Gregory Popovitch

Yes, makes sense. When I played with this repo a while back, I noticed that the benchmarks had variability from one run to the next, so I'm not sure how...

Why does it matter that it is branchless if it is slower?

I believe this version is significantly faster!

Thanks for looking into it @chfast ! I'm not sure how to read the numbers you posted, but I am surprised that you found that add/sub got slower. In my...

Also why diff between add and inline_add?

Thanks! > the "inline_add" can be vectorized vectorized or inlined? Why shouldn't it be relevant for EVM?

@chfast if you are using gcc 11 and therefore the builtin, I am puzzled that my version would be slower.

Wow, this is great, I really appreciate it @bashimao ! I did not investigate the AVX2 code, but the change looks very safe as it is conditioned with by a...

@bashimao I'll check this evening after work how much this improves performance.

@bashimao I did some performance testing and unfortunately I cannot see any performance improvement with the avx2 implementation. Maybe it would show only when the hash map is close to...