Emulator: mame-rr-0.139-v0.1-beta
Modified buttons: User Interface - Reset Game (kbd E), Soft Reset (None), Throttle (None), Fast Forward (None), Show FPS (None), Save Snapshot (None), UI Up (kbd Up), UI Down (kbd Down), UI Left (kbd Left), UI Right (kbd Right), Frame Advance (kbd Space), Toggle Read Only (None), Lua Script Window (None), Load State Dialog (None), Save State Dialog (None), Load State 1 (kbd F1), Load State 2 (kbd F2), Load State 3 (kbd F3), Load State 4 (kbd F4), Load State 5 (kbd F5), Load State 6 (kbd F6), Load State 7 (kbd F7), Load State 8 (kbd F8), Load State 9 (kbd F9), Save State 1 (kbd 1), Save State 2 (kbd 2), Save State 3 (kbd 3), Save State 4 (kbd 4), Save State 5 (kbd 5), Save State 6 (kbd 6), Save State 7 (kbd 7), Save State 8 (kbd 8), Save State 9 (kbd 9), Play Movie From Beginning (None), Stop Movie (kbd F10), Start Movie Recording (kbd F12), Start Movie Playback (kbd F11), RAM Search (None), RAM Watch (None). Player 1 Controls - P1 Button 1 (kbd Right Control), P1 Start (kbd Q), P1 Select (kbd W). Other Controls - 1 Player Start (kbd Q), Coin 1 (kbd W).
Checksum of my ROM: 2E7C565EBE62348B9B10221AE418A4279FE4E01E
BIOS: none? That is what ChatGPT told me
Glitches: This run uses the ladder glitch, which lets Mario partially climb a ladder to bypass collision checks and skip sections of the stage, also affecting RNG timing.
Goal: The goal is to complete one full loop by rescuing Pauline at the end of the Rivets stage as quickly as possible.
Game Description: Donkey Kong is a 1981 arcade platformer developed by Nintendo. Players control Mario as he climbs construction girders to rescue Pauline from Donkey Kong. The game features four unique stages—Girders, Elevators, Rivets, and Conveyor Belts—each with different obstacles and enemies. It’s one of the earliest examples of a story-driven action game and marks Mario’s first appearance.
Samsara: Claiming for judging.
Samsara: Hey, welcome to TASVideos.
The JP version used in the published run is preferred for this game as it plays through all four unique stages in a single loop. This single loop on the US version is effectively just half of the game as a result. Even on a multiple loop run, the JP version goes through the exact same 4 stage loop while the US version chops and screws the order significantly, notably repeating 25m and 75m in later loops.
On the note of the published run, it uses a number of tricks that you don't use here, meaning this run is noticeably and provably suboptimal in comparison. All known tricks should be used in the creation of a run, and since you're also using MAME-rr you should easily be able to take the published run and study its inputs. Admittedly, we'd prefer that Bizhawk or libTAS loading MAME be used, as they're not only using much more up to date versions of MAME, but they're still being maintained to this day and should be a lot more user-friendly to boot.
...Also, don't use ChatGPT for submission texts. Submission texts are extremely important for us to understand what runs do and how they work, and using AI of any sort adds an extremely high likelihood of misinformation that can and will affect how the submission itself is processed.