SigTekX

>_ GPU-Accelerated Signal Processing Engine v0.9.5

A high-performance C++/CUDA signal processing engine architected for real-time ionospheric scintillation analysis. SigTekX bridges low-level GPU optimization with high-level Python research workflows, delivering sub-millisecond processing latency for ULF/VLF radio signal monitoring across batch and streaming execution modes.

System Overview

FIG 01: High-level architecture — Data Plane (C++/CUDA) and Control Plane (Python API)

Core Stack

C++17 CUDA / cuFFT Python AWS Streamlit

Fast Facts

171.1 μs Mean Latency

237.6 MSPS Peak Throughput

138.4 dB SNR Accuracy

0.003 RTF Real-Time Factor

Deep Dives

Module 01

Core GPU Engine

The C++17/CUDA backend — batch and streaming executors, zero-copy memory architecture, lock-free SPSC ring buffers, and cuFFT spectral analysis pipelines.

C++17 CUDA cuFFT Zero-Copy

Module 02

Cloud Infrastructure

AWS EC2 GPU benchmarking, ECR container registry, automated deployment pipelines, and CloudWatch observability for production workloads.

AWS EC2 ECR CloudWatch Docker

Module 03

Researcher API

Python bindings exposing the C++ engine, CLI tools for dataset management, and the dual-plane architecture separating control from data flow.

Python pybind11 CLI Dual-Plane

Module 04

Performance Analytics

Streamlit dashboards for latency and throughput benchmarking, Nsight profiling results, and real-time factor validation across execution modes.

Streamlit Nsight Benchmarks RTF

Access_Source_Repository Return_to_Projects