Hi, this may be a naive question but as per the instructions for resubmitting to mpw-7, we have to perform a static timing analysis. Is this also valid for analog projects in #caravan ? If so, do we follow the same instructions as for caravel? #tapeout-job @Mitch Bailey @Tim Edwards @Abhinav Uppal

Sorry, @Soumil Jain, I’m not involved with the STA portion.

If you didn't connect to the wishbone bus or logic analyzer, or your logic analyzer connections are for fixed or slow-rate configuration or read-back, then you don't need to worry about running STA.

Hi @Tim Edwards, we have a custom ReRAM array where we utilize LA connections as our comparator clock and select lines. We are planning to clock these at 40 MHz / 25 ns (so we can get the fastest throughput on computing with our ReRAM). We looked through the Makefile for running STA and it expects a gate-level netlist, however, since our design is fully-custom and made on caravan, we do not have any useful information in our GL verilog (its essentially an empty wrapper). If we should still attempt to run STA, how can we go about doing that from the final mag / gds files?

@Abhinav Uppal: For one, you're not going to get a logic analyzer output to run at 40MHz, since that's the core clock rate, and you can only drive a logic analyzer output by setting it in software, which implies a maximum rate of 20MHz, and in practice much lower, since your program will have to loop through some code, executing other instructions as well. Ignoring that detail, though, if you are driving only output from the logic analyzer into your project, then you only care about the timing of clock vs. select, where both are changing on the same core clock cycle, and presumably travel a similar path to get to the user project. And STA isn't going to tell you anything useful unless you can either make a liberty file representing your comparator controls, or if those controls are made with standard cells, you could mock up just the digital part and check that the signals switch with sufficient margin. But all that is kind of beside the point, anyway, since we don't have any known good timing for the sky130B variant, so if you got an STA result, it wouldn't necessarily be accurate.