add bitcoin benchmarks from 2025-12-10 #326
Conversation
| "best_value": 785.51375, | ||
| "median_name": "62", | ||
| "median_value": 800.2093333333335, | ||
| "worst_name": "17", |
There was a problem hiding this comment.
Why are some of these names like "17" and "62"?
There was a problem hiding this comment.
I guess they have always been like this.
There was a problem hiding this comment.
I think these ones have no name (there is a single unnamed distribution) and the name-parsing code gets confused by it.
roconnor-blockstream
left a comment
roconnor-blockstream
left a comment
There was a problem hiding this comment.
LGTM.
Probably FeNormalize is the most notable outlier, going from around 289 to 360, but even this is probably acceptable. I don't think there is any reason to believe FeNormalize's benchmarking has changed in any fundamental way.
|
Great, thanks! Yeah, I suspect these low values are just noisy. (Though it does seem like the distributions are all pretty tight.) It is certainly plausible that the marshalling code has changed since 2024 in ways that would affect these values. When I get a chance I can try to bisect this. Anyway I think we should consider the new ones to be more authoritative than the old ones. |
|
|
2 participants
No description provided.