Abstract:Overlay is an effective approach for creating FPGA-based AI accelerators, enabling software-programmable specialized hardware datapaths to flexibly support various DNN operations. Traditional DNN overlays typically base their instruction set design on the von Neumann model but adapt them to be more coarse-grained. These instruction sets control execution at the layer granularity and impose restricted patterns for mapping computation and bandwidth resources. Such constraints cause inefficiencies from the imperfect match between supported execution patterns and diverse DNN layer shapes and types. This work proposes a Reconfigurable Stream Network architecture, a unique ISA abstraction tailored for flexible FPGA overlay execution at low cost, marking it as the first known FPGA design to support dynamic sequential linear layer pipelining. This novel architecture presents a datapath abstraction modeled after a specialized circuit-switched network with stateful functional units (FUs) as nodes and data streaming on edges. Programming a computation corresponds to triggering a network path in this stream-connected datapath. The program can individually control FUs to form paths that exploit both spatial and pipeline parallelism between independent and dependent concurrent computations. We present a proof-of-concept design RSN-XNN on the Versal VCK190. Evaluations show a 22x latency reduction for BERT compared to the state of the art, along with throughput improvements of 3.2x, 2.4x, 2.5x, and 2.8x for BERT, VIT, NCF, and MLP, respectively. RSN-XNN matches the latency of the T4 GPU with the same FP32 performance but only 18% of the memory bandwidth. Compared to the A100 GPU under the same 7nm process node, it achieves 2.1x/4.5x better operating/dynamic energy efficiency in FP32.

Reliability-Aware Overlay Architectures for FPGAs: Features and Design Challenges

The Research of FPGA Reliability Based on Redundancy Methods

A Cross-Layer Framework for Early-Stage Full Chip Oxide Breakdown Reliability Analysis

Reliability Modeling of Fault-Tolerant FPGA-Based Architectures in Space Applications for Soft and Hard Error Recovery

An Adaptive Cross-Layer Fault Recovery Solution for Reconfigurable SoCs.

Design of a Reconfigurable Cryptographic Engine

Enabling Effective FPGA Debug using Overlays: Opportunities and Challenges

A survey on Dependable Digital Systems using FPGAs: Current Methods and Challenges

Towards Cross-layer Reliability Analysis of Transient and Permanent Faults

Fault-tolerant Technology of SRAM-based FPGA

Ada-FA: A Comprehensive Framework for Adaptive Fault Tolerance and Ageing Mitigation in FPGAs

Improving FPGA resilience through Partial Dynamic Reconfiguration

Fault modeling and characteristics of SRAM-based FPGAs (abstract only).

Cross-Layer Design for Reliability in Advanced Technology Nodes: an EDA Perspective

A new architecture for online error detection and isolation in network on chip

A Cycle-Accurate Soft Error Vulnerability Analysis Framework for FPGA-based Designs

Systematic Reliability Evaluation of FPGA Implemented CNN Accelerators

Improving Reliability, Security, and Efficiency of Reconfigurable Hardware Systems

A Flexible Soft Error Mitigation Framework Leveraging Dynamic Partial Reconfiguration Technology

Izip: In-Place Zero Overhead Interconnect Protection Via Pip Redundancy

Addressing Architectural Obstacles for Overlay with Stream Network Abstraction