# Evaluating Built-in ECC of FPGA on-chip Memories for the Mitigation of Undervolting Faults

## TL;DR

This paper investigates the use of FPGA built-in ECC to mitigate undervolting faults in on-chip memories, enabling significant power savings with minimal impact on neural network accuracy.

## Contribution

It demonstrates that FPGA built-in ECC effectively corrects most undervolting-induced faults, allowing aggressive voltage underscaling and energy savings in neural network accelerators.

## Key findings

- Over 90% of faults are correctable by built-in ECC.
- 40% power savings achieved through undervolting.
- Negligible neural network accuracy loss with ECC mitigation.

## Abstract

Voltage underscaling below the nominal level is an effective solution for improving energy efficiency in digital circuits, e.g., Field Programmable Gate Arrays (FPGAs). However, further undervolting below a safe voltage level and without accompanying frequency scaling leads to timing related faults, potentially undermining the energy savings. Through experimental voltage underscaling studies on commercial FPGAs, we observed that the rate of these faults exponentially increases for on-chip memories, or Block RAMs (BRAMs). To mitigate these faults, we evaluated the efficiency of the built-in Error-Correction Code (ECC) and observed that more than 90% of the faults are correctable and further 7% are detectable (but not correctable). This efficiency is the result of the single-bit type of these faults, which are then effectively covered by the Single-Error Correction and Double-Error Detection (SECDED) design of the built-in ECC. Finally, motivated by the above experimental observations, we evaluated an FPGA-based Neural Network (NN) accelerator under low-voltage operations, while built-in ECC is leveraged to mitigate undervolting faults and thus, prevent NN significant accuracy loss. In consequence, we achieve 40% of the BRAM power saving through undervolting below the minimum safe voltage level, with a negligible NN accuracy loss, thanks to the substantial fault coverage by the built-in ECC.