Jim GardnerOk I think I get it. numpy is base R
Christos Argyropoulos MD, PhD@jimgar but slower
@ChristosArgyrop The knives are out folks
@jimgar Somewhat tongue in cheek but this older benchmark shows a couple of things about base #Python vs base #Rstats vs #Perl and (surprisingly for those who don't know it exists) how fast the Perl authothreading library #PDL is vs #numba
Base #Rstats is highly optimized and the #Pythonistas unfortunately take a performance hit that they are mostly unaware about.
This and a few other other examples made me not transition to Python. #rstats ftw
The Quest for Performance Part II : Perl vs Python
Having run this toy performance example, we will now digress somewhat and contrast the performance against a few Python implementations. First let’s set up the stage for the calculations, and provide commandline capabilities to the Python script. ```python import argparse import time import math import numpy as np import os from numba import njit from joblib import Parallel, delayed
@ChristosArgyrop It’s funny you shred that, because I was thinking about Perl too. I’ve never used it, but 5 years ago when I started programming and doing basic bioinformatics, Perl was still spoken about wistfully. @scottishlass has been posting the Perl advent calendar https://perladvent.org/2025/index.html so there’s clearly still some interested parties around… including yourself!
@jimgar @scottishlass my stack nowadays is
#perl + #rstats on top of #clang , #sql .
I am using the former for application architecture development , management of data flows and for resource profiling of #rstats code
Given the extreme portability of both languages I don't have to worry about dependency hell (and CRAN/CPAN are better managed than any #Python repo).
#PDL is an extremely fascinating library especially for feature generation
Profiling Peak DRAM Use in R With Perl - Part 2
In the second part of this we implement the solution that was outlined at the end of Part 1: utilize a Perl application that probes the operating system in real time for the RSS (Resident Set Size), i.e. the DRAM footprint of an application fire the application from within R as a separate process, provide it with the PID (Process ID) of the R session and put it in the background do the long, memory hungry application upon the end of the calculation kill the Perl application and obtain DRAM usage for use within R
Allan Engelhardt@wiknin @jimgar @scottishlass Not only that but we just finished a two day hybrid conference Winter Perl Community Conference in applications (many scientific in nature).
https://perlcommunity.org/
(I delivered a talk on a GPU accelerated library that will be submitted to @joss at some point)
The community conference webpage gives an idea of the topics being discussed since its inception as a track of the Perl & Raku conference
https://www.papercall.io/perlcommunityconferencew25