Loading video...
Video Failed to Load
Pandas is a powerful data analysis and manipulation library for Python! NVIDIA just made Pandas 150x faster with zero code changes🔥 All you have to add is just a couple of lines of code: %load_ext cudf.pandas import pandas as pd
180,800 views • 2 years ago •via X (Twitter)
0 Comments
No comments available
Comments from the original post will appear here
![100x faster alternative to Pandas! (It can even beat GPU dataframe libraries) While Pandas is the most popular DataFrame library, it has some major limitations: - Pandas only uses a single CPU core. - It often creates bulky DataFrames - Its eager (immediate) execution prevents global optimization Introducing FireDucks a highly optimized, drop-in replacement for Pandas. Just change one line of code and you're good to go: ↳ import fireducks.pandas as pd The video below shows a comparison of FireDucks with cuDF—a GPU DataFrame library. In this case, it's even faster than cuDF. That said, the query in the video has chaining operations and it uses all columns. By manually optimizing and working only with the necessary columns, the run-time decreased to: - Pandas: 14 seconds (down from 48 seconds) - FireDucks: 0.8 seconds (unchanged) [SIMILAR] - cuDF: 0.9 seconds (down from 2.6 seconds) This demonstrates that FireDucks' compiler automatically performs the same optimizations that you would need to explicitly implement in cuDF and Pandas. Most importantly, the optimization does not affect the final result. I've shared a link to the Colab notebook in the next tweet. _____ Find me → Akshay 🚀 ✔️ For more insights and tutorials in AI and Machine Learning.](https://image.24vids.com/tw-1888205722451943477/ext_tw_video_thumb/1888205544659595264/pu/img/R-MrsueVTkmwcTNi.jpg)
