cufft

CUDA Library Samples

Python interface to GPU-powered libraries

Hooked CUDA-related dynamic libraries by using automated code generation tools.

fat_llama is a Python package for upscaling audio files to FLAC or WAV formats using advanced audio processing techniques. It utilizes CUDA-accelerated calculations to enhance audio quality by upsampling and adding missing frequencies through FFT, resulting in richer and more detailed audio.

A parallel implementation of image denoising using Cuda and cuFFT.

Wrapper for the CUDA FFT library

MoE-routed GPU signal processing framework — batch cuFFT, kernel dispatch, zero-copy pipelines

This is a benchmarking test for convolution reverb with single core/sequential code and a parallelized implementation using CUDA and cuFFT. This is in fulfillment of my Music Technology Undergraduate Capstone Project.

Six implementations of the same FFT — hand-written CUDA, OpenACC and cuFFT — benchmarked on P100 and V100 to measure what a directive-based GPU port actually costs

GPU-accelerated electrophysiology (EEG, iEEG, LFP) transforms for large batch jobs.

A safe Rust wrapper around a subset of cuFFT.

CUDA sparse binary 2-D FFT with compact CSC input, Bluestein transforms, and cuFFT/SpFFT baselines.

FFT and complex-valued tensor operations for AWS Trainium via NKI (cuFFT-equivalent) — Cooley-Tukey, Bluestein, STFT, ComplexTensor and complex NN layers.

cuFFT helpers for pycu_interface.

GPU-powered batch FFT analysis of instrument audio samples using CUDA and cuFFT. Visualizes both time and frequency domains with Python—fast, scalable, and beautifully plotted.

Radix-2 Cooley-Tukey FFT kernel in CUDA/C with shared-memory stage fusion and a coalesced bit-reversal transpose. Reaches 25% of cuFFT throughput and 70x over a single-threaded CPU on 1M-point transforms (NVIDIA T4). Every figure reproducible from the included Colab notebook.

Export cuFFT through gRPC

Hand-written batched 1D FFT (Cooley-Tukey radix-2) in C++/OpenMP and CUDA, benchmarked against Intel MKL and cuFFT on NVIDIA P100/V100 GPUs

Sparse binary 2D FFT on CUDA/cuFFT with memory-footprint optimization, streaming tiles, Hermitian symmetry, and Nsight analysis.