CL_INVALID_BINARY for larger design

Thanks for the reply. Both are plausible reasons, but I double-checked that both the --board parameter and the host memory allocation are correct.

I wonder if the number of local memories instantiated may cause a problem. In the somewhat cryptic sys_description.txt, the highest ID of a local memory in the working design seems to be 138, whereas the failing design has the highest ID 178. In both cases, some intermediate IDs seem to have been optimized away.

Is anyone successfully running designs with that many local memories?

I right now work on a synthetic example with a high number of local memories in order to find out if this might cause the problem.