>> As for the memcpy, I'd rather memcpy from buffers than wait for a
>> disk transfer - even a transfer that hits the disk's track cache.
> given that choice, I'd choose the memory-to-memory copy as well.

Gee, what a surprise. :-)

> but in the contexts where direct i/o is really targetted, the choice
> is between "wait for a disk tranfer" and "wait for a disk transfer
> and then also do the memory-to-memory", since the data is not going
> to be in the file system's cache.

That's why I've been trying to find a paradigm that lets the OS do
readahead, so that it _can_ be in the cache, and the choice _is_ then
between a memcpy and a disk transfer.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B