Flash-SD-KDE: Accelerating SD-KDE with Tensor Cores
This work addresses the computational bottleneck for researchers and practitioners using SD-KDE, enabling faster density estimation at previously infeasible scales, though it is incremental as it optimizes an existing method.
The paper tackled the slow performance of score-debiased kernel density estimation (SD-KDE) by re-ordering its computation to leverage Tensor Cores on GPUs, achieving up to 47x speedup over a strong baseline and making it practical for large-scale tasks like 1M samples in 2.3 seconds.
Score-debiased kernel density estimation (SD-KDE) achieves improved asymptotic convergence rates over classical KDE, but its use of an empirical score has made it significantly slower in practice. We show that by re-ordering the SD-KDE computation to expose matrix-multiplication structure, Tensor Cores can be used to accelerate the GPU implementation. On a 32k-sample 16-dimensional problem, our approach runs up to $47\times$ faster than a strong SD-KDE GPU baseline and $3{,}300\times$ faster than scikit-learn's KDE. On a larger 1M-sample 16-dimensional task evaluated on 131k queries, Flash-SD-KDE completes in $2.3$ s on a single GPU, making score-debiased density estimation practical at previously infeasible scales.