Minutes, IBIS Quality Committee 11 October 2011 11:00-12:00 EST (08:00-09:00 PST) ROLL CALL Cisco Systems: * Tony Penaloza Ericsson: Anders Ekholm Green Streak Programs: Lynne Green Huawei Technologies: Guan Tao IBM: Bruce Archambeault Greg Edlund IOMethodology: * Lance Wang Mentor Graphics: John Angulo Micron Technology: Moshiul Haque, Randy Wolff Nokia Siemens Networks: * Eckhard Lenski QLogic Corp.: James Zhou Signal Consulting Group: Tim Coyle Signal Integrity Software * Mike LaBonte Teraspeed Consulting Group: * Bob Ross Texas Instruments: Pavani Jella Everyone in attendance marked by * NOTE: "AR" = Action Required. -----------------------MINUTES --------------------------- Mike LaBonte conducted the meeting. Call for opens and IBIS related patent disclosures: - None AR Review: - Mike post AMI checklist to archive - Just done - Greg update AMI checklist items 2.1 through 2.3 - No report - Mike add descriptions to AMI checklist - No changes made - Mike add QUAILs to Mantis - Done - Bob: I couldn't get in to Mantis - Mike: The first page requires a login - I may need help reconfiguring New items: Mike showed the AMI checklist: - Mike: BUG 104 was marked "not a bug" in the last Open Forum - Bob discussed bug processing issues - BUG122 was difficult because it applied to all many settings - This was about errors interpreting Vmeas and Rmeas - It can be difficult to get to the root problem - Michael Mirmak filed a bug on the Touchstone parser Mike showed the [Model Spec] section of IBIS 5.0 - Bob: The issue was making it so the overrides were correct - It should be obvious but is not clear in the spec - Mike highlighted a sentence on page 42 "If one rising or falling subparameter is used, ..." - This makes no sense. - Eckhard: If you use Cref rising/falling then you also should use Rref rising/falling and Vref rising/falling. - The working is strange. - Bob: This is a tack-on to a tack-on to a tack-on to the spec - Mike: Is this a quality issue? - Bob: I don't think anyone uses it Mike returned to the AMI checklist: - Mike: 2.4 through 2.6 are related - They are about multiple instances stepping on files - Bob: It's confusing unless you know how the AMI models work - Mike: It could be reduced to one check - Multiple AMI instances should not conflict - These checks could be sub-bullets - Bob: Check 2.3 should be a business decision - We should not be endorsing operating systems - Mike: Solaris seems to be falling off - Eckhard: Solaris is a form of Unix - Bob: The check should be that supported OSes are listed - Eckhard: I need to know which OSes a model runs on - We find out too late when it doesn't work - Mike: Need to start changing the document instead of adding comments. - 2.4 through 2.6 are now 2.4 - Bob: We can suggest that Windows and Linux are widely used - Mike: 2.5 (formerly 2.7) "Unrecognized parameters" was already discussed - This is a model issue - It should be very easy to test, just create a parameter called "junk" - Bob: It could be an unrecognized Out parameter - That would be a tool issue - Mike: Back to 2.4, most of the work to prevent problems could be done by the EDA tool - We can address what the model should and should not do - In a crosstalk simulations there may be 5 of the same TX, 5 of the same RX - Models need to generate unique file names as needed - Bob: Often different vendors are used for TX and RX - You don't have control over the names - Mike: There is a question of how much the EDA tool does vs. the model - Mike: Separate simulations are often run by a farm in separate directories - There could be conflicts with other shared resources such as memory and sockets Mike showed BIRD 121.1 IBIS-AMI New Reserved Parameters for Data Management: - Mike: This proposes DLLid to help models remain separated - Bob: That may help in the future but there are solutions now - It may be that models bypass the problem today - Mike: We could say a quality model would use the DLLid parameter - Bob: We can't have that in the spec - Tools can handle this problem now - Mike: Tools could copy entire directories to keep instances separate - Bob: Do we have a problem with this now? - Mike: Greg may have some anecdotes - When you have this kind of problem it can be very hard to debug - Bob: We can't rewrite the spec - Mike: True, we will not have checks requiring BIRDs - We don't have to require DLLid, but just that there is no conflict - There might be tool specific troubles though - Someone needs to work it out so it always works for users - Bob: The audience for quality checks is model makers, but users can do it too AR: Mike post latest AMI checklist version to work archive Next meetings: - Next meetings Oct 18 and Oct 25 Oct 18 agenda: - Continued IBIS-AMI checklist review Meeting ended at 12:11 Eastern Time.