At 06:38 PM 5/11/2002 +0200, Andreas Raab wrote:
>When I ran a few benchmarks I always came up with at least a factor of
>two in advantage of internal buffering even if the OS does file
>buffering. Of course, if you have large read operations (starting
>somewhere around 1/4 buffer size) you should go straight to the file
>system but for small reads internal buffering has huge advantages.

FYI, at least on Windows, a friend of mine who played with socket
buffering reports that not all buffer sizes are equal.  There are
some "sweet spots" where it seems to be notably more efficient.
Our hypothesis was that if the buffer nearly filled, but did
not exceed, the default TCP something (window, I think) that
it went faster.  The magic number for him was near 8k.
Note that these TCP parameters can vary from machine to machine.

Sorry not to have more definite information -- we discussed it
over the phone, so I have to rely on memory.

David Chase