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
For RNG with small state, maybe there is some performance boost when converting them into out-of-place version? Or just copy them in such a way that Julia's escape analysis can kick in? Not sure sure about this, though. (See https://github.com/sunoru/RandomNumbers.jl etc.)
The text was updated successfully, but these errors were encountered:
Use
randjump
or pre-calculated state for each thread in__reduce__
. (See: https://discourse.julialang.org/t/multithreaded-broadcast/26786/6, parallel RNG JuliaLang/julia#94 (comment), Per thread RNG state in multi-threaded code JuliaLang/julia#10441)For RNG with small state, maybe there is some performance boost when converting them into out-of-place version? Or just copy them in such a way that Julia's escape analysis can kick in? Not sure sure about this, though. (See https://github.com/sunoru/RandomNumbers.jl etc.)
The text was updated successfully, but these errors were encountered: