much improvments with highest resampling quality or different frequency filtering (from the options, I guess you are using a similar 3-band EQ as the ones I picked from http://www.musicdsp.org/archive.php?classid=3#236 , I added this upon user suggestion but never personnaly find it very impressive).

Also I don't know if this is due to poor CPU performances (my laptop with Core2 Duo P8400@2.27GHz can only run 2.5 SMS cores and cannot render the "3D room" nicely enough to make it worth the use) but I noticed constant sound crackling when loading a new game and from time to time during gameplay (like sound buffer underruns). There are also sometime odd noisy sounds that I did not remember hearing before (tried in Sonic 2 vs mode), seemed like an inaccurate PSG channel.

The sound on the NES on the few games I tested have noises and cracks in it :

Zelda 1, Zelda 2, Super Mario Bros 2 and 3.

No need to play the game for hours, just listen to the attract mode, and you will listen cracks and noises sometimes.

Another issue with my mouse.

On Windows, I am using 750 DPI settings. After launching RC, seems like my mouse is in 200 DPI mode.

I have to switch it from the mouse to 4000 DPI to avoid spending 2 hours to move from one part of the screen to the wished option.

Of course, once I shut down RC, then my mouse is still in 4000 DPI, and I have to readjust it to 750 DPI to have things come back to normal.

(I have a sidewinder X8)

Keep up with the good work !

Thanks for the testing Eke. Something I did broke a few homebrew demos (charles 1500 color one to name one) and I didn't have enough time to check out why. I'll be adding a proper system wide debugger for my RC testing soon as it's too difficult doing it without one.

In regards to 3D room, there are issues with DirectSound3D, basically you need to make sure the sample rate used in RC matches your system rate (only applicable to Vista/Win7). I really dislike DirectSound3D and need to add my own 3D sound engine soon enough.

I know from other testers that only manage to get 20-30fps in the 3D room that it still runs smooth in regards to emulation output, this is because it's all handled on separate cores. So even if you get low video performance it should be fine provided you've found the right tweaked setup.

Things to watch out for :-
1) Disable threaded optimization if you have a NVIDIA card.
2) Don't enable "3D Emulator filter" if you have a NVIDIA card

Currently if you're getting 2.5 on the benchmark it would mean you could get the same with Genesis (roughly same performance due to use of non cycle accurate musashi core). So CPU speed isn't the issue.

The sound settings now are more intensive than last versions, the "good" is something close to 0.700, but I want to do a bit more work on my linear resampling. It is however fast. The audio quality affects benchmark speed, this is on purpose as I want the benchmark to show what you can get based on the quality settings you've chosen.

Hmm I have no sound issues at all, but they can occur for a variety of reasons.

Press F2 when running a game and up will come some debug information. Anytime a "repos" occurs that is a buffer underrun or overrun for sound, this will result in an audible noise.

It's usual to have 1-3 repos as RC finds the right balance with your system, you shouldn't be getting recurrent repos though. If you aren't getting them constantly and have sound pops/clicks then it could be an issue with sampling rates. Ensure the RC sample rate matches system rate, by default RC is at 44100, trying 48000 (and restarting) may alleviate the issues.

In Vista/Win7 if the rates don't match it resamples the audio for you which can introduce a bit of noise (this includes the bit depth setting, some audio drivers use 24 bits by default which means audio will be filtered in RC as it uses 16bits).



Thanks for reporting that. I upgraded to a newer SDL since 0.700 but it has a few issues, I've also noticed some mouse oddities time to time, it is frustrating. In regards to DPI settings though, RC uses the windows speed so it should be correct (ie how fast the mouse cursor moves in normal windows is what RC uses). Are you saying it's not as fast as it moves in normal use?

Yeah well there are definitely differences between Kega and RC as I showed you the images a week or so ago. I don't know how much filtering Kega is doing of the PSG, but in RC its sampled at native speeds and then downsampled using the selected filters.

To my ear Kega sounds a bit more noisey/raw in the FM chip but I don't know if that is how it should be or not. RC sounds very crisp to me, almost too crisp so maybe it's a sound card option on your end or something.

I took some high quality 96KHz samples from my SMS and matched them up to what I'm doing now and it's almost the same sound wise (including noise channel), though it looks like the SMS positions the center a bit differently than how most people do it. I don't have a MD to do the same though so it's hard to gauge that.

A few testers have said they prefer the sound of RC to Kega but to be honest Steve has spent like 15 years working on the Genesis sound emulation so I'm almost inclined to think what he has is more accurate at this stage. And until I can do some high quality real Genesis audio testing I won't really know.



You use that one do you? I did have a look at that one and tried it briefly but it's broken, doesn't do a good job of filtering frequencies (just try adjusting the default settings and you will see). I spent some time asking around some DSP forums/chat rooms and implemented a Linkwitz Riley 4th order filter for the EQ. I was recommended to try something else though :-

http://musicdsp.org/files/biquad.c

I didn't try it as I liked the output I got with what I did, but maybe you'll find some interest in that over the one you linked to.

Thanks for reporting that. I upgraded to a newer SDL since 0.700 but it has a few issues, I've also noticed some mouse oddities time to time, it is frustrating. In regards to DPI settings though, RC uses the windows speed so it should be correct (ie how fast the mouse cursor moves in normal windows is what RC uses). Are you saying it's not as fast as it moves in normal use?[/quote]

Yes, this is correct, the cursor is slow as hell, and therefore I need to compensate by using the internal 4000 DPI of the mouse once RC is launched.

I will keep you updated for the sounds issues after further investigations

Thanks for the link, will give it a test (there is also some 4th order Linkwitz-Riley filters on MusicDSP, just below the one I linked).

Regarding sound issues, I'm not using the 3D room (sorry, don't really like it) but fullscreen mode and I have ATI Radeon chip. Sound scratch seems to be constant when reloading games and audio garbage seem to happen from time to time durign gameplay, more or less frequently, from what I can tell, it's either a problem with sound sync (when feeding the sound buffers) or sound clipping when adding streams.

From my limited tests, I don't think there are really accuracy issues in your FM implementation (all FM cores should be equally accurate by now anyway) and I have some doubts the tiny sample differences you showed before are going to affect sound output that much, it's more related to the way sound is mixed/resampled/filtered, it's very hard to describe anyway, it's just that Kega output is more "pleasant" in my ears on the same setup (even if it does not have 48Khz support here). It's very likely this can be fixed with specific customization on the user side though and I'm pretty sure it's related to audio backend implementation.


I also have slow menu response like reported above me, this is not REALLY slow but enough to make the interface a little frustrating to use compared to other emulators.

Hmm didn't notice that, probably should have tried that before doing my own... oh well.

If you give it a go tell me how it sounds.



Interesting, I'll look into it, maybe I'm not passing on enough events I'm getting from SDL (I had to change a lot of SDL because it was broken). I did testing originally and for me it is perfect so it is strange. I have access to raw mouse input but due to it not being accelerated to same speed as OS I simply use what Windows reports. Does VSYNC off affect it at all?

I have an Nvidia Geforce 250GTS with this enabled. Everything runs fine.