Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.programming.threads > #2787 > unrolled thread

Parallel sort and more about scalability

Back to article view | Back to comp.programming.threads

Contents

  Parallel sort and more about scalability Ramine <ramine@1.1> - 2015-01-02 15:29 -0800

#2787 — Parallel sort and more about scalability

Hello,


I have to be frank, we have to be smart when inventing or doing parallel 
sort algorithms, my new parallel sort algorithm is smart, cause it is 
more cache-aware, but you have to be carefull cause look
at my other parallel quicksort algorithm here:

https://sites.google.com/site/aminer68/parallel-quicksort

You have to know that this parallel quicksort that uses the
classical way of sorting is not so good, cause  when its partition
fonction is used recursively, this parallel quicksort algorithm will 
dispatch the arrays to be partitioned each time to different threads,
and this will make it less cache-aware than my new parallel sort 
algorithm, and since it will make it less cache-aware , so you will
not get much than 3X scalability by sorting strings and you will get 
less that 3X scalability by sorting integers or doubles for example,  So 
i advice you to use my new parallel sort algorithm of my parallel sort 
library version 3.3 that is more cache-aware and that gives you super 
linear scalability when sorting strings and it gives you good 
scalability when sorting integers and doubles etc.

You can download my new Parallel Sort library version 3.3 from:

https://sites.google.com/site/aminer68/parallel-sort-library



Thank you,
Amine Moulay Ramdane.

[toc] | [standalone]

Back to top | Article view | comp.programming.threads

csiph-web