Re: [radeonhd] 0x7187:0x174b:0x3000 ?Sapphire OEM? X1550 512MB
On Thu, Dec 13, 2007 at 09:24:24PM +0100, Lo?c Greni? wrote:
1) I don't known its id. It's an OEM card, I've a "Sapphire Inside" label on my case and one of the stickers on the back reads "X1550 512MB DDR2 PCI-E VGA/TVO/DVI-I". It looks nearly exactly like the Sapphire X1550 512MB that is on Newegg, except that it has a RoHS mention on one sticker (I'm in Europe) and the SKU# is 11093-19 (I cannot decipher it on Newegg's photo, but it does not seem to be the same).
Don't worry too much about this atm :)
2) The BIOS maps only 256MB of ram; it is related to: 3) The card and driver do not seem to know whether the card has 256 or 512MB of RAM (it prints 0xE00000000/28 as main memory, but the driver claims 512MB somewhere else). The bios message starts with "512mb".
This is no big deal right now... In the future it'd be nice if it were possible to remap the 512MB somewhere else (above the 4GB mark, because I've 4GB of DRAM).
Both are the exact same issue. The memory available on the card does not equal the memory we can access from the cpu directly. Whatever is in the non-directly-addressable part can be used by the card itself, but we currently are not doing anything that could make use of that memory anyway.
4) There is a secondary chip with id 0x71a7. It could be another X1550 without output registers (lspci -x below).
This exists only to fool some windows version. Ignore it.
5) Both outputs are changed together by xrandr. I've compiled a fairly recent version of the X server and xrandr (git of beginning of September) and, when I act on one output with xrandr, both change together. For instance, if I type
xrandr --output VGA_1 --right-of DVI-I_1/digital
both output go to (1024,0). I always get same resolution and same position for both outputs. The outputs are not in parallel because the EDID are different for both screens.
That's what I've found after some tries. With a single screen it works ok. With two it's a problem (and that's unfortunate because I'd like to make a two-seat computer).
I have no idea whether this is an RandR FAQ, a known RandR issue, or an issue that still has to be dug out. I hope that Matthias can shed some light on this one. Thanks, Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2007/12/15, Luc Verhaegen
On Thu, Dec 13, 2007 at 09:24:24PM +0100, Lo?c Greni? wrote:
4) There is a secondary chip with id 0x71a7. It could be another X1550 without output registers (lspci -x below).
This exists only to fool some windows version. Ignore it.
No problem. It'd be nice if the driver could match the corresponding (virtual) chipset with a driver associated to no screen. That would remove the warning. (No big deal obviously) Thanks a lot, Loïc Grenié
Luc Verhaegen. SUSE/Novell X Driver Developer. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Dec 15, 07 03:20:53 +0100, Luc Verhaegen wrote:
5) Both outputs are changed together by xrandr. I've compiled a fairly recent version of the X server and xrandr (git of beginning of September) [...] For instance, if I type
xrandr --output VGA_1 --right-of DVI-I_1/digital
both output go to (1024,0). I always get same resolution and same position for both outputs. The outputs are not in parallel because the EDID are different for both screens.
Fixed in git commit f7aaf8947a0f216ffedc7040b0a1ef153b471425 in
*November*. So September is certainly not recent enough.
And yes, this is a FAQ :P
Matthias
--
Matthias Hopf
2007/12/15, Luc Verhaegen
On Thu, Dec 13, 2007 at 09:24:24PM +0100, Lo?c Greni? wrote: [snip]
5) Both outputs are changed together by xrandr. I've compiled a fairly recent version of the X server and xrandr (git of beginning of September) and, when I act on one output with xrandr, both change together. For instance, if I type
xrandr --output VGA_1 --right-of DVI-I_1/digital
both output go to (1024,0). I always get same resolution and same position for both outputs. The outputs are not in parallel because the EDID are different for both screens.
That's what I've found after some tries. With a single screen it works ok. With two it's a problem (and that's unfortunate because I'd like to make a two-seat computer).
I have no idea whether this is an RandR FAQ, a known RandR issue, or an issue that still has to be dug out. I hope that Matthias can shed some light on this one.
I've just discovered today that "ati" driver seems to be able to drive the chip also (I'm not in front of the screen, so I can't check it effectively works, but it looks very much like it). RandR seems to work ok with the "ati" driver, while it does not with the "radeonhd" driver (everything is equal in both setups, I just change the Driver in xorg.con). I'll be seeting in front of the screen next week, probably on monday, so I can make further visual tests by then. Loïc Grenié -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Jan 03, 08 21:28:39 +0100, Loïc Grenié wrote:
5) Both outputs are changed together by xrandr. I've compiled a fairly recent version of the X server and xrandr (git of beginning of September)
That is not recent. The patches I've committed to xrandr were in late November. The radeon driver (to the best of my knowledge) does not reflect the full capabilities of the hardware in randr, therefore it probably does not fall into these pits. Please retry with a recent version of xrandr. This is actually a FAQ. If it doesn't work with current git, we would have to dig deeper into that.
and, when I act on one output with xrandr, both change together. For instance, if I type
xrandr --output VGA_1 --right-of DVI-I_1/digital
both output go to (1024,0). I always get same resolution and same
It fails to unclone the CRTC. The radeon driver probably doesn't use
cloning at all, and all bugs I encountered in xrandr were related to
cloning. The radeonhd driver uses cloning.
Matthias
--
Matthias Hopf
2008/1/7, Matthias Hopf
On Jan 03, 08 21:28:39 +0100, Loïc Grenié wrote:
5) Both outputs are changed together by xrandr. I've compiled a fairly recent version of the X server and xrandr (git of beginning of September)
That is not recent. The patches I've committed to xrandr were in late November. The radeon driver (to the best of my knowledge) does not reflect the full capabilities of the hardware in randr, therefore it probably does not fall into these pits.
Please retry with a recent version of xrandr. This is actually a FAQ. If it doesn't work with current git, we would have to dig deeper into that.
and, when I act on one output with xrandr, both change together. For instance, if I type
xrandr --output VGA_1 --right-of DVI-I_1/digital
both output go to (1024,0). I always get same resolution and same
It fails to unclone the CRTC. The radeon driver probably doesn't use cloning at all, and all bugs I encountered in xrandr were related to cloning. The radeonhd driver uses cloning.
It fails with current git xrandr and radeonhd. Should I also recompile the X server ? I've checked and all outputs are shown as clones. Loïc Grenié -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Jan 07, 08 14:18:20 +0100, Loïc Grenié wrote:
It fails with current git xrandr and radeonhd. Should I also recompile the X server ? I've checked and all outputs are shown as clones.
No, the Xserver had a patch as well, but that was IIRC not about this
issue. So there's probably still a bug lurking.
Could you please run the Xserver with -logverbose 7, and then invoke the
two xrandr commands? It would help a lot if you could annotate at which
places in the logfile you invoked the two commands.
Thanks
Matthias
--
Matthias Hopf
On Jan 07, 08 15:27:37 +0100, Matthias Hopf wrote:
Could you please run the Xserver with -logverbose 7, and then invoke the two xrandr commands? It would help a lot if you could annotate at which places in the logfile you invoked the two commands.
You might want to add yourself to freedesktop bugzilla #13905.
I think that is the same issue.
Matthias
--
Matthias Hopf
2008/1/7, Loïc Grenié
both output go to (1024,0). I always get same resolution and same
It fails to unclone the CRTC. The radeon driver probably doesn't use cloning at all, and all bugs I encountered in xrandr were related to cloning. The radeonhd driver uses cloning.
It fails with current git xrandr and radeonhd. Should I also recompile the X server ? I've checked and all outputs are shown as clones.
I've redownloaded and recompiled everything (btw, radeonhd driver build fails with current git xserver because xf86* libwrapper functions seem to have disappeared, I've just removed "xf86_ansic.h" from the sources). All outputs are still clones. Loïc Grenié -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Jan 08, 08 21:30:33 +0100, Loïc Grenié wrote:
I've redownloaded and recompiled everything (btw, radeonhd driver build fails with current git xserver because xf86* libwrapper functions seem to have disappeared, I've just removed "xf86_ansic.h" from the sources). All outputs are still clones.
???
xf86_ansic.h is only included if HAVE_XF86_ANSIC_H is defined - if the
file is not available any longer, but HAVE_XF86_ANSIC_H is defined this
is clearly a bug. Have you reconfigured radeonhd? configure should
detect that no xf86_ansic.h is available any longer.
CU
Matthias
--
Matthias Hopf
2008/1/9, Matthias Hopf
On Jan 08, 08 21:30:33 +0100, Loïc Grenié wrote:
I've redownloaded and recompiled everything (btw, radeonhd driver build fails with current git xserver because xf86* libwrapper functions seem to have disappeared, I've just removed "xf86_ansic.h" from the sources). All outputs are still clones.
???
xf86_ansic.h is only included if HAVE_XF86_ANSIC_H is defined - if the file is not available any longer, but HAVE_XF86_ANSIC_H is defined this is clearly a bug. Have you reconfigured radeonhd? configure should detect that no xf86_ansic.h is available any longer.
Sorry, my bad: xf86_ansic.h had not been removed from a previous install. I have no problem for compiling radeonhd. Loïc Grenié -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
participants (3)
-
Loïc Grenié -
Luc Verhaegen -
Matthias Hopf