Hi,
On Tue, 01 Feb 2000 21:29:10 -0800, "Jeffrey B. Siegal"
<jbs@quiotix.com> said:
> You might consider RAID, or even having your application separate the
> tracks onto multiple drives; you're pushing against what a single
> drive is capable of. Also, buffer much more than 20ms worth of data
> in your application before sending 20ms of it to the hardware buffer.
> This isolates streaming playback from variable disk access latencies.
In general, if you're having to combine multiple file streams in real
time, this is the only way to go -- you need to reduce the number of
seeks in order to get the necessary performance, so you just have to
read from each file in large blocks and then combine them internally,
rather than reading a bit from each file and expecting the O/S to
schedule many MB of readahead to reduce the seeks.
Eventually we hope to have a page cache prefetch API in the kernel which
will help in this sort of case, but right now you really do need to do
the chunking in the application.
--Stephen
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Feb 07 2000 - 21:00:08 EST