Hacker News Reader

You'd be surprised, there's a lot of low-hanging fruit in there.

Years ago I was poking at the 'blobby' generator code and managed to make it a lot slower by adding multi-threading -- once it overcame the overhead of multi-threading then it ran faster but, IIRC, that was not really within the typical use case so I just gave up. Undoubtedly I made some mistakes...

Having looked at some parts of the code, I think there's still a lot of room for improvement. There is a suprising amount of code which is similar to mentioned in the article where the approach is very complex in the name of optimisation but it's about as likely that it's making things slower (texture painting, for example, does something like 3 pointer indirections per pixel! And does chug a lot on my PC).

Animation tools have a ton of room for improvement, that they're working on. ...and the move to Vulcan will help a ton it looks like.

You seem to be arguing with somebody else. I didn't say anything that disagrees with your statement.