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/

Global: Zendy Tools annual

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

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The wide adoption and significant computing resource of attention-basedtransformers, e.g., Vision Transformers and large language models (LLM), havedriven the demand for efficient hardware accelerators. There is a growinginterest in exploring photonics as an alternative technology to digitalelectronics due to its high energy efficiency and ultra-fast processing speed.Photonic accelerators have shown promising results for CNNs, which mainly relyon weight-static linear operations. However, they encounter issues whenefficiently supporting Transformer architectures, questioning the applicabilityof photonics to advanced ML tasks. The primary hurdle lies in theirinefficiency in handling unique workloads in Transformers, i.e., dynamic andfull-range tensor multiplication. In this work, we proposeLightening-Transformer, the first light-empowered, high-performance, andenergy-efficient photonic Transformer accelerator. To overcome prior designs'fundamental limitations, we introduce a novel dynamically-operated photonictensor core, DPTC, a crossbar array of interference-based optical vectordot-product engines supporting highly parallel, dynamic, and full-range matrixmultiplication. Furthermore, we design a dedicated accelerator that integratesour novel photonic computing cores with photonic interconnects for inter-coredata broadcast, fully unleashing the power of optics. Comprehensive evaluationsshow that ours achieves >2.6x energy and >12x latency reductions compared toprior photonic accelerators and delivers the lowest energy cost and 2 to 3orders of magnitude lower energy-delay product compared to electronicTransformer accelerators, all while maintaining digital-comparable accuracy.Our work highlights the immense potential of photonics for advanced MLworkloads, such as Transformer-backboned LLM. Our work is available athttps://github.com/zhuhanqing/Lightening-Transformer.

Lightening-Transformer: A Dynamically-operated Optically-interconnected Photonic Transformer Accelerator