Silhouette: Toward Performance-Conscious and Transferable CPU Embeddings

NeurIPS 2023 Workshop MLSys Submission25 Authors

Keywords: Hardware embedding, CPU embedding, Transfer Learning

TL;DR: We present our approach Silhouette, that leverages publicly-available CPU performance data sets to learn CPU performance embeddings

Abstract: Learned embeddings are widely used to obtain concise data representation and enable transfer learning between different data sets and tasks. In this paper, we present our approach Silhouette, that leverages publicly-available CPU performance data sets to learn CPU performance embeddings. We show how Silhouette enables transfer learning across different types of CPU and leads to a significant improvement in performance prediction accuracy for the target CPUs.

Submission Number: 25