On Fri, Jan 07, 2000 at 08:30:01PM +0100, Robert Waldner wrote: > On Fri, 07 Jan 2000 18:42:30 GMT, Alisdair McDiarmid writes: > <ioload> > >> This sounds reasonable, as the effect also appears when cdparanoia uses > >> only 20 % of the CPU. (K6-2 350 / 64 MB RAM) > > > >Hrmph. So the solution is Don't Do That Then. > > hrmph.
Yeah :-) > so what for us who Already Did[1]? How to ease the pain? No idea. You could email the author of cdparanoia and ask him to work out a way of reducing interrupt usage. You could report it as a kernel bug - ppp and /dev/ttyS* should have higher priority than PIO ATAPI devices, perhaps. > btw, ripping from /dev/hdb, encoding with gogo (nice +20) _and_ > burning on /dev/scd0 (which is ide-scsi-emulation, in fact > /dev/hdc), speed=2, at the same time gives a high load, but works > fine otherwise. Uh, what? It stops PPP packet loss? And if you're not burning a CD on /dev/hdc, PPP drops packets? > is there some mechanism that balances the load between the two > ide-controllers or something? Probably not. > 1: buy an atapi cdrom ATAPI CDROMs really do suck. -- alisdair mcdiarmid [EMAIL PROTECTED] [the day is done but i'm having fun i think i'm dumb maybe just happy]
