Re: lantency scheduling benchmarks of audio playing tasks during high disk I/O

Alan Cox (alan@lxorguk.ukuu.org.uk)
Mon, 21 Jun 1999 19:05:38 +0100 (BST)

> i think this is a red herring - IRQ latencies _never_ cause even near to
> 15 msec delays even on low class 486 systems, unless there is some really
> bad device in the system. So unless someone really analyzes what is going
> on (i dont use soundcards myself so i cant really help) you might end up
> solving the wrong problem.

I've chased some down to two things

1. Poor device drivers - especially ISA ones on 2.2 not dropping locks Also intentionally bad drivers (IDE transfers without -u1 set etc)

2. Some socket7 chipsets dont seem to default to any kind of sane fair
sharing on PCI, a big UDMA IDE transfer stalls the CPU for ages, also
on some the CPU<->PCI transfers always beat busmastering PCI cards
meaning big graphics blits cause sound dropouts.

I asked for a quick config check - neither seem to be the case for this user.

-
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/