I was able to run the pipetest demo, using the original bit file only, and it runs to completion. However, I believe that the LEDs on the XEM6110 are configured to show the least significant bits of the error counter. If everything is good, I would expect them to stay unlit for the duration of the test.
They remain unlit for the “read” portion, but during the “write” portion, all of the LEDs begin to twinkle, indicating that at least 256 errors have happened, and they twinkle fast enough that I could believe that every write transaction is failing.
I can’t debug the design because when I resynthesize the sample HDL, it doesn’t work at all (cf my previous posting to this forum).