Submission #9694: yellowdice's NES Punch-Out!! in 18:51.94

Nintendo Entertainment System
Punch-Out!!
FCEUX 2.6.6
56605
50.0069789081886
9160
PowerOn
Punch-Out!! (E) [!].nes
Europe
82fd9bf5bba5c6fc1dbf39ee76aa73c2
Submitted by yellowdice on 5/10/2025 6:47 AM
Submission Comments
This submission is intended to be an improvement over https://tasvideos.org/9538S , which beat the PAL version of Punch-Out!! (the one featuring Mr. Dream) in 10:57.63 of in-game time. This TAS finishes the game in 10:55.79, with in-game time improvements in 4 of the 14 fights. Note that while Mike Tyson's Punch-Out!! and Punch-Out!! featuring Mr. Dream are practically identical (give or take a couple differences that have no bearings on the regular speedrun TAS that bans Clockstop Glitch), the PAL versions of MTPO and Mr Dream's PO are drastically different, likely due to the different measures taken when they ported each version to 50fps. Punch-Out!! PAL is almost 2 minutes faster than Mike Tyson's Punch-Out!! PAL due to some different mechanics, namely the PAL PO exclusive facejabs and gut punches used on Great Tiger, Bald Bull 1, Soda Popinski, and Mr. Sandman. It is also worth noting that this TAS prioritizes in-game time over real time, meaning that it aims to achieve the fastest real time completion of PO PAL while still getting the fastest (known) possible in-game time on each of the 14 fights.
Here are the times for each of the 14 fights:
Glass Joe - 41.97
Von Kaiser - 34.00
  • Using a facejab for the second punch is the same speed as a delayed misdirected left gut punch.
Piston Honda 1 - 33.82
  • In the previous PO PAL TAS, this fight finished on the 3rd frame of the .82 decimal. By replacing a left gut punch with a face punch through some rerouting (which is faster if the punch gets a star), I was able to save 2 frames. However, the fight now finishes on the first frame of the .82 decimal, meaning that no in-game time is saved. Additionally, this doesn't constitute a real time improvement either, as you're only able to progress from the blue screen after each fight every 8 frames, resulting in an 8-framerule.
Don Flamenco 1 - 14.61
  • Initiating the fight with a facejab is the same speed as using a left gut punch.
King Hippo - 35.00
  • Countering Hippo's open mouth punches with right gut punches is the same speed as doing so with left gut punches, except for the very last counter, where a left gut punch is 1 frame faster. I decided to use as many right gut punches as possible in this TAS because I thought it looked funnier :)
Great Tiger - 24.25
Bald Bull 1 - 47.82 (was 47.97)
  • This improvement comes from a slight reroute in Phase 1 of the fight. By reordering a face punch and a star punch, I'm able to cut out 3 frames of waiting time, which is enough to bump the fight a decimal lower.
Piston Honda 2 - 37.25 (was 37.97)
  • This improvement partially comes from cutting out a punch in Phase 1 and getting an extra star in Phase 2 through the reordering of some punches to account for the missing punch, but this only lowered the fight to a 37.82. The rest of the improvement comes from a frankly ancient technique known as the "Moment of Weakness". This was first used in the first ever MTPO TAS by Phil and Genisto back in 2004, in their 50.25 Honda 2 fight, before it was eventually obsoleted in 2013. The benefit that "Moments of Weakness" provide is that I'm able to replace a gut punch with a face punch, which is 1 or 3 frames faster (depending on whether the replaced gut punch is a left gut or right gut) when getting a star on Honda 2 (on PO PAL). This technique is often suboptimal as it takes a while to perform on NTSC, but the extra waiting time is significantly shorter in PAL, which allows me to perform a total of 4 of them across the fight.
Soda Popinski - 39.82 (was 39.97)
  • Right gut punch cancels are faster than left gut punch cancels by a single frame in PO PAL, and this 1 frame of advantage was just enough to bump the decimal down to a .82.
Bald Bull 2 - 1:02.61
Don Flamenco 2 - 55.82
Mr. Sandman - 1:13.00 (was 1:13.82)
  • This improvement might take a while to explain, but I'll try my best to keep it succinct: In the Sandman fight in PO PAL, a left gut punch that doesn't get a star is 4 frames faster than a face punch that doesn't get a star. The previous TAS of this fight used exclusively face punches (aside from 1 gut punch in Phase 3), which is significantly faster than using only gut punches as left guts only deal 4 damage to Sandman while face punches deal 5 (right guts also deal 5 damage but PO PAL exclusive right guts aren't a thing afaik, meaning you can't just buffer right gut after right gut unlike left guts or face punches). Sandman starts with 96hp in Phase 1, 88 in Phase 2, and 64 in Phase 3 (if you get the low refill), meaning that the final facejabs of Phases 1 and 2 only deal 1hp and 3hp respectively (since every other punch does 5 damage), which obviously means there's wasted damage in each of the phases (except Phase 3, because a 4 damage left gut is already used there). You probably know where I'm going with this.
  • It isn't just as simple as replacing 4 facejabs in Phase 1 and 2 facejabs in Phase 2 with left gut punches though, as of course it has to be more complicated. Sandman's guard starts in the down position at the start of the fight, which allows for a facejab to be buffered. However, for a left gut to land, his guard has to be in the up position, which requires waiting 6 frames to raise his guard, and another 6 frames for his guard to drop after throwing the gut punches in order to go back to throwing face punches (Sandman's guard retains its position across phases, so you can't negate this 6 frame wait by knocking him down with a gut punch). This could still result in a net gain of 4 frames by only replacing the facejabs in Phase 1 (4 frames x 4 left guts = 16 frames saved, 6 guard movement frames x 2 instances = 12 frames lost, 16 - 12 = 4) and omitting doing so in Phase 2 (because 2 extra left guts would only save 8 frames compared to the timeloss of 12 frames, which will result in a net loss of 4 frames), but it is possible to do significantly better.
  • Overshadowed by each facejab being buffered one after another is Sandman throwing Rolling Jabs, though none of them are ever seen as they are intercepted before he's able to finish throwing them. However, doing so requires delaying 7 frames every couple of punches (you don't need to do so on every punch as most of the time Sandman's move timer is still ticking down). With this knowledge, it's possible to work out that the most optimal course of action is to spend 6 frames raising Sandman's guard, delay 1 frame, use a gut punch to intercept a Rolling Jab, then delay another 6 frames letting Sandman's guard drop, and delaying 1 more frame before going back to throwing facejabs. This essentially means that the 6 frames required to raise Sandman's guard then let it drop is irrelevant, letting all 16 frames of improvement in Phase 1 and all 8 frames of improvement in Phase 2 be taken advantage of. This, all in all, results in 24 frames of improvement across the fight, which, because the clock speed is so slow on Sandman, only bumps the fight down 4 decimal points.
Super Macho Man - 37.82
  • The Macho fight in this TAS finishes on the same in-game frame as the Macho fight in the previous TAS, but finishes significantly earlier in terms of real time. By getting an extra star in Phase 1 of the fight, which cost an extra second, I'm able to reroute Phase 2 to finish with a star punch knockdown instead, which is noticeably faster than the gut punch knockdown used in the previous TAS. This was enough to save 3 framerules worth of real time over the previous TAS, which is actually 48 frames of timesave instead of 24, since you're only able to progress from the bluescreen after the Macho fight every 16 frames.
Mike Tyson (well, Mr. Dream actually) - 1:58.00
Final Time - 10:55.79 (was 10:57.63)
Thanks to mPap for helping me figure out how PO PAL exclusive left guts work on Mr. Sandman, and both he and DJ_Incendration for making the previous TASes of PO PAL, which acted as baselines for the creation of this TAS.
Last Edited by yellowdice 6 days ago
Page History Latest diff List referrers Change Log