Speed Kills

[xelnia]

Unofficial Speedrun Leaderboard

1) Daniel Desjardins: 1:16:40* (Arcade) - 8/10/2014
2) Dean Saglio: 1:19:39 (MAME) - 2/23/2011
3) Vincent Lemay: 1:21:11 (MAME) - 2/23/2011

*The broadcast of Dan's game is broken up into 2 minutes chunks (Twitch pls), so if he re-broadcasts it I will link to the highlight.
---------------------------------------------------------------------------------------------------------

As the 2014 NH Tournament was getting started I began to wonder about DK speedruns. The topic came up in Ross' stream last night so I thought I'd post some questions and information to see if anyone is interested in this "variation."

First, it seems the two fastest known killscreens are by Vince and Dean. Vince's game can be seen here: Video Game World Champion on Donkey Kong: First Man Kill Screen Speed Run! Subscribe! I don't think Dean's game is available to watch anymore, but the details for it can be found in this CAGDC thread: Speed run for Donkey Kong: 1h19 on my first man! Ironically, that thread was started by Vince to announce his own speedrun time (thanks Ben, for pointing out this thread).

Dean measures his overall time from when he presses Player 1 Start to when "Game Over" appears on the killscreen. Arguments can be made for what it means to reach the killscreen (final rivet? break screen? loading of girders? first control of Jumpman? first activation of the bug? final activation of the bug on the last man? "Game Over"?), but I feel that Player 1 Start -> "Game Over" encompasses the FULL game and should be used for any speedrun times.

Using Player 1 Start -> "Game Over" we get these times for Vince and Dean:

1) Dean (self-reported): 1:19:39
2) Vince: 1:21:11

Vince, if you read this: do you still have the INP for this game?

So, is anybody interested in seeing a speedrun high "score" list? What kind of rules would we need? INPs would be easy to verify and the internal frame counter could be used for exact times. Old games could even be used to populate the list if people were interested in being on it.

I've played around with this for about a week and here are some "best segments" from my games:

Player 1 Start -> Break Screen after L4 (sum of best): 7:58.99
1-1   40.05
1-2   42.04
2-1   28.31
2-2   21.37
2-3   44.18
3-1   29.96
3-2   23.47
3-3   23.20
3-4   54.70
4-1   31.65
4-2   23.57
4-3   30.55
4-4   28.32
4-5   57.62

Best L5+:
Barrel: 29.59 (x51 = 25:09.09)
Pie: 23.37 (x17 = 06:37.29)
Spring: 24.34 (x17 = 06:53.78)
Rivet: 55.92 (x17 = 15:50.64)
Sum of best L5+ segments = 54:30.80

Sum of all "best segments" before killscreen: 1:02:29.79

I'll continue to play around with this even if there isn't widespread interest...what say the rest of you?

[VON]

There's definitely interest.

[Xermon54]

Vincent betting 100$ for someone to beat me and Dean does it the next day. This bet started all my bet that I lost! With Vincent Lemay math, I thought nobody would beat my time since I reached the kill screen on my first man, but I guess Vincent Lemay math failed again! And yeah, as usual, I never respect any bet, I apologize Dean for always making fake bets  .

But joke aside, this is a cool track. You need a lot of good luck to get a good speed run time, and every kill screeners can have a shot at getting the record. But this track is also very hard, since you always have to decide whether or not you want to take the risk to finish the board faster while taking some risks because if you die, it will take you a lot more time to restart the board!

I remember Dave reaching like stage 18-19 on a 1h15 pace. A sub 1h15 is definitely the goal in speed run dk, it's fking hard!

[TheSunshineFund]

Top shelf elevators all day

[LMDAVE]

I do wish a speed run would get more attention. Yeah Vince, the few times I tried it I had one run that  went to 17-5 and was on pace for 1:15:00. I did always wonder what a good theoretical min would be for a speed run. 1:02:00 is a good calculation which means, a sub 1 hr would be possibly attainable and phenominal!

[marinomitch13]

Hey, Jeremy, did you happen to keep enough data to get a good average of your times per level/board? I think that would also be important to know, as we'd be able to start figuring out deviations from the average and knowing when/when not to attempt significant YOLOs with each man. My gut thought is that yolo-ing would be best on the pies, as it offers the least amount of loss but a lot to gain.

[xelnia]

Hey, Jeremy, did you happen to keep enough data to get a good average of your times per level/board? I think that would also be important to know, as we'd be able to start figuring out deviations from the average and knowing when/when not to attempt significant YOLOs with each man. My gut thought is that yolo-ing would be best on the pies, as it offers the least amount of loss but a lot to gain.

I haven't done enough deep runs to get good averages beyond level 3 or 4. I've been focusing on a speedy start and have done quite a few YOLO runs trying to get to Level 5 as fast possible, but even then I've had significantly fewer runs of, say, L3 rivets than 1-1. My averages for L1-L3 are:

Splits begin with "How High Can You Get?" break screen:
Start 12.90 (should be equal for everyone, this is the first loading screen)
1-1   29.63
1-2   46.56
2-1   29.82
2-2   22.90
2-3   49.41
3-1   34.36
3-2   23.54
3-3   24.71
3-4 1:00.35

This is probably after 100 or so successful 1-1s, but only a couple dozen 3-4s...

Also, the above splits/times in both of my posts are all based on me manually hitting the split button. Tonight I was able to create an AutoHotKey script that will scan for the break screen and automatically hit the split button for me. It seems to be accurate within a few frames...much better than me reacting to or anticipating the screen change.

[danman123456]

Arggg just did 1:20 Speedrun with THREE DEATHS!!! Could have beaten Dean's time but I died on 21-6 due to being a bonehead - I'm pretty sure that 1:19:39 is going DOWN cuz I had some realllllly bad boards too where I went almost to zero on the timer! Dan Flexes! 

http://www.twitch.tv/danman123456/c/4262284

[xelnia]

Arggg just did 1:20 Speedrun with THREE DEATHS!!! Could have beaten Dean's time but I died on 21-6 due to being a bonehead - I'm pretty sure that 1:19:39 is going DOWN cuz I had some realllllly bad boards too where I went almost to zero on the timer! Dan Flexes! 

http://www.twitch.tv/danman123456/c/4262284

Love it, Dan! 

[xelnia]

Layout I'm using for LiveSplit and **rough** splits from Danman's game that I'm comparing to. lit 

Added AutoHotKey script that I'm using. This is setup specifically to my screen resolution and the colors in WolfMAME 0.153. You'll need to adjust to your needs.

[danman123456]

Wow awesome. I'll try this and just hotkey em in one at a time

[f_symbols]

I hope Jeremy doesn't mind me posting this here 

http://www.twitch.tv/f_symbols/c/4433621

07:50.63 end of level 4 time

[andrewg]

I've been inactive with DK for some time, but I may actually do a somewhat serious speedrun for DK. I'm not really one for point pressing...


EDIT: Since I'm a speedrunner more than I am a high score player, here are my thoughts after some small discussion on this:

1) A speedrun of this game should obviously reach 22-1.
2) Time should end upon death from the game timer glitch from the kill screen itself, and not just any death in 22-1.
3) The player should kill off any remaining lives and enter initials.

I feel this results in the cleanest finish, meets the goal of fastest "kill screen death from kill screen," and results in the shortest speedrun time total.

EDIT2: I also think timing should start at control of character (SDA timing set), which I completely agree with considering sometimes title screen times can vary for some games. It's a good standard.

[danman123456]

Yeah I concur the only difference is my time started when I hit the 1 Player start vs "control of the character". Everything else was after my final death at the KS of which at least one death is due to the timer killing you.

Dan

I've been inactive with DK for some time, but I may actually do a somewhat serious speedrun for DK. I'm not really one for point pressing...


EDIT: Since I'm a speedrunner more than I am a high score player, here are my thoughts after some small discussion on this:

1) A speedrun of this game should obviously reach 22-1.
2) Time should end upon death from the game timer glitch from the kill screen itself, and not just any death in 22-1.
3) The player should kill off any remaining lives and enter initials.

I feel this results in the cleanest finish, meets the goal of fastest "kill screen death from kill screen," and results in the shortest speedrun time total.

EDIT2: I also think timing should start at control of character (SDA timing set), which I completely agree with considering sometimes title screen times can vary for some games. It's a good standard.

[Barra]

Probably worth posting
http://www.twitch.tv/barranz/v/15828732
Got to 18-2 on pace for under 1:10

All top shelf jumps on the elevators. Missed 1. 3 pies deaths
6500 elevator average with no prizes

I think 1:07 is possible.