On 06/19/2012 03:45 AM, Andrew Martens wrote:
We recently found a bug in the FFT for a specific parameter set that
causes data corruption that includes overflows. If you use 'embedded' as
your multiplier implementation of choice you may have this bug.
Thank you, Andrew and Jason, for that advice. I'll check with Joseph
Trinh when I get to JPL. He built the model.
Perhaps that will also deal with the problem I discovered next. I
changed the clock frequency to 900 MHz (in both cam.py and the Agilent
synthesizer) and powered cycled the ROACH. The roach passed rfi_init.py
just like at 800 MHz (except for the new system clock frequency).
However, rfi_spectrum.py died with a very unhelpful message. Something
like "problem detected". I'll try to find where that happens before we
go to the new bof file.
Cheers
Tom
