Tim Waugh writes:
>On Fri, May 02, 2003 at 11:34:23PM -0700, Gora Mohanty wrote:
>> Is it really true that the speed difference is negligible?
>Well, it's only really true for bulk transfers in a defined IEEE 1284
>transfer mode like ECP, to be fair. For custom protocols it will be
>much slower in user-space, but for IEEE 1284 modes: it depends. When
>user-space write()s a large amount, the kernel is doing the transfer
>protocol stuff anyway, so in that sense the difference should be
>negligible for that situation.
>Basically I imagine it depends how much of the logic is in user-space.
OK, that makes perfect sense. As we are using a custom protocol, at some
point I will try and move to kernel space. I will report back on how much
that improved things, but it might take me a little while. Thanks also
for the information on the max. no of parallel ports supported.
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2b29 : Sat May 03 2003 - 22:39:56 EDT