This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
Lee, When I run this on my SPARCstation, the output is: $ ./timeit Starting data initialization. Array sizes = 25MB. Clock resolution = 0.001 ms. Time for 25MB binary write = 0 msec. Time for 25MB 10 binary rewind/writes = 0 msec. Time for 25MB 8 binary file writes = 0 msec. Time for 25MB netcdf write = 0 msec. Spd ratio = 1.05 Time for 25MB 10 repeated netcdf write 0 msec. Spd ratio = 2.26 Time for 25MB 8 netcdf file writes = 0 msec. Spd ratio = 2.34 The "0 msec." times are a symptom that something is not right, maybe the program should be using nanoseconds instead of milliseconds for this. Also, I'm not sure how valid these times are, since a write may return immediately on Unix systems, since the write doesn't actually take place until the system buffers are flushed to disk, every 30 seconds or so, unless you actually do a close or explicit flush. I/O benchmarks like this can be difficult to interpret unless you know how the system is dealing with buffering and caching. I know almost nothing about how Fortran-90 I/O works on Unix systems, so take this with a little skepticism ... --Russ Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: NOJ-562667 Department: Support netCDF Priority: Normal Status: Closed