Re: initiator questions

["Richard Iachetta" <iachetta@us.ibm.com>]

>Beyond that, if you are doing a long burst that is broken up into smaller
>bus transactions, you should keep trying to do the same bus width that was
>accepted on the first bus transaction.

I'm not sure I agree with that.  If a transaction is disconnected by the
target (or even by the master for that matter), the master doesn't really
know if the continuation address will be to the same target or another.

>Also, remember that it doesn't hurt performance to try a 64-bit burst if
>the burst length is 4 DWORDs or longer, and the starting address has
>AD[2:0]==0.

Right, so why not just always try 64-bit?

Rich Iachetta
IBM Microelectronics Division -- Austin
World Wide Field Design Center
Phone: 512-838-6305   Tie Line: 678-6305