PCI experts...
I have a new machine 733Mhz with the i820 chipset that pulses the GNT# line inactive for 1 clock cycle in the middle of a bus master cycle. The REQ# line is continuously asserted during this event.
While I don't believe that this is illegal, it certainly is unusual. Why remove GNT#, unless the arbiter is going to grant the bus to some other master. Yet if another master was requesting the bus, then GNT# would not be reasserted on the next clock edge?
Anybody have a reason for this behavior or have a reason that it is illegal?
Tony Clark