Abstract:Tensor algebra plays a major role in various applications, including data analysis, machine learning, and hydrodynamics simulation. Different tensor algebra inherently varies in dimension, size, and computation, leading to different execution preference, including parallelization, data arrangement, and accumulation. Another critical aspect for tensor algebra is the involved tensors can be with varying mixes of dense and sparse representation. Such diversified applications are notoriously difficult to accelerate. Prior ASIC architectures do not meet the needs due to fixed dataflow and prior fine-grained fabrics (e.g., FPGAs) solutions offer limited performance and power improvement due to bit-level reconfigurable structure. In this article, we propose Morphling, a reconfigurable architecture that can flexibly handle both dense and sparse tensor computation. We first generalize a flexible execution model that decomposes tensor operations into three steps, including tensor vectorization, vector computation, and output reduction. The dense and sparse tensor computation share the same execution model, but differ in the vector computation step where the multiplications are conducted. Depending on the number of inputs and outputs that are linked together in the computation step, we define three parallel patterns, including many-to-one, one-to-many, and one-to-one, which correspond to different implementations for dense and sparse computation. Furthermore, to efficiently support sparse tensor, we design a tiled-BCSR format that enables high parallelism and balanced workload. At the architecture level, we propose a reconfigurable design to support the execution model. The hardware units can be reconfigured to support different datapath and enable different types of data reuse. We evaluate Morphling using various tensor operations and compare it with CPU, GPU, FPGA, and state-of-the-art ASIC designs. Overall, Morphling achieves 13.4X, 677.7X, 44.7X energy efficiency over Xilinx ZC706 FPGA, Intel i7-9700K CPU, and NVIDIA TitanX GPU.

MorphoSys: an Integrated Re-configurable Architecture

Modeling, Implementation and Scalability of the Morphosys Dynamically Reconfigurable Computing Architecture

PRAM: A Novel Model For 2-Dimensional Reconfigurable Arrays

Model of Reconfigurable Computing System

Design of a Media-Enhanced and Reconfigurable RISC ,Microprocessor

Design Of A Simulator For Mesh-Based Reconfigurable Architectures

A Reconfigurable Architecture Specific for the Butterfly Computing

Asynchronous reconfigurable computing array design

Morphling: A Reconfigurable Architecture for Tensor Computation

Multi-core System with Reconfigurable Datapath

A Hierarchical Local-Interconnection Structure for Reconfigurable Processing Unit

A Reconfigurable Processor Architecture Combining Multi-Core and Reconfigurable Processing Units

A Mixed-Signal Data-Centric Reconfigurable Architecture Enabled by RRAM Technology (abstract Only).

Novel many-core architecture design for real-time image processing

Reconfigurable Computing: a Promising Microchip Architecture for Artificial Intelligence

Reconfigurable Computing: Toward Software Defined Chips

Examples of Reconfigurable Cryptographic Processor Design

Accelerating Edge AI with Morpher: An Integrated Design, Compilation and Simulation Framework for CGRAs

Coarse-grained reconfigurable multimedia processor

A Reconfigurable Multi-Processor Soc For Media Applications

Reconfigurable Neuromorphic Computing: Materials, Devices and Integration