RAPIDS cuDF Accelerates pandas Nearly 150x with Zero Code Changes

Technical Blogs & Events Technical Blog
1

Originally published at: https://developer.nvidia.com/blog/rapids-cudf-accelerates-pandas-nearly-150x-with-zero-code-changes/

NVIDIA announced that RAPIDS cuDF can now bring GPU acceleration to 9.5M million pandas users without requiring them to change their code. Learn more about this feature and other updates at the AI and Data Science Virtual Summit on November 8. pandas, a flexible and powerful data analysis and manipulation library for Python, is a…

2

Cool!

Did you run the benchmarks using NVIDIA Grace Hopper? How much does it cost? I can’t find it

3

I see this is only for a Linux environment. Is there a roadmap for cuDf GPU acceleration in standalone Windows (local Jupyter notebook) without WSL.

4

I followed the instructions for WSL2, and it installed fine, but in Python it says

>>> import cudf.pandas
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'cudf.pandas'

Note that the cudf install verifies:

eafpres@EAFLLCML:~$ python
Python 3.9.18 (main, Aug 25 2023, 13:20:04)
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import cudf
>>> print(cudf.Series([1, 2, 3]))
0    1
1    2
2    3
dtype: int64
>>>
5

Fantastic performance, when can we expect the Windows version?