Super Nintendo consoles appear to be running ever-so-slightly faster as they age and speedrunning detectives are hot on the case
The legendary Super Nintendo Entertainment System is 35 years old and still a mainstay in every classic gamer’s heart. It’s also much beloved by the speedrunning community, which has found itself in a bit of a quandary recently—as it seems the elderly console might run slightly faster as it ages.
That’s according to 404 Media, which has been speaking to TASBot administrator Alan Cecil. Cecil has been collecting data from BlueSky users as to how fast their SNES consoles are currently running and, while the dataset is small, for now it does appear that some ageing consoles are showing small, but not insignificant speed boosts compared to when they were first released.
The phenomenon appears to be related to a Sony audio processing unit called the SPC700. Documentation provided to game developers in the 90s suggests that the DSP rate of the SPC700 should be 32,000 Hz, which is governed by a ceramic resonator.
However, ceramic resonators have been known to run at higher frequencies under certain condition changes, like higher temperatures, and it appears that ageing might also change the speed at which it processes audio data.
How much of a change? The fastest SPC700 recorded so far was running at 32,182 Hz. That’s not much of a difference compared to stock, but in terms of speedrunning, there’s some concern that it might have a noticeable impact.
In a scene transition, for example, if the audio data loads faster than its supposed to on some consoles compared to others, that might translate to a fractionally faster load time—and these sorts of millisecond differences might result in a slightly faster speedrunning result overall compared to emulated results, which now run at 32,040 Hz to account for some SNES systems running faster than others.
So does this translate into meaningful, real world gains for human speedrunners? Probably not, according to Cecil. However, for accurately emulating tool-assisted speedruns, where inputs are tied to specific frames, it might make a difference. The jury’s out in this regard, but Cecil is continuing to gather data to document the effect and provide more information on how to accurately emulate SNES consoles in future.
A similar analysis suggests the emulated difference between a theoretical 31,900 Hz and 32,200 Hz DSP sample rate would be 7.8 ms, which would equate to less than half a frame.
Ah, nerdery. You’ve got to love it. Apparently due to the temperature-sensitive nature of the ceramic resonators, you can actually change the speed at which a SNES console runs by sticking it in the freezer overnight, which sounds like a terrible thing to do to a now-much-sought-after piece of old-school tech.
Still, performance is performance, and I can only salute the level of meticulous data-gathering and debate going on here. The speed at which Super Metroid should technically run might be the smallest of details, but in these topsy-turvy times, it’s nice to see immense accuracy being applied to this most gamer-like of endeavours.
