*This is a tool-assisted speedrun!*

For https://dsdarchive.com/

Author: RockyGaming4725
Date: July 6, 2026
IWAD: plutonia.wad
PWAD:
Map(s): 03
Skill: 4
Category: NoMo
Sourceport (Tool): XDRE-R v2.24
Sourceport (Video): DSDA Doom 0.29.4
YouTube: https://youtu.be/Qwx5mHoGN1A

Time: 23.23
Comments:

After almostmatt1 and I finished the Plutonia D2ALL, I knew there was room for improvement on this map in the NoMo category. I started rebuilding this over a year ago. The first file saved from this specific demo was in May 2025. While over a year ago, I wasn't actively working on this for a while. In fact, it took me almost a whole year to open up my XDRE build to work on this horrifying nightmare. However, the break was long enough to pretty much inspire me to finish this project.
I'm honestly very happy with the result. I saved frames in nearly every section compared to 30plx854, and I have some very impressive diagonal wallruns, made possible by what I've done to XDRE-R 2.24 - I wrote some code that lets me choose to use XDRE's native brute forcing, or parse the checks and commands in the XDRE boxes, split the range (VERY simply.. Matt's done an excellent range splitting job with BRS), and spawn however many DSDA-Doom processes I wish. With some Linux terminal commands, I can also get each of these processes to use more CPU power, which I need to be a hair cautious with as my CPU can easily become melted cheese from that sort of thing. I'd clean up the code to make it part of an official release, but to be honest I don't have enough time, and I'm very lazy when it comes to programming. It doesn't help I haven't even opened the source code in almost 2 years.

I'd take a guess and say that maybe 30-40% of tics in this demo are brute forced. Most the smooth turning into speed redirection collisions were brute forced, too. And yeah, maybe you could argue that if a major portion of this demo was brute forced, then there's not a lot of TAS art in it. But, I'm the TASer of this demo, and I say this demo is freaking beautiful. :)

So, is 22s possible? I'd say most likely. While it might be doable with current tools, including brute force range splitting and multithreading, whether that's with Matt's clever BRS or with my less clever implementation into XDRE-R (Not currently available. Sorry), the next second barrier would probably be easier broken by using a smarter movement calculation tool, maybe when the JaffarPlus TAS bot tool can be used by some intelligent individual (that isn't me).

This may be my last TAS demo ever. I'm heading to university 10 hours away and I think bringing a single ounce of Doom TASing into my Macbook would be unhealthy for me haha. and I say this as I've already tried to compile XDRE on it...

I don't have any unfinished demos. I may have a few completed ones that I never uploaded, but those are definitely not optimized enough for me to officially share them.
So, for real, as my life is about to transition, it may be fitting that I finally refuse to launch XDRE with any serious intent again.

30plx854 from Matt and myself is what I'm most proud of. But this demo may be a close contender for 2nd, just for how fast it is. :)


Cheers,
Rocky

rockygaming4725@gmail.com
