Variations, margins, and statistics

  • Authors:
  • Patrick McGuinness

  • Affiliations:
  • Freescale Semiconductor, Austin, TX, USA

  • Venue:
  • Proceedings of the 2008 international symposium on Physical design
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Design margining is used to account for design uncertainties in the measurement of performance, and thereby ensures that actual manufactured parts will operate in within predicted bounds. As process and environmental variations become increasingly severe and complex in nanometer process technology, design margining overheads have increased correspondingly. This paper describes the types of process and environmental variations, their impact on performance, and the traditional design margining process used to account for these uncertainties. We consider statistical timing (SSTA) in the context of its ability to reduce timing margins through more accurate modeling of variations, and quantify potential benefits of SSTA for setup and hold time margin reduction. Combining SSTA with complementary techniques for systematic variation-aware and voltage-variation-aware timing provides meaningful design margin reduction. We introduce the concept of activity based operating condition as a supporting construct for variation-aware STA flows