IEEE 754R notes from January 20, 2005

Notes taken by Dan Zuras.

January 20, 2005

Notes for meeting at Sun in Building 14 in Santa Clara at 1:00 on Thursday 1/20/05. Joe Darcy hosted us in the Space Mountain Room. Mike Cowlishaw, Dick Delp, David Hough, Prof Kahan, Alex Liu, Jim Thomas, Peter Markstein, Jon Okada, & Dan Zuras attended. Jeff Kidder & Leonard Tsai were on the phone.

Prof Kahan asked for people who are attending the ARITH conference who are also willing to be on a panel discussing the "non-progress" of the 754 work. Prof Kahan, David Hough, Mike Cowlishaw, Jeff Kidder, Peter Markstein are going. Getting a letter from Isreal Koren inviting Jeff & Peter would help.

We began with the proposals for the new definition of underflow & the new notion of abrupt underflow. It took a long time to present given all the questions & discussions. It was as much a problem of teaching as of convincing. In the end, I don't think either teaching or convincing went well.

Prof Kahan proposed that underflow be defined by tinyness after rounding. It will be signalled if the result is inexact. It does not signal in the case of gradual underflow where the result is exact. It is signalled in the case of a nonzero abrupt underflow. I so proposed & Kahan seconded. It passed with no objections.

Its 3:00. Time to pee.

We went on to contractions after the break & Prof Kahan began by explaining his sic() notation. Oddly enough, even this generated a great debate. As well as an interesting discussion of the difference between finite & non-infinite. Much was discussed. Little was decided.

David pointed out that this entire discussion might have gone smoother if it was confined to the high level concepts rather than getting bogged down in the text itself. While I agree, I'm not sure how to do it.

I asked everyone how we make better progress in the near future & got no new ideas beyond that which we are doing in the current subcommittee.

Invite Ivan to next Thursday's subcommittee meeting.

754 | revision | FAQ | references