When extra pools were reported by Cray API, it was not multiplying
  the buffer count by granularity to get the total size, but was
  just getting the counts.