In my testing of recent kernels 2.6.0-test5 and 2.6.0-test4-mm4 (mm6
wouldn't cooperate with X for some reason and I didn't do much
investigation) I've experied an easily replicable and highly annoying
problem with Warcraft III and WineX 3.1 (prebuilt).
After playing 1 or 2 games, or leaving the game idle in the chat room,
the sound will eventually start to stutter and chop badly. In the
presence of this incredibly bad sound, the mouse and game respond just
fine (kudos to the scheduler on that point). Considering the game is
played in "real-time" and is full of audio cues I hate to imagine that
Con's scheduler will be the "official" scheduler of 2.6 without having
this issue addressed.
The kernels I use are tainted with nvidia's video drivers, 1.0.4496.
Nick's scheduler in 2.6.0-test4-mm5 seems to be the only thing capable
of correcting this problem. In general operation, mm5's scheduler
seems better at handling about everything I threw at it, with a rare
xmms skip once in a week of use.
I'm not a developer but I'd love some feedback and or questions to
help figure out why this happens with Con's scheduler patches in mm4
and test5 to help improve 2.6.0 altogether.
Other stats: gcc 3.2.3, AMD Athlon XP 2100+, 512MB RAM, ATA 100
drive, GF4 Ti4200. I'll also post CFLAGS upon request.
I am not subscribed to the mailing list. Please cc me on any replies,
thanks!
Tony Jones
aka
Sir-Tez
On Tue, 2003-09-09 at 23:52, Tony Jones wrote:
> In my testing of recent kernels 2.6.0-test5 and 2.6.0-test4-mm4 (mm6
> wouldn't cooperate with X for some reason and I didn't do much
> investigation) I've experied an easily replicable and highly annoying
> problem with Warcraft III and WineX 3.1 (prebuilt).
>
> After playing 1 or 2 games, or leaving the game idle in the chat room,
> the sound will eventually start to stutter and chop badly. In the
> presence of this incredibly bad sound, the mouse and game respond just
> fine (kudos to the scheduler on that point). Considering the game is
> played in "real-time" and is full of audio cues I hate to imagine that
> Con's scheduler will be the "official" scheduler of 2.6 without having
> this issue addressed.
>
> The kernels I use are tainted with nvidia's video drivers, 1.0.4496.
>
> Nick's scheduler in 2.6.0-test4-mm5 seems to be the only thing capable
> of correcting this problem. In general operation, mm5's scheduler
> seems better at handling about everything I threw at it, with a rare
> xmms skip once in a week of use.
>
> I'm not a developer but I'd love some feedback and or questions to
> help figure out why this happens with Con's scheduler patches in mm4
> and test5 to help improve 2.6.0 altogether.
>
> Other stats: gcc 3.2.3, AMD Athlon XP 2100+, 512MB RAM, ATA 100
> drive, GF4 Ti4200. I'll also post CFLAGS upon request.
Hi!
I'm testing the above kernel with mplayer, playing a non interleaved avi
video generates a heavy load even on a xp2100+, and tried to play the
same video under 2.6.0-test5 with con's O20 patch. My experience is
similar to yours, nick's patch is far better on the moment in heavy load
situation. The O20 patch has a side effect, it's somehow slowing down
the machine, e. g. when I compile the kernel under it, it takes much
more time to finish that.
Paco
2003-09-10, sze keltezéssel 07:01-kor Tony Jones ezt írta:
> I am not subscribed to the mailing list. Please cc me on any replies,
> thanks!
>
> Tony Jones
> aka
> Sir-Tez
>
>
> On Tue, 2003-09-09 at 23:52, Tony Jones wrote:
> > In my testing of recent kernels 2.6.0-test5 and 2.6.0-test4-mm4 (mm6
> > wouldn't cooperate with X for some reason and I didn't do much
> > investigation) I've experied an easily replicable and highly annoying
> > problem with Warcraft III and WineX 3.1 (prebuilt).
> >
> > After playing 1 or 2 games, or leaving the game idle in the chat room,
> > the sound will eventually start to stutter and chop badly. In the
> > presence of this incredibly bad sound, the mouse and game respond just
> > fine (kudos to the scheduler on that point). Considering the game is
> > played in "real-time" and is full of audio cues I hate to imagine that
> > Con's scheduler will be the "official" scheduler of 2.6 without having
> > this issue addressed.
> >
> > The kernels I use are tainted with nvidia's video drivers, 1.0.4496.
> >
> > Nick's scheduler in 2.6.0-test4-mm5 seems to be the only thing capable
> > of correcting this problem. In general operation, mm5's scheduler
> > seems better at handling about everything I threw at it, with a rare
> > xmms skip once in a week of use.
> >
> > I'm not a developer but I'd love some feedback and or questions to
> > help figure out why this happens with Con's scheduler patches in mm4
> > and test5 to help improve 2.6.0 altogether.
> >
> > Other stats: gcc 3.2.3, AMD Athlon XP 2100+, 512MB RAM, ATA 100
> > drive, GF4 Ti4200. I'll also post CFLAGS upon request.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
Tony Jones wrote:
>In my testing of recent kernels 2.6.0-test5 and 2.6.0-test4-mm4 (mm6
>wouldn't cooperate with X for some reason and I didn't do much
>investigation) I've experied an easily replicable and highly annoying
>problem with Warcraft III and WineX 3.1 (prebuilt).
>
>After playing 1 or 2 games, or leaving the game idle in the chat room,
>the sound will eventually start to stutter and chop badly. In the
>presence of this incredibly bad sound, the mouse and game respond just
>fine (kudos to the scheduler on that point). Considering the game is
>played in "real-time" and is full of audio cues I hate to imagine that
>Con's scheduler will be the "official" scheduler of 2.6 without having
>this issue addressed.
>
>The kernels I use are tainted with nvidia's video drivers, 1.0.4496.
>
>Nick's scheduler in 2.6.0-test4-mm5 seems to be the only thing capable
>of correcting this problem. In general operation, mm5's scheduler
>seems better at handling about everything I threw at it, with a rare
>xmms skip once in a week of use.
>
>I'm not a developer but I'd love some feedback and or questions to
>help figure out why this happens with Con's scheduler patches in mm4
>and test5 to help improve 2.6.0 altogether.
>
Actually, I'd love some feedback from you.
Use this: http://www.kerneltrap.org/~npiggin/v14/sched-rollup-v14.gz
It will apply against 2.6.0-test4 or test5 (not mm). See how you go.
Thanks