@Dinesh A congratulations on the bring-up! Did you have to use the clock skew adjustment features of your part to make it work? Were the final tap settings different from what you anticipated?
d
Dinesh A
03/18/2023, 5:21 PM
Yes, Clock skew adjust logic really helped me to bring-up the chip; Compare to Timing analysis, I see One Tap point shift was there in silicon in two of the blocks. In Timing analysis also I don't had common working Tap value for all three corner (Slow/Typical/Fast) - I was expecting Some Tap adjustment needed in silicon.
e
Eric Smith
03/18/2023, 5:28 PM
Interesting! I can see you expected to need them but was that always part of your bring up plan (in which case can you share when you did it in the overall bring up) or was there an error you observed that lead you to adjust? Do you have a published plan that you use to adjust them? I ask because this is a really useful circuit technique, that I think a lot of people could benefit from in their design, and I think this might be the only open source example of a working adjustable system. Certainly the only one I know of! ;)
d
Dinesh A
03/18/2023, 5:34 PM
In MPW-2 time, there was no clean Hierarchical Time closure methodology. Based on my experience i have add this clock adjustment block. Two month back only i have re-validated Timing analysis based the script released by efabless. I actually thought My MPW-2 chip will be dead on arrival chip.
Dinesh A
03/18/2023, 5:36 PM
One more thing helped: I have mapped individual macros important internal signals are mapped to 128 bit caravel la_data_out port. This helped me understand where individual blocks are hanging & which block clock adjustment need to be tried.