epestr reports speeding up the BF JIT interpreter and produces a render
2Yesterday 8:33 PM · 17h ago · 2 posts · 2 comments · 2 sources · development 2 of 2
Responding in the comments, the author said the render could be made a reasonable benchmark by dropping to one sample per pixel at a quarter resolution, cutting the workload roughly 1600x from an estimated ~100 days to under one day, and reported having sped up the JIT interpreter, based on TSoding's JIT compiler, producing an actual rendered image described as 'slightly artsy/funny looking' in an update to the post.
“I've also been working on the JIT interpretor I used and managed to speed it up a lot and got a render out of it.”
epestr
epestrAuthor of the Brainfuck ray tracerTSodingAuthor of the JIT compiler epestr's interpreter is based on
The whole story articlespostscommentsthe bright band is this development · numbered dots are the others · click one to jump
I must sleep soon, missed the additional. If you want it to be a reasonable benchmark, you could change the source `ray_ssa.c` to 1 sample per pixel and a quarter of both the width and height. That should reduce the work by 1600x (from ~100ish days to <1) I've also been working on the JIT interpretor I used and managed to speed it up a lot and got…
I've been tring to optimize expressions too. My implementation is based on TSoding's JIT compiler. I've just added a license instead of repackaging it.