Yesterday’s tool-assisted speedrun (TAS) block at Summer Games Done Quick ended with a Super Mario Bros. 3 run that lasted barely longer than a blink.
The ridiculously rapid run is the result of a newfound glitch discovered by TAS speddrunner ais523, which basically involves pressing buttons at an extremely fast rate — some 6,000 times a second. Ais523 explained the glitch on Reddit following the run.
…if you’re reading the controller repeatedly until you get two values the same (in order to work around the DPCM/controller conflict), then if the controller reads a different output each time (because you’re mashing the controller really fast), it’s going to get stuck in a loop, potentially allowing for the code that handles the start of a frame running recursively. If the game isn’t designed to expect that to happen (and if the code in question isn’t really laggy, why would it?), bad things happen, and it was a case of finding a game in which the bad things in question would happen to let us win instantly.
Aided by total_, ais523 got the glitch working with Super Mario Bros. 3. The run was performed by TASBot, the Nintendo R.O.B. robot modified by marathon regular dwangoAC to run controller instructions.The game began, the game ended, and people cheered, as seen in the YouTube video posted by Countryclubguy.
You can catch more of the Summer Games Done Quick TAS block over at the event’s Twitch archive.
Comments
6 responses to “Robot Finishes Super Mario Bros. 3 In Two Seconds”
So thats the any percentage speed run covered for a while…
Exciting stuff..
When speed running was new, it was impressive to see people finish something like Quake in 15 minutes because I’d played the game and knew how long it took me to finish (Longer than that).
Finding hacks like this one that just break a game don’t count as any kind of speed run in my book. It just seems pointless.
That’s why there are categories. Any% versus Any%-no-glitch, for instance.
The point is the technical exercise of not just finding the glitches, but figuring out how to exploit them quickly and reliably. The point of the Any% category *IS* to break the game.
Fair enough, whilst I can see why the hack ones are a technical challenge I find them annoying to watch. This article is good cause it shows the cool shit people can find 20+ years later. But I hate watching a speedrun and then halfway through a game they jump a portion due to a glitch. I’d personally call that a glitch run and not a speed run, but that’s semantics.
I guess all I have to watch out for is ‘no-glitch’ speedruns and that should do the trick, cheers.
Yeah, no-glitch or glitchless depending on which particular site you’re on. It’s not as popular a category as Any% though, because they tend to be quite boring.