Design Con 2015
Welcome Guest Log In | Register

Advanced Yield Analysis and Optimization with 3 to 6 Sigma Statistical Simulations for Memory, Logic, Digital and Analog Designs

Original Air Date: Oct 4, 2012 | Duration: 60 minutes Webinar
0 2
0 1 View


    Please disable any pop-up blockers for proper viewing of this webinar.

More Info Less Info

Overview:
Handling process variations in process development, SPICE modeling and design optimization has become a big challenge, yet a must for advanced process nodes. This Webinar will demonstrate how to make statistical yield analysis (3-6 sigma) more practical, more reliable and faster, by fully utilizing the foundry SPICE models, integrated statistical SPICE engine and hardware-validated statistical sampling technologies. Attendees are invited to discuss with ProPlus and IBM experts, and learn about the technologies, solutions and application experience sharing in these related areas.

Who should attend:
Engineers and management who may:

  • Have problems and concerns in foundry SPICE models, especially the statistical variation models
  • Be concerned about process variations in advanced process nodes and want to make better use of foundry variation models
  • Be looking for better solutions on statistical yield analysis with regular Monte Carlo simulations for analog designs
  • Need an integrated solution for High Sigma Monte Carlo analysis for your memory or digital designs
  • Be interested in learning about High Sigma design and application experiences shared by an IBM expert

What attendees will learn:

  • How process variations are handled by advanced SPICE models and understand how to better use those models and make them application specific.
  • The keys to a practical yield analysis solution and how to make it more reliable and faster.
  • The technologies and the seamlessly integrated DFY solution from ProPlus, covering the advanced modeling solutions, high performance high accuracy statistical SPICE simulation engine and hardware-validated statistical sampling technologies.
  • The experience sharing on yield analysis and design optimizations from IBM experts, especially High Sigma analysis for memory designs.
Presenters:
Dr. Bruce McGaughy, Chief Technology Officer and Senior Vice President of Engineering, ProPlus Design Solutions, Inc.
Dr. Bruce McGaughy currently serves as the Chief Technology Officer and Senior Vice President of Engineering of ProPlus Design Solutions, Inc. He was most recently the Chief Architect of Simulation Division and Distinguished Engineer at Cadence Design Systems Inc., responsible for all Cadence simulation products and technologies, including Spectre/SpectreRF, APS, UltraSim, etc.

Dr. Rajiv V. Joshi, Research Staff Member, T. J. Watson Research Center, IBM.
Dr. Rajiv V. Joshi is a research staff member at T. J. Watson research center, IBM. He received his B.Tech degree from Indian Institute of Technology (Bombay, India), M.S. degree from Massachusetts Institute of Technology and Doctorate in Eng. Science from Columbia University, USA. He developed novel interconnect processes and structures for Aluminum, Tungsten and Copper technologies, which are widely used at IBM for various sub-0.5um memory and logic technologies as well as across the globe. His circuit related work includes design of register files, registers, latches, L1, L2 Caches, development of physical design tools, and CAD-based library generation and circuit designs in SOI technology. He received an outstanding technical achievement award for his contributions to IBM microprocessor designs. His recent work related to 8T stable 6 GHz SRAM cell was covered by EE times.

2 comments
write a comment

No Avatar

Larkin X Posted Oct 4, 2012

I'm quite interested in this area but found the presentation too 'thin' for me to begin to conclude this tool might be as robust as I need. There is a strong burden of proof when one claims 6-sigma in simulation. This presentation fell far short of convincing me that this tool works as promised. The speedups claimed for 6-sigma seem to make the dangerous assumption that the results are Gaussian. Real life doesn't fit any distribution out to 6-sigma. The second part of the talk seemed to talk about circuit sensitivities in the most general intuitive terms rather than demonstrate cases where non-intuitive process variation occcured in the tails but was efficiently handled by the tool.

reply

No Avatar

ProPlus Posted Oct 12, 2012

Thank you for your interest and comment. Here's our response: NanoYield does not assume that the results are Gaussian. In fact, it does not even assume the model parameter distributions are Gaussian. Furthermore, the tool can handle non-intuitive situations. This is illustrated by the fact that the technology has been proven and verified against silicon at IBM for multiple generations of technology and on a wide range of applications.

reply

Please Login

You will be redirected to the login page

×

Please Login

You will be redirected to the login page

×

Please Login

You will be redirected to the login page