Re: Native Map faster than MCE::Map.

I think we had the same question not long ago.

Point is that parallel jobs have a communication overhead, hence MCE::Map will only show it's potential if the jobs themselves are considerably heavier than the overhead.

edit

yep, see I thought MCE::Map is faster than usual map. What am I missing?

update

next time please use <code> not <pre> tags

Cheers Rolf
_{(addicted to the Perl Programming Language :)

Wikisyntax for the Monastery}

Comment on Re: Native Map faster than MCE::Map. Select or Download Code

Replies are listed 'Best First'.
Re^2: Native Map faster than MCE::Map. by Anonymous Monk on Jun 21, 2021 at 16:35 UTC
Sure, I will use the code tag next time. As for the heavier workload, if I use a higher number, the program is "killed" by the OS (with or without MCE::Map). So not sure how to generate a "heavier" work load. Any pointers would be helpful.	[reply]
Re^3: Native Map faster than MCE::Map. by LanX (Saint) on Jun 21, 2021 at 16:42 UTC
The job has to be heavier not the number of jobs. `$_ * $_` is simply to fast to justify the overhead. Please read the thread I linked to AND the documentation of MCE::Map ! Cheers Rolf _{(addicted to the Perl Programming Language :) Wikisyntax for the Monastery}	[reply] [d/l]