[mythtv-users] Optimizing MythTV for AMD64

[Kyle Rose]

> Thanks! This alone has actually been pretty good for me. I'm using NBC
> "Law & Order" as a test clip. I think that is 1080i based on what I've
> read about various networks,

CBS and ABC are generally 1080i; FOX is generally 720p; I'm not sure
about NBC.

> but I don't know how to just examine the
> data file to extract this information. I think it would be kind of
> cool for the Myth GUI (or MythWeb) to provide some interface for
> looking at technical data for recorded programs (for me, "technical"
> would be gratuitous geek information like 720p/1080i, SD broadcast
> over HD, etc.).

You can get this information from (I believe) --verbose playback on
mythfrontend.

> What kind of deinterlace algorithm are you using? I'm using Bob.
> (Would that even be relevant to this performance discussion?)

Bob, for sure.  It's the only one that makes any sense, IMO; and rather
fortuitously, it's also the cheapest (other than one-field, I guess, but
that isn't really a deinterlacer. :) )

> 	- Have you been using amd64-gcc-3.4 from the start, or had you
> 	  tried pure64 (gcc-3.3) before? I'm still in my enamored
> 	  honeymoon phase with MythTV, so I haven't started really
> 	  poking at my setup yet.

Originally, 3.4 didn't exist, so I was using pure64.  It may be that
pure64 doesn't have any of the problems I've seen 3.4 exhibit, which
would make sense since Ubuntu is based on it (or rather, they are both
based on the same set of packages).  Nonetheless, I'm really impressed
by Ubuntu's stability and maturity for being such a new distribution.
Being based on a distribution like Debian will do that. :)

> 	- What was the motivation for going to Ubuntu?

Twofold: Erik suggested it for reasons I'll explain Monday... and I had
a bad enough taste in my mouth from the 3.4 archive that I decided I
just wanted to go with something different for a change.  FWIW, once I
log in, there is very little difference.  The bootup looks different,
and gdm originally looked different until I configured it back to
sanity, but otherwise it really is Debian unstable-made-stable: newer
packages but much smoother.

> I've been wanting to upgrade, but pchdtv.com only has driver patches
> to the 2.6.9 kernel.

Separate frontend and backend.  Backend runs 2.6.9, frontends both run
some variant of 2.6.11.  There are some advantages to this setup, though
the electric bill is a source of some complaint.

> So this is implicitly --disable-xvmc, right? I also can't use XvMC.
> Myth can't start with some kind of xvmc "can't setup context" problem.

I just explicitly disabled it because I didn't want someone changing the
defaults on me.  I don't bother with XvMC because it seems buggy and my
machines are fast enough to do without it.

> Which drivers are you using for your HD-2000/HD-3000 cards?

Just 3000 at this point.  I'm using 1.6.

> As I mentioned above, I use gcc-3.4. The reason is because with
> gcc-3.3 I get this problem:
> 
> 	gcc-3.3 -c -pipe -march=k8 -w -O3 -Wall -Wno-switch
> 		-fomit-frame-pointer -fPIC -fno-common -D_REENTRANT
> 		-DMMX -DUSING_OPENGL_VSYNC -D_GNU_SOURCE
> 		-D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr/local\"
> 		-DQT_NO_DEBUG -DQT_THREAD_SUPPORT -DQT_SHARED
> 		-I/usr/share/qt3/mkspecs/default -I.
> 		-I/usr/local/include -I/usr/include/qt3 -o cpu_accel.o
> 		cpu_accel.c
> 	cc1: error: bad value (k8) for -march= switch
> 	cc1: error: bad value (k8) for -mcpu= switch

It sounds like you've got both pentium-builder installed, AND you aren't
  specifying --cpu=x86_64 on your configure line: the --cpu argument
keeps the Myth build system from adding -march=... to the compile line.
  You don't need it for AMD64 since there aren't any other variants of
that architecture to optimize for...yet.

Cheers,
Kyle