peda__: hi, can anyone tell me if mergedfb is supposed to work with radeonhd?
deen: when i run mplayer to watch a video the video signal gets interrupted for a short time (at least both displays are black and have to "reconnect" again, just like when you change the resolution). has anyone experienced the same problem and maybe knows a solution?
ciaranm: deen: -vo what?
deen: no matter
deen: -vo {x11,gl,gl2,sdl}
ciaranm: well, of those only x11 works
deen: but even on the ones not working my screens go black
agd5f: peda__: use xrandr
peda__: agd5f: I did, just wanted to know if I didn't use it correctly or if clone mode is the only thing that works
agd5f: peda__: dualhead or clone should work
peda__: ok, thanks
agd5f: xrandr --output XXX --right-of YYY
agd5f: replace XXX/YY with the outputs you want to use
sirjaren: Hey guys, I was wondering if you could provide a bit of insight as to whether which hardware acceleration to use. I have an M56 ATI Mobility Radeon X1600 and I'm using the latest driver (1.2.1). I have found performance (2D acceleration) with 'exa' to be lousy compared to 'xaa' even though 'exa' is supposed to be the new standard. Shadowfb provides the best acceleration but at the cost of CPU cycles. I'm currently trying to build a few G
sirjaren: entoo systems via SSH and would like to be able to have a full blown desktop OS (Gnome) and to maximize the amount of CPU I can get in order to compile the OS's on the other computers in a speedier timeframe (without dropping out of X if at all possible). I have my mouse hardware accelerated as well. Any ideas or thoughts?
yangman: any CPU cycles used for 2D graphics will be negligible compared to compiling apps
yangman: it's really a non-issue
sirjaren: All right. Thank you though. I have even tried to optimize the CHOST for my exact compiler (I'm an optimization freak).
tulcod: sirjaren: ehm, optimizing cpu till that extreme PLUS x is a weird ieda
tulcod: *idea
sirjaren: Thanks again yangman.
ciaranm: oh dear
ciaranm: frickin' gentoo ricers
sirjaren: idea? :P
ciaranm: CHOST has no effect upon performance
ciaranm: go and stand in the corner
yangman: FWIW, gcc is far from a good optimizing compiler for modern x86 CPUs
sirjaren: What would you suggest then?
yangman: just stick with the standard -O2, and maybe -fomit-frame
kf: march=native works great for me
ciaranm: on x86, all that matters is -O2 and getting march right
yangman: anything beyond that is generally not worth the potential problems
ciaranm: where by right i mean none at all if you have a core2 and a gcc that doesn't do core2
tulcod: sirjaren: if you're an optimization freak, then don't use friggen X :-/
kf: wich is the point of march=native
tulcod: sirjaren: and even if you would, then don't use gnome
ciaranm: if you have march=native, you have march=core2
tulcod: sirjaren: and even if you would install gnome anyway, just switch it off
kf: detect cpu features and enable corresponding optimizations
tulcod: sirjaren: if you really want the maximum performance, go with 32 bit
ciaranm: looks at tulcod
ciaranm: you go and stand in the corner too!
sirjaren: Haha! Okay guys. Thank you!
tulcod: sirjaren: it's not a joke
sirjaren: ciaranm: lovin' the corner. :P
sirjaren: No, I understand.
tulcod: ciaranm: i wasn't saying i agreed wiht it
tulcod: just saying what could be improved
ciaranm: this is why gentoo users should be taken out and shot
sirjaren: What?
tulcod: ciaranm: racist :p
sirjaren: For asking a question? Did I impose elitism?
tulcod: ...
ciaranm: i should've talked daniel out of including "extreme performance" and "about choice" in the 'about' thing a long time ago
sirjaren: Well thanks for the information. I guess 'shadowfb' is fine and dandy. Anyhoo thanks for the driver though, much better than the blob that is fglrx. :)
ciaranm: unfortunately when he wrote it we were shipping egcs, which was a lot faster than gcc and a major selling point
tulcod: sirjaren: if you want even more performance, use gcc 3.4
ciaranm: ok, now we know tulcod is trolling
tulcod: ciaranm: well
tulcod: assuming you *NEED* gcc ;)
agd5f: sirjaren: radeon has pretty solid exa acceleration, but requires the radeon drm. this will eventually be available in radeonhd as well
sirjaren: agd5f: Looking forward to that. :)
rmh3093: is there a gitweb view of the radeonhd code?
tulcod: sirjaren: look forward to doing rc-update del clock, cause you won't need it in an optimized system
tulcod: rmh3093: yes
rmh3093: can you send me the link
tulcod: rmh3093: try google
sirjaren: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-radeonhd
rmh3093: thanks