Re: A7N8X-E Deluxe Stability with HD2000 / FBUS, FDSR, IRQ Lockup
john_freer at yahoo.com
Tue Nov 29 11:41:09 EST 2005
JF REPLY IN CAPS
On Mon, 2005-11-28 at 11:56 -0800, John Freer wrote:
> I've got an XP2600, 333FSB, in an Asus A7N8X-E Deluxe motherboard
> with a MX440 at 8x AGP. As tuners, I've got a PVR-250, HD-2000
> (dvb-bt8xx DVB Driver), and a Win-TV dbx (software btttv card).
That's almost exactly what I have. Same motherboard, AMD 2500+,
PVR-250, HD-2000, HD-5000 (new Air2PC).
> Assuming nothing else is going on, I can playback 1080i content
> ~5% free.
I only use a small percentage of the CPU, even with 1080i, thanks to
ON A SIDE NOTE, HOW DO YOU DEAL WITH THE XMVC STUTTER WITH OSD? THE
BEST SOLUTION I'VE FOUND SO FAR IS TO ENTER THE DATABASE AND MAKE THE
OSD FIELD NULL. THAT WAY, NO OSD IS USED AND THERE'S NOTHING TO
STUTTER. I ACHIEVE SIMILAR PERFORMANCE BENEFITS USING XVMC, BUT
GENERALLY DON'T USE IT BECAUSE OF STABILITY, ESPECIALLY WITH THE
CORRUPTED STREAMS I'M GETTING BECAUSE OF THE FBUS ERRORS.
> I'm running FC3 with 2.6.12 / 18.1 mythtv. All dist-upgraded to the
> latest at-testing.
I'm running 22.214.171.124 with DVB drivers from CVS. All with the latest
ANY REASON TO BELIEVE I'D GET BETTER RESULTS WITH 2.6.14 THAN 2.6.12?
> I have intermittant problems with HD recordings from the HD-2000.
> Then they end with something like
> bttv(0) IRQ lockup
I see a few of those in my /var/log/messages. I haven't correlated it
with failed recordings, but it's possible.
WHEN YOU SEE AN IRQ LOCKUP, THE CARD HAS QUIT EVEN THOUGH MYTH STILL
THINKS ITS RECORDING. THERE IS NO MORE MPEG2 STREAMING OFF THE CARD.
> There is also pixelation / data corruption while the FBUS errors
> going on. Its not a signal quality issue. Seems to correlate with
> processor load and/or HDD load.
This is a huge issue for me, and it's the same with the HD-5000 as
the HD-2000. I found that recording on the SATA drive caused too much
activity, but recording on the PATA drive usually works. I had to use
hdparm -d1 -m16 -a16 -A1 -u1 /dev/hda
to get it to work (the -m16 was the critical one); otherwise it was
worse with the PATA drive.
CURIOUS. MY DRIVE(S) DEFAULT SETTINGS WERE -m16 AND -U1. I COPIED
YOUR HDPARM SETTINGS AND NO IMPROVEMENT.
Ideally, the cards would have a large enough buffer on them to allow
bus contention, and the driver would make full use of it. I believe
that the video4linux driver for the HD-2000 didn't have glitches even
without the hdparm settings on the PATA drive, so it may be something
that improved drivers can solve.
MAYBE, BUT I AM STARTING TO GET THE FEELING THAT THIS MIGHT BE AN
ASUS / CHIPSET SPECIFIC ISSUE. I BELIEVE THAT SOME OTHERS ARE
RUNNING SIMILAR CONFIGS WITHOUT THE STABILITY PROBLEMS. AT THIS
POINT I WOULD GLADLY TRY A DIFFERENT MOBO IF SOMEONE COULD SUGGEST
More information about the mythtv-users