Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Almost sounds like savestates don't properly save/restore memory card state?
Is it possible to dump memory card so one could try to analyze what is going on?
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
IIRC, all accepted do, except M64, DTM (input frame based), JRSR and OMR (time-based).
I think some might also have deterministic 3 latches per frame.
That would only work on games that poll a fixed number of times per ingame frame and don't have anything depending on number of video frames (this is why SM64 was verifiable!). Otherwise clocking still matters.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
MP3 and AVI hasn't been used for a long time (last use I can find was June 2009 for both).
Any AVI is subject to re-encoding (there are also 7 MKV MP3 encodes).
You mean Vorbis, right?
That's what the current guidelines are (first use of Opus was 2460M, September 2013).
Also, MP4 is disfavored among encoders due to the amount of BS required to produce dedupped MP4 encode.
Thread #14314: Opus codec quality testing
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
I don't think it would be. Specifically, some of the problems:
- The APIs normally exported by the hypervisor.
- Stopping the execution in deterministic manner (including the SPUs).
- Saving and Restoring GPU state (much easier on CPUs, as those are designed differently).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Obviously any movie using reset emulation requires a new version, but it could be possible to write old-format movie if there are no resets (as long as there is a way to tell formats apart).
However, this might not be easy, depending on how exactly Gens writes the movies...
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
ZIP the DSM file before uploading, and the 4MB limit is for the zip itself, not the file inside (there are some more powerful compressors supported, but those are more exotic).
Submissions also have the same 4MB limit.
Also, the the files are stored compressed, and quota usage is calculated from compressed size.
As to if binary formats after compression would be smaller: Probably, but the difference wouldn't be that great.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
At least with some systems apparently whatever electrical protocols of controllers can represent is fair game (including messing with controller type identification).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
If A and C are near enough (which might actually be over hundred pixels apart, don't know about the API), fast enough mouse motion can jump from one to other in one step, bypassing B.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Note that libopus 1.1-beta (and 1.1 when it is released) should be used instead of libopus 1.0.X.
Use 'opusenc --version' to see what version it is:
That's using 1.1-beta.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
I didn't compute that yet, but I looked at what kind of distribution time between submissions has.
Looks like one of those ill-behaved ones.
For intervals of about 1.5 min to 24 hours, the intervals grow roughly proportional to ~1.6th power of rank. For intervals longer than that, the intervals grow even faster (the slope of log-log plot increases rapidly).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Whole emulated system memory or whole host process memory?
The first isn't the complete state (there are e.g. CPU registers), and the second contains all sorts of irrelevant stuff (that likely differs even if states are identical).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
It is 9, not 6. But all those .fm2 files should have identical input tracks, so any should work with any of those 9 games (albeit 8 get wrong ROM warning).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Different emulators make it impossible, because there is no one movie file all the relevant emulators (1-6 is NES, 7 is SNES, 8 is PSX and 9-10 is Wii) could play.
It even wouldn't be possible (at least currently) to do multi-runs across systems even with one emulator (some support more than one system), because the system is written into movie file.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Nope, it is not, there is single correct frame rate per system,region pair. If the correct value is known is separate issue.
Previously, each emulator had its own framerate, now all files for given system (and region) are evaluated with the same framerate.
There are loads of publications affected by the framerate issue. E.g. Any older .fmv, .fcm, .smv, .vbm.
There are also some pubs where publication time and submission time slightly mismatch because the framerate change (ugh, didn't see that).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1142)
Joined: 5/1/2010
Posts: 1217
Also, for NES, there are so many dodgy mappers that support in anything except original cart is questionable.
SNES is much more regular. Unless game uses expansion chips, pretty much the only source of variability is memory mapping (which can be expressed as a few memory address ranges and some properties of those).