Have you ever wondered about how the DAC technical program comes into existence? This month Peggy Aycinena gives us an in-depth look at how the papers are reviewed and selected. The process may surprise you.

DAC is a community event that exists to support the chip design ecosystem - from research to silicon - and, as with most community efforts, is succussful only through the efforts of many volunteers. This month's feature article highlights two of those volunteers and the near Herculean task of organizing the DAC technical program.

Enjoy!

Let us know what you think: news@dac.com.

View the 42nd DAC Proceedings

View the Exhibitors

A Labor of Love

by Peggy Aycinena

Grant Martin has a day job. So does Sachin Sapatnekar. By day, Grant is Chief Scientist at Tensilica, Inc. in Santa Clara; Sachin is a professor at the University of Minnesota in Minneapolis, where he holds the Robert and Marjorie Henle Chair in ECE.

Both of these fellows have plenty to keep themselves occupied during the day, so why take on this whole other job; one that comes with no pay, yet is guaranteed to consume all of their waking hours other than those spent at the office?

For both Sachin and Grant, the answer is the same: "For the love of the technology."

Grant Martin and Sachin Sapatnekar are Co-chairs of the Technical Program Committee for the 43rd Design Automation Conference, DAC 2006. It's a two-year assignment; this is Grant's second year and Sachin's first.

I had a chance to speak to with Sachin and Grant recently. Although they live in different states and time zones, and hail from different countries - Grant from Canada and Sachin from India - they think as one when describing their efforts to facilitate the massive task of procuring, sorting, and vetting the technical content for DAC.

They have lots of help, of course - the DAC staff, the 83 people who have volunteered their time to sit on the 17 Technical Program Sub-committees this year, plus the 800 additional volunteers who serve on an as-needed basis as reviewers as well.

The DAC staff handles the initial receipt of the papers and the task of obscuring the names of the authors, if the authors have not already attended to that themselves. It's important that the paper submission review process be as impartial as is humanly possible.

Grant and Sachin then double check for personal references within the text that might reveal identities, and then sort and review the allocations of papers to one of the 17 Technical Program Sub-committee Chairs.

Authors can choose a subject category when they submit their papers, but the Co-chairs frequently reassign submissions to a different category - approximately 100 papers this year were reassigned - to ensure that the subject area of the paper is well matched to the expertise of the reviewers.

The Sub-committee Chairs, in turn, assign the various papers to their sub-committee members or to additional on-call reviewers if necessary - and then the real fun begins.

By the way, between the December 19th deadline for DAC 2006 paper submissions, and the January 1st deadline for delivery of the papers to the Sub-committee Chairs, Grant and Sachin processed 865 papers. That's 432.5 papers per Co-chair - over the course of 11 days!

For Sachin and, Grant, this is indeed a labor of love. You couldn't pay them enough to get them to do this if they weren't full committed to the task.

Of course, were you to ask any of the other folks involved why it is they are donating their time and energy to the paper selection process, and certainly they would profess to the same sense of commitment. Why do they do it? "For the love of the technology."

The sub-committees are organized along subject lines; and populated with people highly knowledgeable in the particular technology niche assigned to that committee.

Sachin said, "The people who are on these sub-committees are very capable of reading the submissions and passing judgement on the technical merits and currency of each paper they review."

Straightforward enough, but it's hard work nonetheless. Do the math.

Most sub-committees have 5 or 6 members. Each sub-committee is assigned 40, 50 or sometimes 60 papers to review. (Two of the sub-committees this year actually received 100 papers to review!) The sub-committees range in size from 3 to 9 members. Each reviewer spends somewhere between 2 and 6 hours reading a paper. With at least 3 reviews to a paper, each sub-committee member must read at least 25 to 35 papers to be carrying his or her share of the review workload, because each paper must be read by at least 3 different reviewers.

Let's presume an average of 3 hours per paper times 25-to-35 papers, and we're looking at 75 to 100 hours at a minimum to do the reviews, per sub-committee member.

However, there's more.

After each reviewer has come to a conclusion about a paper, they will then want to compare their conclusions about the acceptability of the submission with the others individuals on the sub-committee who were also assigned to read that particular paper. More open-ended hours of work are involved in this process. Often additional reviews are sought from the legions of external revieweróthis year additional 2500 reviews were solicited from that group.

Sachin explained: "People on the sub-committees need to trust each other and their judgment. Each member is sorting and ranking the papers they have received to review, and then there is follow-on discussion to clarify the judgments of each reviewer."

Grant added: "The reviewers put the papers they see into three categories: definitely yes, definitely no, and then everything in between. It's this middle category that takes up most of their time. The people who serve on the committees have great respect for each other. There can be honest disagreement over the technical merits of a paper, but in general the people on these committees love what they do and enjoy interfacing with other technologists."

At this point, the initial selection portion of the process may be finalized, but the meter is still running. Now the overall sub-committee will recommend one of their papers for a DAC Best Paper Award - a prestigious designation for any researcher in the field. Sometimes a sub-committee will want to recommend two papers.

Those recommendations are built on yet more hours of consultation, committee-wide discussions, and consensus building to settle on the paper, or papers, that should receive the nomination.

Have you done the math?

Between the committee reviewers, additional external reviewers who may have been pulled into the process, the Sub-committee Chairs, and the overall Technical Program Co-Chairs - we're talking here about upwards of 15,000 man-hours needed to fully review and attend to the important work of selecting the technical content for DAC. Happily, none of the people involved are afraid of hard work!

Of the 865 papers submitted this year, 150 to 160 will be accepted - a 20% acceptance rate. Grant and Sachin would like to see more papers presented, and they're working with the DAC committee to try to stretch the meeting space available in San Francisco. It's going to be tough to do that this yearóhowever as the popularity of DAC as a launching place for new ideas and technologies continues to grow, more room will definitely be needed at future Design Automation Conferences.

Grant said, "We're not trying to hit a certain percentage acceptance rate each year. People will come to DAC anyway - whether the acceptance rate is higher or lower - but DAC has a role to fulfill as a leader in the technical area of CAD tools and design. To fulfill that role, the space for technical papers at the conference must expand going forward."

What about the oft-heard complaint that bias and/or the politics may play a part in the technical paper selection process.

Sachin answered easily: "In general, politics have not played a role in the technical paper selection. The discussions are about the technology alone - they're not about controversy. Reasonable people can differ on whether a paper is good or not, but these arguments have a technical basis."

Grant added, "It's true that the larger companies often have more technologists on their staff who tend to submit more papers than smaller companies. In addition, some companies are very keen on DAC - they submit more papers than other companies of equal size. Not surprisingly, a particular company might end up with more acceptances than others, but there are no quotas whatsoever."

If there are no company quotas, are there quotas with respect to papers are from industry versus those from academia?

Sachin said no: "We don't really categorize the papers between those submitted from industry versus those submitted from academia. The truth is that it's just not that clear. So many papers have joint authors across industry and academia, that sort of tracking would be irrelevant."

What if someone has a paper turned down - can they ask for an explanation or even appeal the decision?

Grant said: "After a paper is turned down, there really isn't too much of an appeals process. There is, however, an attempt to clarify to the authors of the paper why it was not accepted. This process doesn't lead to a reversal, but it does help the authors to better understand and hopefully improve on their efforts next time around!"

And he's not kidding. Sachin and Grant have poured so much of their energy and creativity into this process - their enthusiasm is contagious.

If I were a researcher, I would welcome the opportunity to write up my research and present it to these fellows, and the ranks of the sub-committees they lead, for a review and some feedback.

Better yet, if I were a researcher, I think I would also want to spend some off-hours volunteer as a reviewer in order to work closely with Grant and Sachin, and the whole team involved in assembling the technical program at DAC.

Who wouldn't want to work closely with people who find their highest calling is a labor of love?

A Note from the DAC Technical Program Committee to authors:

"This year, DAC is using a clarification process in its paper review cycle. Reviewers will have an opportunity to raise a 'Question for Clarification' about your paper. These questions will be made available to you for response during the period of March 8 through March 13, 2006. Not all papers will inspire a Question for Clarification from reviewers. If you are contacted regarding a review of your paper, you may enter a clarifying response to these questions. The Program Committee will consider your response when reaching a final judgment on your paper. Proposing a response will not guarantee acceptance of your paper, as DAC reviews many more submissions than can be accepted."

And a gentle reminder from the DAC Technical Program Committee to their reviewers:

"Your reviews will be available via this website beginning January 18, 2006. Reviews are due March 3, 2006."

Copyright © 43rd Design Automation Conference
5405 Spine Rd. Suite 102 - Boulder, CO 80301 USA
You are receiving this email because you have attended the Design Automation Conference or a related ACM or IEEE sponsored event. To unsubscribe, click here or send an email to: unsubscribe-43240@elabs3.com and you will be removed from future correspondence. Thank you.

Highlighted Video Sessions:

SESSION 46 | Special Session: DFM and Variability: Theory and Practice Papers 46.1-46.6

These videos present multiple facets of the design for manufacturability paradigm. The papers focus on theory related to circuit and CAD solutions to deal with variability, and updated assessments of the impact of BEOL variations; and practical issues of economic justification of DFM, as well as the interaction between foundry and designer.