Post subject: Native platform "emulation" such as PS3 rerecording on a PS3
Moderator, Senior Ambassador, Experienced player (898)
Joined: 9/14/2008
Posts: 1007
Greetings, This is a potentially controversial suggestion but I feel it warrants a theoretical discussion. I do not know if this suggestion is or ever will be plausible but it strikes me as worth exploring. With those disclaimers aside, here's my suggestion: Historically it has not been practical to emulate most advanced consoles within any reasonable timeframe after their release, although there have some been some notable exceptions such as the first N64 emulators. The PlayStation 3 looks to be as difficult to emulate as the Sega Saturn has been due to the sheer complexity of the platform and as such would appear to be a very unlikely emulation target. However, if the goal is to allow rerecording instead of pure emulation of the console it might be possible to implement the solution natively on the target hardware itself. In the case of the PS3, there have recently been an amazing number of breakthroughs in executing arbitrary code all set off by Sony's decision to remove the OtherOS feature from the console. The most recent development is the release of the PS3 private key by the fail0verflow group which ultimately allows arbitrary code to be signed and run as if it were legit software approved by Sony. This extraordinary key recovery development means it is theoretically possible to create a replacement hypervisor or other TSR code which can run on the PS3's reserved system SPE core and provide re-recording functionality for PS3 software. It is already possible for the current system software to pause game execution at any time by hitting the PS button on the controller (although this statement is a generalization); if it were possible to write out and subsequently restore the system state to the internal hard drive of the PS3 it is theoretically possible to implement a rough rerecording framework natively on the PS3 target platform itself. The key challenge would be making it deterministic which may be such a difficult problem as to make the idea impractical. I have no ability to develop anything like this myself nor do I have any idea if it is actually possible but it's such an intriguing idea that I wanted to throw it out there for discussion. Any thoughts? A.C. ******
I was laid off in May 2023 and could use support via Patreon or onetime donations as I work on TASBot Re: and TASBot HD. I'm dwangoAC, part of the senior staff of TASVideos as the Senior Ambassador and BDFL of the TASBot community; I post TAS content on YouTube.com/dwangoAC based on livestreams from Twitch.tv/dwangoAC.
Emulator Coder, Skilled player (1141)
Joined: 5/1/2010
Posts: 1217
Few of the major challenges are: * Keeping track of component state. It might not be possible to read the state of some hardware component. * Saving and loading system components in consistent state. * There are likely multiple free oscillators, and multiple free oscillators mean nondeterminism if those interact.
Joined: 1/5/2012
Posts: 52
Location: Maridia
Where are you going to store memory dumps? And how long will they take to create?