[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[f-cpu] ROP2 layout


as a "little exercise", i started to "synthesize"
the ROP2 unit using Alliance. In fact, it's mostly
a matter of placing and routing because i did the
synthesis by hand long ago.
I am using Alliance's "portable" cell library
and i did some routing attempts : i can probably
automate some part of it ("procedural placing"
will ease the work).
FFs take around one half of the surface
(we need 4+1 FF and 5 "logic" cells per row), which
corresponds approximately to the time ratio (one half
is spent in the wires...). A "normal" ROP2 unit
will be 64*50=3200 lambdas high and maybe 2500 lambdas wide.
it's already getting big.

BTW routing with Alliance is not "yet" OK because
even though placing is almost ok (at least it yields
working masks), the routing is still performed by
a commercial tool (there's a script that wraps around
"silicon ensemble"). Yagle seems to work, but parasitic
extraction and the likes are not yet ok : i don't know
what frequency the circuit can reach.

WHYGEE, special reporter in a strange world...
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/