( ESNUG 255 Item 4 ) -------------------------------------------- [11/14/96]
From: dchapman@goldmountain.com (Dave Chapman)
Subject: My Internal Tri-State Vs. MUXes Buddhist Rules Of Thumb
Yo, John,
I liked your column in this week's EE Times about the endless discussions
over internal tri-state busses and muxing. The rule of thumb I use is:
- Less than/equal to 4 devices on a bus, multiplex.
- Less than/equal to 8 devices, probably mux, maybe tri-state.
- Less than/equal to 12 devices, probably tri-state.
- More than 12 devices, tri-state.
This is based of the trade-offs in synthesis of keeping track of tri-state
bus enables versus keeping track of the additional logic for the mux. It
is also based on the observation that the fan-in delays for muxes scale,
while the high-current driver delays are constant. (In other words, in
small cases the mux will be faster, while in large cases, tri-state will be
faster.) Also, for designs where there is a common enable for, say, 32
data lines, the control logic for tri-state will requires fewer gates than
the logic to support 32 instances of a 10-1 mux.
Note that future changes in logic technology will cause the trade-offs to
change (like you mentioned) but probably not by much.
- Dave Chapman
Goldmountain
|
|
