ALF Conference call, Sept. 19 2001 Participants ------------ Alex Zamfirescu - ASC Jake Karrafalt - ASC John Zao - ASC Jay Abraham - Silicon Metrics Joe Daniels - Accellera Greg Dufour - Mentor Graphics Srinivas Madaboosi - Fujitsu Kim Nguyen - Tera Systems Tim Baldwin - Monterey ALF IEEE status --------------- PAR number has been changed to 1603 Need to fill out form with list of workgroup participants for IEEE A.I. send email with IEEE membership # to Wolfgang - all A.I. post pertinent PAR information on web - Wolfgang ALF test suite -------------- ASC explained their activity on ALF test suite. Parser is under development. Could be made available to the industry. At this point in time, terms should be negotiated between interested parties and ASC. Contact person: Jake Karrfalt President and CEO Alternative System Concept Inc 22 Haverhill Road, P.O. Box 128, Windham, NH 03087 phone 603-437-2234 fax 603-437-2722 email jake@ascinc.com web www.ascinc.com A.I. publish development schedule - ASC A.I. set up ftp site for testcase submission, logistics to be negotiated between Accellera and ASC - Wolfgang Comments and questions from audience: Jay: interested in seeing the spec and also ALF writer Wolfgang: interested in intelligent translator between .lib and ALF. Greg: informative interest Srinivas: informative interest, need to discuss interally w Fujitsu Kim: Tera Systems is developping ALF today Tim: Support OLA for a key customer, ALF will also be customer-driven Jake got feedback from his customers, indicating demand for OLA. Srinivas wants to learn more about ALF and OLA in order to enable Fujitsu to make a decision about what is the way to go. A discussion about the relation between ALF, OLA, .lib followed. Summary by Tim Baldwin: OLA is a library API based on a compiler from IBM, distributed by Si2. The compiler uses DCL as an input language. DCL is a control-centric language, targeted for delay calculation, whereas ALF is data-centric, targeted for general purpose cell characterization. One objective of the OLA project was to enable the compiler to use ALF as primary input by extending the OLA semantics to be compatible with ALF. However, this work has not resulted in a production-worthy solution so far. Therefore OLA supporters prefered to stick with DCL, ALF supporters prefered to stick with native ALF. Today there is an almost equal split amongst ASIC vendors between strong OLA supporters and strong ALF supporters. Comments from Wolfgang: The rationale for OLA vs ALF comes down to make vs buy. If an ASIC vendor wants to use proprietary software for delay calculation in conjunction with commercial EDA tools, a standard API is needed. If an ASIC vendor wants to use native delay calculation available in EDA tools, a standard library format is more practical. An other issue is availability of features. The OLA workgroup has just started signal integrity specification development, whereas ALF-compliant commercial signal integrity tools have been developped and publicly announced earlier this year. Comments from Jay: There is a tendency in the industry towards using native delay calculators in EDA tools, especially with the improvements of commercially available signal integrity analysis. Primetime-SI which uses .lib format has a strong position. Assignment of IEEE doc chapter review ------------------------------------- The assignee is meant to take some responsibility for continuous review. One rough review cycle will be necessary by the time of the next meeting. Tim Baldwin: 9.6 - 9.9, 11.13, 11.15 (related to physical modeling) Kim Nguyen: 6 and 8 (lexical rules, generic objects) Jay Abraham: 11 except 11.13 and 11.15 (analog behavior) Jake Karrfalt: 5 and 7 (language construction principles, generic objects) Tim Ehrler (offline): 10 (digital behavior) A.I. and related doc review --------------------------- Note: Due to the fact that most attendees were new, the detailed technical review and A.I. follow-up got defered. Alex committed to complete his previous A.I.s before the next meeting. Wolfgang will follow up with Tim Ehrler and Sergei Sokolov on their A.I.s Parser issues ------------- Joe Zao has sent email to Wolfgang about ASC's parser issues. Wolfgang answered (see reflector). Currently, ASC's parser supports only predefined keywords, no custom keywords. Tim Ehrler has sent email to Wolfgang: no parser issues from Philips. A.I. send follow-up email to reflector - Tim Ehrler Next meeting ------------ October 9 at NEC, face-to-face Tim Baldwin: plan to call in - evt send expert for physical modeling Kim Nguyen: plan to attend part of meeting, need detailed schedule Srinivas: need to confer with his management Alex: plan to attend Joe: plan to attend Tim Ehrler (offline): plan to attend