I've observed that the function wavread, which needs to retrieve
some specific header data apart from the real audio data, uses an
mget call for each of them. For instance, it is necessary to
detect several string ID's each one followed by 4 byte integers.
I wonder if this is an efficient way to use the drive. I mean,
each mget requires reading the disk content which means
mechanically positioning the read-write head and picking the
However, it is known that on each read the head retrieves
typically a 4096 byte sector and the operating system may read at
once several sectors into a cluster.
Do operating systems or Scilab, at a low level, ensure that if one
performs several mget calls to retrieve closely packed data, the
information will be retrieved from some buffer secured during the
first physical read, instead of reading again the disk?