James, Comments below. Walter From: James Zhou [mailto:james.zhou@xxxxxxxxxx] Sent: Tuesday, September 04, 2012 2:15 PM To: wkatz@xxxxxxxxxx; IBIS-ATM Subject: RE: [ibis-macro] Package modeling comparison Hi Walter, Thanks for providing the example files. Here is a list of questions I have on the example files (there are many of them which may take some time to resolve). package.emd (1) It refers to both pin and pad in "interconnect" section, however only pins are defined in "Pin" section. Should there also be a section to define "pads"? a. <WMK> There are only pins of the EMD module, and instances of either .emd, .ibs, or .ibsx files. Connections are made between pins of the EMD file and the pins in the .emd, .ibs, or .ibsx files. (2) "Pkg", "Tx" and "Rx" are instantiated using two different keywords of "Package_Model" and "Class". However, their syntax and most of the contents look exactly the same. What are the differences between them? a. <WMK> The intent is "Package_Model" is the interconnect model for that pin. "Class" is to assign a package model to a class of pins. Similar to [Model_Package_IBIS-ISS] in the incremental .ibs file. (3) "Reference_Designator_Map" points to bare_die.ibsx file. There are no explicit mapping on pad names. Is the mapping implicit, requiring both files using the same pad names? a. <WMK> Yes (4) What is the difference between Tx and Tx_Coupled and, Rx and Rx_coupled? What are the benefits of making such distinctions at model time? a. Quite often all of the Rx channels got to the outside row of a BGA (eliminates vias on the Rx board where the series caps are in 802.3), and Tx go to inner rows of the BGA. (5) can you please provide examples of pkg_typ.mod pkg_min.mod pkg_max.mod. What's the syntax of these mod files? a. <WMK> I purposely left out the .mod files, this should be transparent to the changes to the IBIS files, they might look like: i. .subckt pkg_typ padP padM pinP pinM Length=.017 ii. W1 padP padM 0 pinP pinM 0 N=2 RLGCMODEL=rlgc L=Length iii. .model mname=rlgc . iv. .ends pkg_typ bare_die_ibsx: (1) Are sNp files allowed when N not equal 4 or 2? a. <WMK> The only sNp's that IC Vendors currently are delivering are s4p for differential SerDes drivers and receivers. The intent here was to demonstrate what IC Vendors want. I would like to see from IC Vendors what they require. More complicated on die models will require a section similar to the interconnect section for the package models. (2) Are "Buffer+ Buffer- Pad+ Pad-" reserved node names? What to do when N not equal 4 or 2? a. <WMK> Yes they are reserved node names. More complex on-die interconnect will require a section similar to the interconnect section for the package models with Model_Ports data. (3) Other than Tstonefile, we assume IBIS-ISS is also allowed, what is the syntax to hook it up? a. <WMK> More complex on-die interconnect will require a section similar to the interconnect section for the package models with Model_Ports data. The bottom line is Users are now requiring IC Vendors to deliver on-die interconnect as Touchstone files. IC Vendors do not want to supply IBIS-ISS because it discloses IP and is harder to do and support than delivering sNp. Bottom line is do not expect to see any SPICE circuits for channel on-die interconnect. Things may be different for on-die power distribution, but again we need to see examples from IC Vendors. (4) possible typo in line 80, missing "Pad+": (Tstonefile Ports Pad- Buffer+ Buffer-) <WMK> Typo, should be: (Tstonefile_Ports Pad+ Pad- Buffer+ Buffer-) legacy.ibs (1) is it the intention to introduce keywords in comment lines for future standards (ibis 6.0)? a. <WMK> No, the comment lines where my way of defining the problem that Arpad and I were to use to demonstrate the difference in our methodologies. James Zhou From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Walter Katz Sent: Tuesday, September 04, 2012 8:24 AM To: IBIS-ATM Subject: [ibis-macro] Package modeling comparison All, Enclosed is a zip file containing a presentation and implementations of "legacy.ibs" in legacy IBIS format, using legacy IBIS files with parameter tree sections, splitting the IBIS file into an EMD file (parameter tree format) and a bare-die legacy file, and finally converting the bare-die legacy file into an IBISx file. Mike L, Can you post this please to the IBIS Reflector. Walter Walter Katz wkatz@xxxxxxxxxx Phone 303.449-2308 Mobile 303.335-6156 _____ This message and any attached documents contain information from QLogic Corporation or its wholly-owned subsidiaries that may be confidential. If you are not the intended recipient, you may not read, copy, distribute, or use this information. If you have received this transmission in error, please notify the sender immediately by reply e-mail and then delete this message.