Integrated Circuit Engineering 2
Design submission e-mails should be sent to I.McNally@elec.canterbury.ac.nz and timestamped on arrival no later than 2pm on Monday 12th May.
The text of your e-mail must:
Passed | Failed | Not Attempted | |
---|---|---|---|
XXX_DESIGN Layout DRC | |||
XXX_DESIGN Layout SLS simulation | |||
XXX_DESIGN Layout T-Spice timing simulation | |||
XXX_DESIGN Schematic T-Spice simulation | |||
XXX_DESIGN Layout versus Schematic check |
The following files should be attached to your e-mail:
Details of these files are as follows:
L-Edit library file for your multiplier
S-Edit library file for your multiplier
LVS output file for the top level cells.
To help you with the LVS task a couple of files have been created:
XXX_DESIGN.vdb configures LVS to compare XXX_DESIGN.spc layout netlist with XXX_DESIGN.sp schematic netlist using the prematch file MULTIPLY.pre which lists all inputs and outputs for the multiplier design.
To run the final LVS check you should copy the two files into your own user space and open the XXX_DESIGN.vdb file using LVS. You will then have to modify the configuration to compare the your own layout and schematic netlists. If you find that the LVS fails, you may relax the checking options as follows:
You will need this option where the order of series transistors is not the same in the schematic and netlist. This may happen where the order of transistors in a leaf cell is not the same in the schematic and netlist but may also happen where the order of connections to a gate within a module is not the same. See the web document, Advice on LVS options, for more details.
You will need to turn off device parameter matching if you have not set the length and width values of transistors in the schematic leaf cells. See the web document, Advice on LVS options, for more details.
The file "XXX_DESIGN.out" will be created when the LVS is run. It is this file that you must submit.
SLS simulation output file
To help with final sls simulation, a couple of simulation stimulus files are available:
Both files perform 256 multiplications (0x0 0x1 ..... 15x14 15x15). In the case of multicycle.cmd, new operands are presented once every 6 clock cycles. This should be suitable for all designs. In the case of pipeline.cmd, new operands are presented every clock cycle. This should be suitable for all single cycle designs and pipeline designs.
To run a simulation with one of these files you would copy the file into your unix user space (i.e. save it under \\rockhopper\<username>\ ) and then run a unix simulation command like:
spacesls -cmd pipeline.cmd XXX_DESIGN.gds XXX_DESIGNSee the web document, Switch Level Simulation of L-Edit Designs, for more details.
Note that these are very simple stimulus files. A design which gives the correct results with these stimulus files does not necessarily meet the design specification.
Since there are 256 results to be checked you really need to post-process the output from such a simulation. For this reason, a new script "finalsls" has been provided:
finalsls XXX_DESIGN.gds XXX_DESIGN
When this command is run it invokes the "spacesls" script using the multicycle.cmd stimulus. After the waveforms are displayed you should close the waveform window (File -> Exit). The post processing will then commence, checking each answer against the expected result. The output from this checking will go to the screen and to a file called "XXX_DESIGN.results" which is the file that you should submit.
Please take time to read this document carefully. Marks will be deducted where any of the deliverables listed above are missing or not as specified.
If the one or other of the "XXX_DESIGN.tdb" and "XXX_DESIGN.results" files is not as specified the design will not be fabricated.
Note that it is not a prerequisite that the design passes all the test vectors in order to be fabricated, it is merely required that the cell names are correct, that the design passes the DRC and that a simulation is possible.
In addition to your e-mail you should also submit a printout of your top cell to facilitate the error checking procedure. A single page plot marked with your team name and posted into pigeon hole number 7 is sufficient.
Iain McNally
7-5-2003