Discussion:
Nanosim with Synthesized Verilog
(too old to reply)
Sibi
2006-04-11 08:48:50 UTC
Permalink
Hello,

I have a question about Nanosim from Synopsys. I'm in a team that does
an ASIC design - we use the IBM Cu-11 library. I want to use Nanosim to
estimate the power of my design (I can provide input vectors etc to
Nanosim). Is this possible at all?

First of all, I tried to provide the synthesized verilog (structural
verilog) files as an input to Nanosim. But, I kept getting these
errors:

Netlist compilation took 0.350 s
nanosim: ERROR: illegal # of elements (0) in netlist

Second, I tried synthesizing the same behavioral verilog files and used
dc_shell to write the output in the EDIF format. I fed this input to
Nanosim and it was able to accept the design. But, the *.edif file had
references to the standard cells from the IBM library and Nanosim
complained that it could NOT find the subcircuit definition or the
functional model of these cells.

I do have the functional model files for the IBM Cu-11 library (in
terms of the *.v files) and the *.db and *.lib files. I'm not sure how
to feed these to Nanosim. I tried the cell_lib_path option in the
config file and also, -L flag. Both don't seem to work. Does anyone how
to get this setup to work? Do I need the Cu-11 library files in a
different format? How do I indicate to Nanosim about these library
files?

Or am I in a wrong domain and should probably be looking at PrimePower
or a similar tool?

I'd be very grateful if someone answers the above questions.

Thanks in advance,
G.Sibi
mk
2006-04-11 16:45:04 UTC
Permalink
Post by Sibi
Hello,
I have a question about Nanosim from Synopsys. I'm in a team that does
an ASIC design - we use the IBM Cu-11 library. I want to use Nanosim to
estimate the power of my design (I can provide input vectors etc to
Nanosim). Is this possible at all?
First of all, I tried to provide the synthesized verilog (structural
verilog) files as an input to Nanosim. But, I kept getting these
Netlist compilation took 0.350 s
nanosim: ERROR: illegal # of elements (0) in netlist
Second, I tried synthesizing the same behavioral verilog files and used
dc_shell to write the output in the EDIF format. I fed this input to
Nanosim and it was able to accept the design. But, the *.edif file had
references to the standard cells from the IBM library and Nanosim
complained that it could NOT find the subcircuit definition or the
functional model of these cells.
I do have the functional model files for the IBM Cu-11 library (in
terms of the *.v files) and the *.db and *.lib files. I'm not sure how
to feed these to Nanosim. I tried the cell_lib_path option in the
config file and also, -L flag. Both don't seem to work. Does anyone how
to get this setup to work? Do I need the Cu-11 library files in a
different format? How do I indicate to Nanosim about these library
files?
Or am I in a wrong domain and should probably be looking at PrimePower
or a similar tool?
You can use design compiler for power estimation by reading in your
toggle data and report_power commmand. To use nanosim, you need to
convert your verilog gate level to spice netlist; there must be a
utility in nanosim to do this for your. Then you read your gatelevel
spice netlist, the spice netlist of the ibm library (it must come with
the design kit) and the rc file in to nanosim and run your simulation
and integrate the current from your power supply to get an average.
Also peak current is useful to know too.

Loading...