Hi,

in order to get a better time resolution for benchmarking, I have
added the following in SVN 182:

In C/C++ I have added function cycle_counter() which returns the current cycle counter
of the CPU if the CPU supports it (and -1 if not).

In APL this functions is available via native function FILE_IO like this:

'lib_file_io.so' ⎕FX 'FILE_IO'
FILE_IO

⍝ read CPU cycle counter
FILE_IO ¯1
31854717246282

⍝ read CPU frequency
FILE_IO ¯2
2329921090

/// Jürgen

Reply via email to