Yield Loss Reduction and Test of AI and Deep Learning Accelerators

TL;DR

This paper presents a novel methodology for reducing yield loss in AI and deep learning accelerators by leveraging their fault tolerance, enabling high fault rates with minimal accuracy impact through selective deactivation of faulty processing elements.

Contribution

It introduces an application-driven binning and testing approach that correlates circuit faults with AI workload accuracy, improving yield without sacrificing performance.

Findings

Accelerators can tolerate up to 5% fault rate with less than 1% accuracy loss.

The methodology enables effective product-binning based on fault tolerance.

A new fault isolation and test flow for PEs enhances yield management.

Abstract

With data-driven analytics becoming mainstream, the global demand for dedicated AI and Deep Learning accelerator chips is soaring. These accelerators, designed with densely packed Processing Elements (PE), are especially vulnerable to the manufacturing defects and functional faults common in the advanced semiconductor process nodes resulting in significant yield loss. In this work, we demonstrate an application-driven methodology of binning the AI accelerator chips, and yield loss reduction by correlating the circuit faults in the PEs of the accelerator with the desired accuracy of the target AI workload. We exploit the inherent fault tolerance features of trained deep learning models and a strategy of selective deactivation of faulty PEs to develop the presented yield loss reduction and test methodology. An analytical relationship is derived between fault location, fault rate, and the AI task's accuracy for deciding if the accelerator chip can pass the final yield test. A yield-loss reduction aware fault isolation, ATPG, and test flow are presented for the multiply and accumulate units of the PEs. Results obtained with widely used AI/deep learning benchmarks demonstrate that the accelerators can sustain 5% fault-rate in PE arrays while suffering from less than 1% accuracy loss, thus enabling product-binning and yield loss reduction of these chips.