openMSX "Accuracy"

Page 3/3
1 | 2 |

By mars2000you

Enlighted (5378)

mars2000you's picture

06-11-2018, 22:52

Concerning blueMSX: some ports that don't depend from libreto/RetroArch exist, especially

* CocoaMSX for MacOS : https://github.com/CocoaMSX/CocoaMSX
* blueMSX-Wii for the Wii console : http://www.wiimsx.com/
* MSX.emu for Android and iOS
* blueMSXBOX for XBox

I will not edit myself this wiki, but I think these precisions should be included. Besides, the fact that MSX.emu is actually a blueMSX port requires to change a little the table.

By TomH

Master (242)

TomH's picture

07-11-2018, 03:53

I'll have a look at it if I can muster the enthusiasm. I find that whole school of listing and ranking things by arbitrary jargon very exasperating.

The whole 'cycle accuracy' thing in particular has probably exhausted my patience for the week. Even just an MSX 1 has at least two separate clocks in it — the 3.58Mhz for the CPU and the 5.37Mhz for the VDP. So whose clock are we talking about? If you've a cycle accurate VDP and a cycle accurate Z80, with properly timed communications between them, are you accurate to whatever is the least common multiple of the two clocks? What about if you're modelling a composite signal like me, which is continuous and analogue? Am I allowed to one-up my competitor by just constantly sampling that at a higher and higher precision?

I am grateful for this opportunity to rant.

By mars2000you

Enlighted (5378)

mars2000you's picture

07-11-2018, 13:36

@TomH: you could be interested by this article written by Daniel Vik, the main coder of blueMSX, about cycle accuracy: http://www.danielvik.com/2010/04/cycle-accurate-computer-emu...

Of course, in fine, it's always a problem of definition, eventually limited by the computer resources. You can indeed have a definition that is closer to the real thing, but will add (almost) nothing visible for the end-user, except that he will need a more powerful machine to run the emulator without problem of resources.

By the way, I had already this kind of discussion in 2009 on the SMS forum: http://www.smspower.org/forums/11817-NewSMSEmulatorRetroCopy... and one of the possible conclusions was already that 'cycle accuracy' was more a marketing expression. So, nothing new under the sun 9 years after!

By Manuel

Ascended (14738)

Manuel's picture

07-11-2018, 15:20

By TomH

Master (242)

TomH's picture

07-11-2018, 15:58

Agreed entirely on "cycle accuracy" as a marketing expression, which is why it's a frustration that what attempts to be an authoritative wiki — and seemingly gets a decent amount of traffic — has decided to endorse it as an informal synonym of "perfect accuracy".

Oh well, c'est la vie.

By Grauw

Enlighted (7465)

Grauw's picture

07-11-2018, 17:12

I think the term cycle accuracy isn’t super precise on what exact level of accuracy there is, but in contrast to line or frame accuracy it gives me useful information...

By TomH

Master (242)

TomH's picture

07-11-2018, 16:55

It definitely gives you exclusive information: it excludes some of the false simplifications the author might have made. But I'm not sure it tells you any more than that, and it's all based on the honour system and subjective analysis anyway as per the linked SMS thread getting into the weeds on whether blueMSX deserves the label.

I'd definitely argue that it's nonsense to consider, as that Wiki does, that "Very high" accuracy is worse than "cycle" accuracy. It's like saying that bananas are better than Tuesdays.

Page 3/3
1 | 2 |
My MSX profile