Submission #6218: DrD2k9's C64 Donkey Kong (Atarisoft) in 01:47.81

Donkey Kong

Goal

• Beat all unique sub-stages as quickly as possible.

Game Notes

• This particular port of DK has all four sub-stages seen in the arcade version. This port, however, does not include the "How high can you get?" screen with stacked DKs.
• Tricks
  • Ladder Boost - Only when traveling from left to right, climbing a ladder for 1 frame then immediately returning to the ground positions Mario further to the right on-screen than he would have progressed simply running past the ladder. This works with both ascending and descending for the first frame (broken ladders cannot be descended). This trick is used in both girder stage and rivet stages.
  • Spring pseudo-clip - While jumping normally loses a pixel of horizontal distance compared to running the same number of frames...On the top of the elevator stage, it is possible to jump toward an oncoming spring so that Mario can progress past the spring without being hit even though the spring would hit him if he were just running. This is not jumping over the spring (which I couldn't accomplish anyway), it's jumping through it. This trick is not used in this submission.
  • No other beneficial glitches were found in this port.

DK for C64 can be considered endless.

• There's no kill screen (that I could find in my testing).
• Once level 99 is beaten, the game drops to level 0 then back to level 1 and loops.
• Contrary to the game manual, DK does not increase the rate of barrel rolling at higher levels.
  • In all levels there is a 90 frame wait for DK to turn to grab a barrel. DK turning back to center requires 39 more frames. At this point he will either:
    • Immediately drop the barrel and the 90 frame counter then starts over. The barrel will fall either straight down or along various angles downward.
    • Turn a second time (after 39 more frames) to roll the barrel along the girders. He then takes 24 more frames to return to center for the 90 frame counter to start.
• There were, in fact, no discernible difficulty variations between various levels.

Determining game-end

• Level 1 - Girders, Rivets
• Level 2 - Girders, Conveyors, Rivets
• Level 3 - Girders, Conveyors, Elevators, Rivets
• Level 4 - Girders, Conveyors, Girders, Elevators, Rivets
• Level 5 - Girders, Conveyors, Girders, Elevators, Girders, Rivets

HOWEVER!!!!

• There are 8 total values for the girder stage.
  • The only discernible difference between these is the direction of the first barrel throw; all other gameplay mechanics are the same.
• There are 3 values for the conveyor stage.
  • I couldn't identify any specific variations.
• There are 4 values for the Elevator stage.
  • The first spawning of a spring seems to have a slight difference in timing between these.
• There are 5 values for the Rivet stage.
  • No identifiable variations.

• Generally we allow/encourage starting at the highest difficulty selection available from the beginning of a game. Which would be level 5 in this port.
  • But as already mentioned difficulty itself doesn't vary between levels, so is starting at Level 5 actually a 'higher difficulty' than starting at Level 1, 2, 3 or 4?
• Then there's the issue of unique content...rules state that the game must reach a point where no new content will appear. Thus it must be determined what is considered unique enough for publication purposes:
  • If unique can be considered visually unique, then this submission (starting at level 3) should qualify as the fastest method of seeing all visually unique content; even though it neither plays the 'hardest' difficulty available from the start screen nor plays the last available unique sub-stage memory identifier.
  • If however, unique content is based on the technicality of sub-stage memory identifier; the only way to truly have no further unique content is to complete Level 5. This approach however, results in a longer more repetitive run due to the 2 additional girder sub-stages which themselves offer nothing new or unique visually.
    • If it's determined by the community/staff that this is the necessary approach, I have also completed a run using this method available here. Some tricks are/aren't used in that run compared to those used in this submission.

Special thanks to Fortranm