I'm going to be absolutely thrilled if we find out that for years the only legit way to do SNES speedruns was bsnes/higan.
Turns out SNES consoles might start running faster as they age due to one of the ceramic components resonating faster when warm due to breakdown.
For speedruns this hits vee with an absolutely terrible level of grim amusement after all the crap I went through dealing with "Emulator bad" thought termination. For actual preservation it's concerning because having a gold standard to compare experiences is really important even if it's hella inaccessible to most (not that the SNES is, but some consoles are).
@trysdyn I had to sit and wonder about why APU clock speed would have an effect on speedruns... then I realized, that probably affects communication between the CPU and SPC, especially if you're loading a considerable amount of stuff into the SPC's memory at once, doesn't it?
That could conceivably cause less lag if the APU clock speed increases, if that's the case...
@KS I actually don't know the super fine details myself. I know this caused a TAS validation to desync, and I know Dwango says it could impact long speedruns. They call out causing pitch bending in audio as well.
@KS @trysdyn the SPC700 is a lot slower than the 65c816 (which are on separate clocks as stated) and it's not uncommon for the 65c816 to wait in a loop for the SPC700 to be ready, or to have finished processing a command.
And having a variable amount of loop iterations isn't good when you have a prerecorded stream of inputs that make assumptions about the game behaving in a deterministic way when given those inputs; I've seen games do things like continuously tick a random number generator with all their free time, which would mean timing has to be exact.