In a post very old (" processing large volumes - Comparison Software "- September 2008), I compared the behavior of several software when processing a file with relatively large decision trees.
I was describing inter alia the conduct of Tanagra 1.4.27 version released in August 2008. Since my development machine has changed Tanagra itself has changed, we are so far to version 1.4.37, and Sipina has also been modified (version 3.5), with the introduction of multithreading induction techniques for certain trees. I thought it was time to study the performance by re-editing experiments in the same conditions. On
Sipina Tanagra and the only software I have analyzed in this new, improved processing time is obvious. After, we must discern what is attributable to the change machine, which amounts to changes in implementations. We propose some tracks in our document.
The new results were added in the last section (Section 5) of the PDF.
Link: fr_Tanagra_Perfs_Comp_Decision_Tree.pdf
0 comments:
Post a Comment