Emulator Comparison revisited? New set of tests required.

Page 1/2
| 2

By snout

Ambassador (14810)

snout's picture

08-12-2010, 02:19

Allright guys, don't get too excited just yet. As most of you have noticed by now I'm more or less back in action and I'm even contemplating on writing a new, updated version of the MSX Emulator Comparison. This time, I would like to test MSX emulators in both Linux and Windows. In order not to waste my precious time I will most likely not review every MSX emulator known to man (e.g. fMSXSO and fMSX-SDL have not been updated in ages) -- at least not in the first edition.

As emulators got better and better I'm looking for a new set of tests that will help identify the differences in accuracy of the emulation more. For example, the 'refresh behavior' in Mission2Mir - or the recently discovered SCC bug. How can these easily be reproduced? Which software (games/demos) do you recommend in the testing toolkit? Suggestions are more than welcome!

Login or register to post comments

By JohnHassink

Ambassador (4125)

JohnHassink's picture

08-12-2010, 02:50

My 2 cents (regardless of OS, sorry):

* blueMSX video rendering could use a lot of improvement

* openMSX does strange things with programs such as Hydefos

* NLMSX (/fMSX (?)) is easy, fast and light but terrible with sound; especially Moonsound emulation

* meisei, although very heavy, would slaughter them all on some points if it would just support disks & MSX2 and higher

* ofcourse the SCC bug should be emulated, just as sprite flicker, color spill etc. is emulated. It's a flaw/limitation/special treat which should be included thus the question itself is a total no-brainer if you ask me

* besides this critics, I do not want to give the impression of nagging since I'm thankful for the existence of these emulators every god given freaking day

By Vampier

Paragon (1847)

Vampier's picture

08-12-2010, 03:11

For emulations things such as savestates/loadstates and reverse are good to take on consideration.

Anyway just wait until the new openmsx will be released.

By wolf_

Ambassador_ (9470)

wolf_'s picture

08-12-2010, 03:17

Montana John exposes the SCC bug on real SCC hardware.

By SLotman

Paragon (1130)

SLotman's picture

08-12-2010, 13:36

Multi-Plex stage 2 water is borked on all emulators Smile

By hap

Paragon (1908)

hap's picture

08-12-2010, 14:24

Mission2Mir? scrap that one Smile it was just a bug in the game itself, surfaced with an unbugged emu feature

By hap

Paragon (1908)

hap's picture

08-12-2010, 14:44

Following that, I suggest to scrap the rest of the nitpicky hard-to-determine emulation accuracy tests too, you'd need to be someone with a lot more expertise in the emulation field than the average snout to get a fair test result. Tongue

Overall, emu accuracy is terribly good, really.

Instead, compare them by user-experience and handy features (see Vampier above). Also, please present it as your personal opinion and not as a (factual) comparison approved by the whole MRC. Smile (blueMSX bragging on several occasions that it came out as winner on the previous comparison still makes my eye twitch sometimes)

By Hydragon

Champion (329)

Hydragon's picture

08-12-2010, 19:00

@Demonseed: as far as I know... NLMSX hasn't been updated in ages as well

edit: From Quibus I remember he said he hadn't uploaded Metal Limit.. because of not showing correctly the graphic demos inside of the game.

By Huey

Prophet (2479)

Huey's picture

08-12-2010, 19:45

Instead, compare them by user-experience and handy features (see Vampier above). Also, please present it as your personal opinion and not as a (factual) comparison approved by the whole MRC. Smile (blueMSX bragging on several occasions that it came out as winner on the previous comparison still makes my eye twitch sometimes)

BlueMSX rules with its debugger, Meisei rules with its graphical VRAM (sprite and tileset) viewer.

By max_iwamoto

Champion (333)

max_iwamoto's picture

08-12-2010, 21:01

BlueMSX rules with its debugger,

It is one of the best out there... But I would pay $100 to improve blueMSX debugger, especially Search feature...

By ARTRAG

Enlighted (4635)

ARTRAG's picture

08-12-2010, 22:51

Bluemsx debugger needs a couple of essential features:
1) add/remove watch point for ram/vram variables. Now there is no way to observe changes in the ram while they happens or to to add breakpoits when they happens
2) allow the user to decide the allignment of code and data. Now if it starts disassembling data, it can continue for thousands of bytes before reallinging correctly to the actual code. Moreover any label in the middle of a(wrong) instruction disappears

Page 1/2
| 2
My MSX profile