You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
After testing, when the number of allocations is large enough (like 1‹‹22 ) , quininer/linkedhashmap will become very slow ( 4x ) . Is there any other appropriate improvement method?
PsiACE
changed the title
Convert to 0-unsafe code?
Convert to 0-unsafe code or try Raw API?
Mar 7, 2021
Another possible way to improve is to use the Raw API in hashbrown and griddle, I am not sure whether it can bring a performance improvement, but it is undoubtedly a challenge worth trying.
I noticed that @quininer has implemented a 0-unsafe
LinkedHashMap
.See https://github.com/quininer/linkedhashmap . It seems that 0-unsafe has no loss in performance.
I have no prejudice against unsafe, I just think it is a very interesting work.
The text was updated successfully, but these errors were encountered: