Hi all, After several rounds of discussion on the back-channel BIRD, there are clearly some different views. The proposal we have has a couple of key ideas in it, which I want to state here for clarity: - AMI models should be seen as a black box. The spec should just define the API into/out of them, and give model developers the freedom to put whatever innovative behavior their SerDes has in there. It isn't up to this committee to legislate or hard-code what can be inside the black box. That goes against the basic spirit of the IBIS-AMI approach, and we view it as a dangerous path. Inside the algorithmic model is the modelers' domain. - Let's put the protocol-specific parameters into a separate .bci file that both Tx and Rx access. Protocol_Specific back-channel parameters are analogous to Model_Specific parameters in an AMI file. They can be defined without having to drive a multitude of new Reserved_Parameters into the IBIS spec for a new standard, and have to go through a long spec release and golden parser development cycle to get into the industry. - The job of the EDA tool is to simply pass the communication back and forth between the Tx and Rx, and let the models do their thing from an equalization standpoint. These basic ideas form the basis for our proposal. We will cover some more details on this in the next meeting. There are other ideas on backchannel that certainly have technical merit, but they do take us in a different direction. I encourage other members besides Cadence and SISoft to chime in with their views to help us get to closure on backchannel. Thanks, Ken [cid:image002.jpg@01CF648A.568A6A90] Ken Willis | Product Engineering Director - High Speed Analysis Products P: 980-245-7595 www.cadence.com<http://www.cadence.com>