mebrown47

mebrown47 / CUDA_spectrum

Public

A simple CUDA FFT spectrum tool

cuda dsp fft sdr spectrum-analyzer
76
9
100% credibility
Found May 07, 2026 at 76 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A GPU-accelerated tool that processes raw radio IQ samples into power spectra, with Python scripts for plotting waterfalls and analyzing peaks.

How It Works

1
📡 Discover the Signal Analyzer

You hear about a speedy tool that turns radio wave captures into clear pictures of hidden signals using your computer's graphics power.

2
💻 Get It Ready

You download the files and prepare them on your computer that has a strong graphics card.

3
📻 Capture Radio Data

You record radio signals from your receiver device and prepare the raw data stream.

4
🚀 Launch the Analyzer

You pipe your radio data into the tool, pick settings like zoom level for frequencies, and it quickly processes huge amounts to reveal the spectrum.

5
Choose Your View
📈
Make Pretty Plots

Use the plotting helper to create power graphs and waterfall images showing signals over time.

📊
Get Stats Report

Run the analysis script for peak signal lists, averages, and top frequencies without pictures.

6
🌈 Explore the Results

You see frequency charts with peaks highlighted, noise floors, and signal strengths making weak transmissions pop.

🎉 Spot Hidden Signals

You successfully detect and analyze radio signals with fine detail, like a professional radio explorer.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 76 to 76 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is CUDA_spectrum?

CUDA_spectrum is a GPU-accelerated spectrum analyzer that takes interleaved float32 IQ samples from stdin—like from RTL-SDR or HackRF—and cranks out power spectra via cuFFT on CUDA hardware. It handles massive FFT sizes up to 64M points for fine resolution down to 0.15 Hz bins at 10 Msps, with windowing, dBFS output, and streaming over ZMQ or to files in CSV, binary, or JSON. Python helpers generate waterfalls, PSD plots, and peak stats, making it a simple CUDA program for quick RF analysis pipelines.

Why is it gaining traction?

It blasts through large FFTs way faster than CPU tools, perfect for long integrations on weak signals, with dead-simple nvcc build and pipe-friendly CLI options like -f for FFT size or --zmq tcp://*:5555. ZMQ support lets it feed live data to GNU Radio or plotters, while binary output avoids CSV slowdowns on big frames—real user wins over bloated SDR suites. As a simple CUDA code example, it hooks devs wanting GPU FFT without cuFFT boilerplate.

Who should use this?

SDR tinkerers processing live captures from rtl_sdr or hackrf_transfer, RF surveyors chasing faint sats with 16M+ FFTs, or GPS L1 hunters needing peak freqs and doppler offsets. Suits students on simple CUDA projects or engineers prototyping spectrum tools before scaling.

Verdict

Grab it for a simple CUDA kernel example if you need fast, large-FFT spectra from IQ streams—docs and examples are sharp, build is one nvcc line. Low 1.0% credibility from 76 stars signals early maturity, so test on your RTX first; solid for niches but watch for edge cases.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.