User talk:Tommy

From Emulation General Wiki
Jump to navigation Jump to search

The recommendation checklist consists of two main things: accuracy and usability.

Accuracy is usually weighed stronger than usability, but if two emulators are close, but the slightly less accurate one is easier to use, then the slightly less accurate one will be recommended over the more accurate one. If the accuracy difference is night and day, with the less user friendly one being much more accurate, then it will be recommended over the less accurate one.

While an emulator author may have a hard time deciding usability, accuracy is boolean. I personally prefer hard numbers from accuracy tests and/or game compatibility, but that isn't always necessary to show accuracy. I do not think you should change the recommendation status of your own emulator (for obvious reasons), but showing that accuracy and usability are important goals, and that you're keeping your emulator active are good ways to get it recommended. If you would like to do accuracy testing for us vs. other popular (and even unpopular) emulators, that would go a very long way.

Also you mentioned subcycle accuracy on your page. If you would like to add an entry for that here explaining the difference between that and cycle accuracy, that would be very appreciated. --Syboxez (talk) 17:33, 5 March 2018 (EST)

Also have you tested CLK in Windows with mingw? It may work without any additional effort (I don't use Windows, so I can't test myself). --Syboxez (talk) 17:37, 5 March 2018 (EST)

You are correct in your assumption on the 2600 page about classification, which is why I hate using the term "Cycle" to describe an emulator's accuracy, as people assume that means perfect. For now, I would still put that classification as "Mid" as it was before, even though you are correct in calling it cycle accurate. Ever since bsnes, people have been taking the term "cycle accurate" to mean 100% perfection, when that is demonstrably not the case. This issue extends (perhaps to an even worse extent) to FPGA emulation. Plenty of inaccurate/incomplete FPGA cores attempting to emulate a console exist, but whenever you mention FPGA to someone in the retro gaming world, they generally take that to mean infallible. If FPGA emulation was infallible, then the Super NT wouldn't be getting firmware updates to correct bugs. Same goes for cycle accuracy; if higan was perfect, then there would be no updates to the emulator addressing accuracy. As for the Emulation Accuracy page, I tried adding the bit about higan and FPGA in there, and just forgot to remove the word "perfection" (if you couldn't tell, that part wasn't written by me). Updates to that page to correct any misconceptions are strongly encouraged, and I liked your section on subcycle accuracy quite a bit, although I may add a bit at the beginning so non-techies can understand a bit better. *Rant over (not really)* --Syboxez (talk) 23:06, 5 March 2018 (EST)

Agreed; 'cycle accurate' is virtually a marketing label at this point — it doesn't inherently tell you much about how indistinguishable an emulator is likely to be from a real machine. — Tommy (talk) 09:55, 6 March 2018 (EST)