I am excited to share a project I’ve been developing as part of my academic research: The Turkish Sieve Engine (TSE v1.0.0).
This engine is an implementation of the N/6 Bit Methodology, a novel approach designed to minimize memory footprint while maximizing throughput for prime number generation.
Whether you are a math enthusiast, a cryptography researcher, or a high-performance computing (HPC) geek, I’d love to get your feedback!
Hi everyone,
I've been working on optimizing prime number sieving, specifically focusing on reducing the memory footprint without sacrificing throughput. I've developed a methodology called N/6 Bit and implemented it into a tool called the Turkish Sieve Engine (TSE).
The project is now open-source, and I'm looking for feedback from the community regarding the implementation and performance.
Technical Overview:
Methodology: N/6 Bit structure (explained in the linked paper).
Implementation: Hybrid parallelism using NVIDIA CUDA and OpenMP.
Goal: Maximizing efficiency on modern hardware.
I've provided the pre-compiled binaries for those interested in benchmarking or code review.
GitHub: [https://github.com/bilgisofttr/turkishsieve]
Zenodo (Academic Paper): [https://zenodo.org/records/18038661]
I would love to hear your thoughts on the memory management or if you have results from different GPU architectures.