New message proposal for Fast-20

Gerald Maurer g_maurer at
Thu Jul 14 16:22:35 PDT 1994

 Reply to:   RE>New message proposal for Fast-20

Is increasing the extended message length warranted ?

The additional length doesn't break our hardware.  However, what is
the benefit in being able to specify the Period in 1 or .1 nanosecond
increments from 1 to 65565 ?

If instead this new message resembled the current SDTR message with
the multiplier being 1 nanosecond versus the current 4 nanoseconds, it
would provide enough resolution to specify Periods from 1 nanosecond
(1000 Mxfrs) to 255 nanoseconds (3.92 Mxfrs). This would cover
the original minimum Period of 200 nanoseconds (5 Mxfrs),
the FAST minimum Period of 100 nanoseconds (10 Mxfrs) and the FAST-20
minimum Period of 50 nanoseconds (20 Mxfrs). If a slower transfer
rate is desired revert back to the original SDTR message or go async.

It might also be simpler to implement in firmware since the
only difference between the current SDTR message and the new message
would be the extended message code (byte 2) and the multiplier for the
period. I believe in most cases this would require only minor changes
to the firmware routine that processes the current SDTR message. Instead 
of using a fixed multiplier of 4 nanoseconds the multiplier would
be determined by which of the two SDTR messages (byte 2) was received.

Gerald Maurer
QLogic Corporation

More information about the T10 mailing list