FireDucks

Compiler Accelerated DataFrame Library for Python with fully-compatible pandas API

Get Started
import fireducks.pandas as pd
News
Release fireducks-1.1.6 (Jan 07, 2025)
"Pitfalls of Time Measurement for FireDucks with %%time in Notebooks" is published (blog) (Dec 26, 2024)
"Ensuring compatibility with pandas in the GPU version of FireDucks" is now available(blog) (Dec 19, 2024)
"Exploring performance benefits of FireDucks over cuDF" is now available(blog) (Dec 18, 2024)
"Cache or Eliminate? How FireDucks increase opportunity of optimization" is now available(blog) (Dec 17, 2024)
Updated TPC-H Benchmark: 125x average speedup over pandas, 2.2x average speedup over polars (Dec 06, 2024)
"How to run polars-tpch benchmark with FireDucks" is now available(blog) (Dec 06, 2024)
"Unveiling the Optimization Benefit of FireDucks Lazy Execution: Part #2" is now available(blog) (Dec 05, 2024)
"Unveiling the Optimization Benefit of FireDucks Lazy Execution: Part #1" is now available(blog) (Dec 05, 2024)
"What to do when FireDucks is slow" is now available(blog) (Nov 15, 2024)
Accelerate pandas without any manual code changes

Do you have a pandas-based program that is slow? FireDucks can speed-up your programs without any manual code changes. You can accelerate your data analysis without worrying about slow performance due to single-threaded execution in pandas.

Sustainable and Economical

Concerned about the cost and environmental impact of cloud computing? Our acceleration technology reduces cloud usage fees, while minimizing the CO2 emissions at the same time, making FireDucks an environment-friendly and wallet-friendly choice.

Reliability and Performance

FireDucks is developed by infusing the essence of supercomputers that NEC has refined over the years. Made in Japan, high-quality FireDucks promises reliability and high performance.

Features

Multithreaded ✨

FireDucks is multi-threaded, enabling higher speeds on multi-core CPUs. More

JIT Compile ⚡

A runtime compiler embedded in the library optimizes your code. More

pandas API 🐼

FireDucks is fully compatible with pandas API. The only difference is the import statement. No additional learning is required to start with FireDucks.Get started

AS IS Execution 🚀

You can run your pandas program directly with FireDucks. Its import-hook functionality will automatically replace the import statement for pandas with the import statement for FireDucks. More

Benchmarks 📈

FireDucks shows high performance gain while executing various queries included in the TPC-H and TPCx-BB benchmarks. More.

Use Cases

Automobile

Telecom

Finance

Cloud Hosting

E Commerce

Gaming

From Users

alt
TTDC

Toyota Technical Development Corporation has seen significant benefits within about three months of the introduction of the system.

Read
alt
NEC Solution Innovators

NEC Solution Innovators, Ltd. has used FireDucks for a data analysis task.

Read

From Developers

alt
Pitfalls of %%time

By Osamu Daido | 2024-12-26

Let's take a look at pitfalls of the %%time magic command with FireDucks.

Read
alt
How to take traces in FireDucks

By | 2024-12-20

This article introduces how to get execution traces in FireDucks.

Read
alt
Ensuring compatibility with pandas in the GPU version of FireDucks

By | 2024-12-19

In this article, we will introduce what points we have to be careful about when developing the GPU version of FireDucks to ensure compatibility with pandas.

Read
alt
Exploring performance benefits of FireDucks over cuDF

By Sourav Saha | 2024-12-18

This article demonstrates the optimization benefit of FireDucks to speedup over cuDF even on CPU.

Read