English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Conventional multiply-accumulate (MAC) operations have long dominatedcomputation time for deep neural networks (DNNs), espcially convolutionalneural networks (CNNs). Recently, product quantization (PQ) has been applied tothese workloads, replacing MACs with memory lookups to pre-computed dotproducts. To better understand the efficiency tradeoffs of product-quantizedDNNs (PQ-DNNs), we create a custom hardware accelerator to parallelize andaccelerate nearest-neighbor search and dot-product lookups. Additionally, weperform an empirical study to investigate the efficiency--accuracy tradeoffs ofdifferent PQ parameterizations and training methods. We identify PQconfigurations that improve performance-per-area for ResNet20 by up to3.1$\times$, even when compared to a highly optimized conventional DNNaccelerator, with similar improvements on two additional compact DNNs. Whencomparing to recent PQ solutions, we outperform prior work by $4\times$ interms of performance-per-area with a 0.6% accuracy degradation. Finally, wereduce the bitwidth of PQ operations to investigate the impact on both hardwareefficiency and accuracy. With only 2-6-bit precision on three compact DNNs, wewere able to maintain DNN accuracy eliminating the need for DSPs.

PQA: Exploring the Potential of Product Quantization in DNN Hardware Acceleration