GPU Accelerated data processing and Machine learning workflows with RAPIDS

Accelerating Data Processing with cuDF -> Pandas-like GPU DataFrames

If you know Pandas, you can switch to cuDF, a RAPIDS library, to process data on GPUs. most Pandas code will run with minimal changes by simply replacing pandas with cudf.

Example:

# CPU-based Pandas code
import pandas as pd
df = pd.read_csv('data.csv')
df['new_col'] = df['col1'] + df['col2']

# GPU-accelerated cuDF code (zero code change besides import)
import cudf
df = cudf.read_csv('data.csv')  # Uses GPU instead of CPU
df['new_col'] = df['col1'] + df['col2']  # Same operations, much faster