Blog
Methodology, implementation write-ups and challenge post-mortems.
Privacy-preserving systems depend on a large and fragmented landscape of computational building blocks, with different implementations winning across parameters, workloads and hardware. As AI makes implementations cheaper to create and specialize, the bottleneck shifts from building them to continuously discovering, evaluating and integrating the best available options. FHERMA is the infrastructure layer for that new computational landscape.
Fair Math and NVIDIA are working together to address the performance bottlenecks of Fully Homomorphic Encryption by developing high-performance, GPU-accelerated building blocks through open challenges on the FHERMA platform, leveraging the state-of-the-art NVIDIA cuPQC SDK.
One email a week, when something is actually measured.
New articles, closed challenges and specification changes. No announcements.