> Have someone experienced troble with Csound on SGI machines producing
> gaps in the output file?

no.  can you post a stripped-down orc/sco, maybe alongside with a
more specific analysis of these silences?  I.e.,

  - is the problem pertinent to certain opcodes?
  - are you using the same binary on all machines?
    - if not, are some compiled with aggressive optimization?
  - are the silences "K-rate"-aligned?  Buffer-aligned?