High-performance exact vector search on NVIDIA & AMD GPUs. 9354 QPS (RTX 5090D), 6275 QPS (RX 7900 XTX). Open-source, deterministic, and recall=1.0 — enabling Trustworthy AI at the edge.
cuda brute-force gpu-acceleration cuda-kernels hip amdgpu explainable-ai vector-search edge-ai vector-search-engine trustworthy-ai exact-searching rdna3 secure-ai-practices deterministic-ai rocml ann-benchmarks
-
Updated
Jan 22, 2026 - C++