Radeon Driver Gets Tear-Free X-Video

[phoronix]

Phoronix: Radeon Driver Gets Tear-Free X-Video

While there still is no open-source R600/700 3D documentation or code, Pierre Ossman and AMD's Alex Deucher working on the xf86-video-ati driver have added a feature that should please those with already supported GPUs. They have added what should be tear-free X-Video support! For those using EXA acceleration with this open-source ATI driver, this feature can be enabled through the EXAVSync option in your xorg.conf...

http://www.phoronix.com/vr.php?view=NjkwNQ

[deanjo]

Still tears on the 690g although not as bad. As the blog entries say:

We now have mostly tear free rendering in the radeon driver

/me reinstalls the nvidia card

[val-gaav]

great news ... though I'll wait for packages to try it :) Let's hope for an update in tormod repos in upcomming days :P

[mintcoffee]

Being an eager beaver, I've given this new driver a try. Tear-free accel def works on my FireGL V5200, but only in fullscreen mode. Xv video doesn't seem to work in windowed mode with composite enabled. Anyone else experience this issue? If so, can someone point me to the bug tracker? :)

[oibaf]

Being an eager beaver, I've given this new driver a try. Tear-free accel def works on my FireGL V5200, but only in fullscreen mode. Xv video doesn't seem to work in windowed mode with composite enabled. Anyone else experience this issue? If so, can someone point me to the bug tracker? :)

https://bugs.freedesktop.org/show_bug.cgi?id=18542#c39

[yoshi314]

if only using post-6.9.0-git snapshots of this driver didn't end up with blank, black screen on my box.

oh well, i'll try again and i guess i'll finally file a bug on it.

[Zhick]

My experience so far: On r500 cards you will get corruption in windowed video with the current git-version when you're using compositing. Even without there will be some corruption, though less notable (see the bugreport oibaf linked). But Alex Deucher already created a patch which fix this issue (attached to the same bugreport), but it seems not to be in git yet.
With the patch there's no tearing on videos without compositing, but still with compositing enabled.

[bluefiberoptics]

What does this mean for someone with a R300 card? I have a Mobility Radeon X600 and I run Ubuntu. I'm fairly new using Linux, so I get confused between the names of different sets of drivers. The binary ATI drivers are fglx I think. What are the drivers we are talking about and how can I get them? Thanks!



Nice to meet everyone!

[bridgman]

I think the most recent work was actually done on an r300 so you should be in good shape, although the code was only pushed yesterday so it could have problems even on another r300-family card ;)

There are three commonly used drivers for ATI/AMD graphics these days :

- fglrx (aka Catalyst) is the closed source driver we publish - supports from R3xx and up last time I checked

- radeon (aka -ati) is the original open source driver for Radeon family GPUs - supports from R100 up.

(The "ati" driver is actually a wrapper around mach64, r128 and radeon drivers which loads the appropriate driver depending on what hardware it finds. The wrapper is not really required these days but the -ati name sticks around because the "radeon" driver source is in the xf86-video-ati project)

- radeonhd is a more recent open source driver project, supporting r5xx and up, source in xf86-video-radeonhd

Most distros include both radeon and radeonhd drivers, although you need to check which versions are included to make sure you don't have something really old.

[bluefiberoptics]

Okay, so I'm assuming the driver that is used on a default install of Ubuntu on my notebook would be the "- radeon (aka -ati)" drivers. If they are updated, how do I get those updates? Will I receive updates through the update manager or do I need to enable backports or anything?

Also, do these open source drivers support 3D acceleration or is that reserved for the binary drivers? Thanks!

[TechMage89]

To get the tear-prevention code, though, you would need to build the radeon driver (xf86-video-ati) from git, because it's just been added.

[bridgman]

I believe you will get updates through the update manager but there are three driver components involved (the X driver, ie radeon/radeonhd), the drm (kernel) driver and the mesa (3d) driver) but I don't remember seeing updates for the drm or mesa drivers. Maybe someone who knows Ubuntu better can jump in.

The open source drivers (assuming you have all three components) currently support 3d on everything up to r5xx and rs690. 3D support for r6xx and r7xx (ie HD2xxx and higher) is being worked on now.

If you have a 5xx or earlier card then Ubuntu 8.10 comes up with 3d already enabled, so you can just turn on visual effects under "preferences" and enjoy the eye candy.

[bridgman]

To get the tear-prevention code, though, you would need to build the radeon driver (xf86-video-ati) from git, because it's just been added.

Yep, that's definitely worth mentioning ;)

I wouldn't expect Ubuntu packagers to create an update until the next formal release of the driver, so in the meantime you would need to build from source or pick up one of the "edgers" packages -- I don't remember the URL but maybe Tormod or someone else could help.

[russell_h]

I believe you will get updates through the update manager
I believe Ubuntu's policy is to generally only commit bug fixes to a given version of Ubuntu after release. Backports will have feature additions, but I doubt any graphics drivers are in there.

The open source drivers (assuming you have all three components) currently support 3d on everything up to r5xx and rs690. 3D support for r6xx and r7xx (ie HD2xxx and higher) is being worked on now.
So does that mean that this won't work on (for example) my HD3850? Or just that I just won't be able to get 3d acceleration at the same time?

[bridgman]

So does that mean that this won't work on (for example) my HD3850? Or just that I just won't be able to get 3d acceleration at the same time?

Right now there is no HW acceleration (2d, video, or 3d) on r6xx and up, including your HD3850. There is shadowfb support, which makes 2d and video ("X11" output) surprisingly useful, but no HW acceleration.

We have 2d accel working in house (agd5f got the overlapping src/dst problem fixed yesterday), and textured video "almost working" (it needs some shaders added to do YUV-RGB conversion), along with a demo program that shows how to program the engine for common 3d functions, so I think that's a good start. We are going through what we hope will be the final IP review to get all this released to the public.

The code and docs cover both 6xx (HD2xxx, HD3xxx) and 7xx (HD4xxx) families.

[russell_h]

We have 2d accel working in house, and video "almost working", along with a demo program that shows how to program the engine for common 3d functions, and are going through what we hope will be the final IP review for public release.

The code and docs cover both 6xx (HD2xxx, HD3xxx) and 7xx (HD4xxx) families.

Awesome! I got the card specifically for video playback hoping that you would come through with this, so thats great news (I realized I actually have a 4850, but it sounds like the same deal).

[tormod]

If you use plain Ubuntu 8.04 or 8.10, you can get an updated DDX (card driver) from my repository https://launchpad.net/~tormodvolden/+archive The packages are unofficial, but testing is welcome.

If you want to upgrade libraries, server and dependencies (for instance xserver 1.5 in Hardy or mesa trunk in Jaunty) then the xorg-edgers repository might be of interest to you. All disclaimers apply. Read the small text on https://launchpad.net/~xorg-edgers/+archive

[chrisr]

My machine doesn't lock up, but I still can't login.


(II) Module exa: vendor="X.Org Foundation"
compiled for 1.5.2, module version = 2.4.0
ABI class: X.Org Video Driver, version 4.1
(==) RADEON(0): Assuming overlay scaler buffer width is 1920
(II) RADEON(0): No MM_TABLE found - assuming CARD is not TV-in capable.
(!!) RADEON(0): For information on using the multimedia capabilities
of this adapter, please see http://gatos.sf.net.
(!!) RADEON(0): MergedFB support has been removed and replaced with xrandr 1.2 support
(--) Depth 24 pixmap format is 32 bpp
(II) do I need RAC? No, I don't.
(II) resource ranges after preInit:
[0] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
[1] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[2] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[3] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[4] 0 0 0x000a0000 - 0x000affff (0x10000) MS[B](OprU)
[5] 0 0 0x000b0000 - 0x000b7fff (0x8000) MS[B](OprU)
[6] 0 0 0x000b8000 - 0x000bffff (0x8000) MS[B](OprU)
[7] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[8] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
[9] 0 0 0x000003b0 - 0x000003bb (0xc) IS[B](OprU)
[10] 0 0 0x000003c0 - 0x000003df (0x20) IS[B](OprU)
(II) RADEON(0): RADEONScreenInit e0000000 0 0
Entering TV Save
Save TV timing tables
saveTimingTables: reading timing tables
TV Save done
(==) RADEON(0): Using 24 bit depth buffer
(II) RADEON(0): RADEONInitMemoryMap() :
(II) RADEON(0): mem_size : 0x10000000
(II) RADEON(0): MC_FB_LOCATION : 0xefffe000
(II) RADEON(0): MC_AGP_LOCATION : 0xffffffc0
(II) RADEON(0): Depth moves disabled by default
(**) RADEON(0): Using accelerated EXA DownloadFromScreen hook
(II) RADEON(0): Allocating from a screen of 262144 kb
(II) RADEON(0): Will use 32 kb for hardware cursor 0 at offset 0x007e9000
(II) RADEON(0): Will use 32 kb for hardware cursor 1 at offset 0x007ed000
(II) RADEON(0): Will use 8100 kb for front buffer at offset 0x00000000
(II) RADEON(0): Will use 8100 kb for back buffer at offset 0x007f1000
(II) RADEON(0): Will use 8100 kb for depth buffer at offset 0x00fda000
(II) RADEON(0): Will use 118784 kb for textures at offset 0x017c3000
(II) RADEON(0): Will use 119028 kb for X Server offscreen at offset 0x08bc3000
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 13, (OK)
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 13, (OK)
drmOpenByBusid: Searching for BusID pci:0000:01:00.0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 13, (OK)
drmOpenByBusid: drmOpenMinor returns 13
drmOpenByBusid: drmGetBusid reports pci:0000:01:00.0
(II) [drm] DRM interface version 1.3
(II) [drm] DRM open master succeeded.
(II) RADEON(0): [drm] Using the DRM lock SAREA also for drawables.
(II) RADEON(0): [drm] framebuffer handle = 0xe0000000
(II) RADEON(0): [drm] added 1 reserved context for kernel
(II) RADEON(0): X context handle = 0x1
(II) RADEON(0): [drm] installed DRM signal handler
(**) RADEON(0): Using AGP 8x
(II) RADEON(0): [agp] Mode 0x1f00420a [AGP 0x8086/0x2550; Card 0x1002/0x4153]
(II) RADEON(0): [agp] 131072 kB allocated with handle 0x00000001
(II) RADEON(0): [agp] ring handle = 0xc8000000
(II) RADEON(0): [agp] Ring mapped at 0xb7e3b000
(II) RADEON(0): [agp] ring read ptr handle = 0xc8101000
(II) RADEON(0): [agp] Ring read ptr mapped at 0xb7e3a000
(II) RADEON(0): [agp] vertex/indirect buffers handle = 0xc8102000
(II) RADEON(0): [agp] Vertex/indirect buffers mapped at 0xa7b97000
(II) RADEON(0): [agp] GART texture map handle = 0xc8302000
(II) RADEON(0): [agp] GART Texture map mapped at 0x9ff97000
(II) RADEON(0): [drm] register handle = 0xff8f0000
(II) RADEON(0): [dri] Visual configs initialized
(II) RADEON(0): RADEONRestoreMemMapRegisters() :
(II) RADEON(0): MC_FB_LOCATION : 0xefffe000 0x1fff0000
(II) RADEON(0): MC_AGP_LOCATION : 0xffffffc0
(==) RADEON(0): Backing store disabled
(II) RADEON(0): [DRI] installation complete
(II) RADEON(0): [drm] Added 32 65536 byte vertex/indirect buffers
(II) RADEON(0): [drm] Mapped 32 vertex/indirect buffers
(II) RADEON(0): [drm] dma control initialized, using IRQ 16
(II) RADEON(0): [drm] Initialized kernel GART heap manager, 130023424
(WW) RADEON(0): DRI init changed memory map, adjusting ...
(WW) RADEON(0): MC_FB_LOCATION was: 0xefffe000 is: 0xefffe000
(WW) RADEON(0): MC_AGP_LOCATION was: 0xffffffc0 is: 0xcfffc800
(II) RADEON(0): RADEONRestoreMemMapRegisters() :
(II) RADEON(0): MC_FB_LOCATION : 0xefffe000 0xefffe000
(II) RADEON(0): MC_AGP_LOCATION : 0xcfffc800
(II) RADEON(0): Direct rendering enabled
(II) RADEON(0): Render acceleration enabled for R300/R400/R500 type cards.
(II) RADEON(0): Setting EXA maxPitchBytes
(II) RADEON(0): EXA VSync enabled
(II) RADEON(0): num quad-pipes is 1
(**) RADEON(0): Option "MigrationHeuristic" "greedy"
(II) EXA(0): Offscreen pixmap area of 121884672 bytes
(II) EXA(0): Driver registered support for the following operations:
(II) Solid
(II) Copy
(II) Composite (RENDER acceleration)
(II) UploadToScreen
(II) DownloadFromScreen
(II) RADEON(0): Acceleration enabled
(**) Option "dpms"
(**) RADEON(0): DPMS enabled
(==) RADEON(0): Silken mouse enabled
(II) RADEON(0): No video input capabilities detected and no information is provided - disabling multimedia i2c
(II) Loading sub module "theatre_detect"
(II) LoadModule: "theatre_detect"

(II) Loading /usr/lib/xorg/modules/multimedia//theatre_detect_drv.so
(II) Module theatre_detect: vendor="X.Org Foundation"
compiled for 1.4.99.905, module version = 1.0.0
ABI class: X.Org Video Driver, version 4.1
(II) RADEON(0): no multimedia table present, disabling Rage Theatre.
(II) RADEON(0): Set up overlay video
(II) RADEON(0): Set up textured video
init memmap
init common
init crtc1
init pll1
freq: 119000000
best_freq: 119000000
best_feedback_div: 238
best_ref_div: 27
best_post_div: 2
restore memmap
(II) RADEON(0): RADEONRestoreMemMapRegisters() :
(II) RADEON(0): MC_FB_LOCATION : 0xefffe000 0xefffe000
(II) RADEON(0): MC_AGP_LOCATION : 0xcfffc800
restore common
restore crtc1
restore pll1
finished PLL1
restore FP
(II) RADEON(0): RandR 1.2 enabled, ignore the following RandR disabled message.
init memmap
init common
init crtc1
init pll1
freq: 119000000
best_freq: 119000000
best_feedback_div: 238
best_ref_div: 27
best_post_div: 2
restore memmap
(II) RADEON(0): RADEONRestoreMemMapRegisters() :
(II) RADEON(0): MC_FB_LOCATION : 0xefffe000 0xefffe000
(II) RADEON(0): MC_AGP_LOCATION : 0xcfffc800
restore common
restore crtc1
restore pll1
finished PLL1
restore FP
(--) RandR disabled
(II) Initializing built-in extension MIT-SHM
(II) Initializing built-in extension XInputExtension
(II) Initializing built-in extension XTEST
(II) Initializing built-in extension XKEYBOARD
(II) Initializing built-in extension XINERAMA
(II) Initializing built-in extension XFIXES
(II) Initializing built-in extension RENDER
(II) Initializing built-in extension RANDR
(II) Initializing built-in extension COMPOSITE
(II) Initializing built-in extension DAMAGE
(II) Initializing built-in extension XEVIE
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 14, (OK)
drmOpenByBusid: Searching for BusID pci:0000:01:00.0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 14, (OK)
drmOpenByBusid: drmOpenMinor returns 14
drmOpenByBusid: drmGetBusid reports pci:0000:01:00.0
(II) AIGLX: enabled GLX_MESA_copy_sub_buffer
(II) AIGLX: enabled GLX_SGI_swap_control and GLX_MESA_swap_control
(II) AIGLX: enabled GLX_texture_from_pixmap with driver support
(II) AIGLX: Loaded and initialized /usr/lib/dri/r300_dri.so
(II) GLX: Initialized DRI GL provider for screen 0
(II) RADEON(0): Setting screen physical size to 434 x 270
(**) Option "Protocol" "auto"
(**) Option "Device" "/dev/input/mice"
(II) Mouse0: Setting mouse protocol to "ExplorerPS/2"
(**) Mouse0: Device: "/dev/input/mice"
(**) Mouse0: Protocol: "auto"
(**) Option "CorePointer"
(**) Mouse0: always reports core events
(**) Option "Device" "/dev/input/mice"
(**) Option "Emulate3Buttons" "no"
(**) Option "ZAxisMapping" "4 5"
(**) Mouse0: ZAxisMapping: buttons 4 and 5
(**) Mouse0: Buttons: 9
(**) Mouse0: Sensitivity: 1
(**) Option "CoreKeyboard"
(**) Keyboard0: always reports core events
(**) Option "Protocol" "standard"
(**) Keyboard0: Protocol: standard
(**) Option "AutoRepeat" "500 30"
(**) Option "XkbRules" "xorg"
(**) Keyboard0: XkbRules: "xorg"
(**) Option "XkbModel" "pc105"
(**) Keyboard0: XkbModel: "pc105"
(**) Option "XkbLayout" "gb"
(**) Keyboard0: XkbLayout: "gb"
(**) Option "CustomKeycodes" "off"
(**) Keyboard0: CustomKeycodes disabled
(II) evaluating device (Mouse0)
(II) XINPUT: Adding extended input device "Mouse0" (type: MOUSE)
(II) evaluating device (Keyboard0)
(II) XINPUT: Adding extended input device "Keyboard0" (type: KEYBOARD)
(II) Mouse0: Setting mouse protocol to "ExplorerPS/2"
(II) Mouse0: ps2EnableDataReporting: succeeded
(II) RADEON(0): Damage tracking initialized for page flipping
(II) config/hal: Adding input device ImPS/2 Generic Wheel Mouse
(II) LoadModule: "evdev"

(II) Loading /usr/lib/xorg/modules/input//evdev_drv.so
(II) Module evdev: vendor="X.Org Foundation"
compiled for 1.5.0, module version = 2.0.7
Module class: X.Org XInput Driver
ABI class: X.Org XInput driver, version 2.1
(**) ImPS/2 Generic Wheel Mouse: always reports core events
(**) ImPS/2 Generic Wheel Mouse: Device: "/dev/input/event3"
(II) ImPS/2 Generic Wheel Mouse: Found x and y relative axes
(II) ImPS/2 Generic Wheel Mouse: Found mouse buttons
(II) ImPS/2 Generic Wheel Mouse: Configuring as mouse
(II) XINPUT: Adding extended input device "ImPS/2 Generic Wheel Mouse" (type: MOUSE)


and then it all stops.

[agd5f]

My experience so far: On r500 cards you will get corruption in windowed video with the current git-version when you're using compositing. Even without there will be some corruption, though less notable (see the bugreport oibaf linked). But Alex Deucher already created a patch which fix this issue (attached to the same bugreport), but it seems not to be in git yet.
With the patch there's no tearing on videos without compositing, but still with compositing enabled.

The corruption with compositing has been fixed properly now. The Xv anti-tear stuff won't help with composite since the video is rendered offscreen and then composited onto the desktop later. In that case, you can enable EXA and the EXAVsync option to apply the same techniques to EXA (for metacity or other compositers that use render). Since Compiz uses GL, you'll have to use the compiz vsync options.

[oyvind]

Wow, textured video with the latest xf86-video-ati GIT snapshot is just perfectly in sync. No tearing at all, here. Tested with a 60fps video, and it was smooth. Great ! So I don't need to use VSYNC-ed OpenGL output, anymore. Note that I don't use Compiz, nor EXA acceleration (because of corruption issues).

Ubuntu/8.10 x86, ATI X1400 radeon mobile.

[agd5f]

nor EXA acceleration (because of corruption issues).

If you haven't filed a bug for this yet, please do. We can't fix bugs we don't know about.

[dungeon]

First when i enable EXAVSync i see gtkperf performance drop (from 10s to 22s) and 'man radeon' said "It reduces tearing at the cost of performance." == that is OK... Then i start tvtime (overlay port) and got hard freeze. Reboot on case button. Then i start 'xdtv' (now with textured video port), second hard freeze. Reboot on case button. off for EXAVSync. Write this.

r200 (9250) here

[korpenkraxar]

If you use plain Ubuntu 8.04 or 8.10, you can get an updated DDX (card driver) from my repository https://launchpad.net/~tormodvolden/+archive The packages are unofficial, but testing is welcome.


Just tried the latest version (6.9.0+git20081206.979ad04d-0ubuntu0tormod) on Ubuntu 8.10 64-bit running a Thinkpad Z61m with an X1400 ATI chip and I'll be damned! :eek:

Never seen such smooth video playback of Big Buck Bunny (big_buck_bunny_1080p_surround.avi) on this laptop before using either previous radeon or fglrx drivers (and trust me about using that video as a benchmark, my one and a half year old daughter has us see that one almost once a day).

I need to toggle Compiz off using fusion-icon though or there will still be tearing. I have nothing else manually specified in my xorg.conf than the EXA option:

Section "Device"
Identifier "Configured Video Device"
Option "AccelMethod" "EXA"
EndSection

The radeon driver has come a long long way this year and I can finally fully enjoy using this laptop without breaking into tears :p

To everyone involved in improving, packaging and testing this driver, thanx for your hard work and happy holidays!

bridgman, I'm sure you know what we all want for this Christmas this year. Apart from being grumpy about fglrx and impressed by Nvidias VDPAU I've been a fairly good boy :)

[bridgman]

Umm... World peace ? A chicken in every pot ? Carmen Electra ?

:D

[oyvind]

If you haven't filed a bug for this yet, please do. We can't fix bugs we don't know about.

Ah, well, I am the person responsible for reporting these bugs:

https://bugs.freedesktop.org/show_bug.cgi?id=18399
https://bugs.freedesktop.org/show_bug.cgi?id=18400
https://bugs.freedesktop.org/show_bug.cgi?id=18591
https://bugs.freedesktop.org/show_bug.cgi?id=18398
https://bugs.freedesktop.org/show_bug.cgi?id=18397

All EXA corruption related, four of them originally reported through the Ubuntu bug system ;)

Regards,
Øyvind S.

[susikala]

Hm, maybe I got something wrong, but how do you enable XV at all with the Radeon driver? I'm using the latest one (1:6.9.0+git20081003.f9826a56-0ubuntu4) from Jaunty at the moment, and 'xvinfo' returns:

X-Video Extension version 2.2
screen #0
no adaptors present

Is there any specific configuration option I should enable? I'm using a HD3200 IGP.

[agd5f]

Is there any specific configuration option I should enable? I'm using a HD3200 IGP.

Xv is only available on r1xx-r5xx chips. Accel code (including Xv) for r6xx/r7xx chips (the hd3200 is r6xx based) is in progress and will be released as soon as the IP review is finished.

[susikala]

Xv is only available on r1xx-r5xx chips. Accel code (including Xv) for r6xx/r7xx chips (the hd3200 is r6xx based) is in progress and will be released as soon as the IP review is finished.

Ah, thanks a bunch! I thought that applied only to RadeonHD.

[tormod]

I'm using the latest one (1:6.9.0+git20081003.f9826a56-0ubuntu4) from Jaunty at the moment, and
You might want to try the updated Intrepid driver from my personal repository at https://launchpad.net/~tormodvolden/+archive
It should run fine in Jaunty.

[mirak63]

this is readon driver and not readonhd, right ?

[bridgman]

Correct, although it shouldn't be too hard to port across to radeonhd. I believe Pierre was working with an r300 which is only supported by radeon.

[mirak63]

ok so hd4xx0 are not supported by radeon driver ?

[oyvind]

Blargh, latest radeon GIT snapshot re-introduced tearing for me .. Now I have a triangle in upper right corner which is out-of-sync and thus causes tearing pattern. It was fun while it lasted, back to OpenGL-output :). For anyone else following the GIT-snapshots, here's the reason:
http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=f892154dbd0d8b6dd8abdf2196ab0c7c18327412

I'm sure it'll get sorted out eventually.
ATI X1400 mobile radeon, Ubuntu 8.10 x86.

[bridgman]

ok so hd4xx0 are not supported by radeon driver ?

Not for Xv.

[val-gaav]

There seems to be a regresion in latest svn code ...

With tormod package from 14 december I experience nice tear free video play ...
However tormod update from 16 december re introduced tearing for me...
I'm on rs690 kubuntu 64bit ...

Anyone can confirm this ?

[tormod]

Anyone can confirm this ?
That's what Øyvind said two comments ago, right?

BTW, the hex word after the date in my package version identifies the last git commit in that package, so please use that when referring to which packages you talk about. For instance, latest now is "2346fd7c".

[oyvind]

That's what Øyvind said two comments ago, right?

BTW, the hex word after the date in my package version identifies the last git commit in that package, so please use that when referring to which packages you talk about. For instance, latest now is "2346fd7c".

I simply reverted commit f892154dbd0d8b6dd8abdf2196ab0c7c18327412 in my git-tree ("R3xx-R5xx: switch back to quad rendering for Xv"), and tear-free XVideo is back and working nicely on my ATI X1400. I'm not affected by the problem that this commit addresses, so I thought why not.

[val-gaav]

That's what Øyvind said two comments ago, right?

Right, that makes me look really stupid ... For the future I really shouldn't post after drinking alcohol.
As for hex word, sure I'll use it next time ...

[tormod]

Right, that makes me look really stupid ... For the future I really shouldn't post after drinking alcohol.
lol we got the issue confirmed at least :) I am pretty sure the developers will sort this out soon.

[russell_h]

lol we got the issue confirmed at least :) I am pretty sure the developers will sort this out soon.

I'm seeing a commit with the comment "R3xx-R5xx: better fix for xv primitive tearing issues" from a few hours ago, so you might give it another shot now.

http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/

[chrisr]

Hi,

I have a Radeon 9550 card, and I thought I'd give the latest DDX driver a spin on my Fedora 9 machine (vanilla 2.6.27.10 kernel). The git log's last entry is:

commit c0c33dab44e6966b1702d4e8cfba3537fc6e2d5c
Author: Patrick Haller <patrick.haller@haller-consult.com>
Date: Mon Dec 22 03:06:23 2008 -0500

Fix off by one in EXA composite limit checking

Patch from Patrick, with some updates from me:
- fix r200 limits
- note about r100 limits

However, as soon as I tried logging in, Xorg went into an endless spin-cycle with 100% usage of one CPU. I was eventually able to ssh in from a remote box and swap the driver back out again, but although I managed to "neutralize" the rogue process, I couldn't remove it from the process table completely without rebooting!

Am I missing a dependency somewhere? E.g. does the DDX driver in git need Fedora 10's version of Xorg?

[sundown]

Wow, I'm just extreamely happy! This brought some tears in my eyes.

Recently Debian moved this to the experimental repos and only now have I been able to fully test it.

It freaking does work!

Thank you, guys!!!

[agd5f]

Hi,

I have a Radeon 9550 card, and I thought I'd give the latest DDX driver a spin on my Fedora 9 machine (vanilla 2.6.27.10 kernel). The git log's last entry is:

commit c0c33dab44e6966b1702d4e8cfba3537fc6e2d5c
Author: Patrick Haller <patrick.haller@haller-consult.com>
Date: Mon Dec 22 03:06:23 2008 -0500

Fix off by one in EXA composite limit checking

Patch from Patrick, with some updates from me:
- fix r200 limits
- note about r100 limits

However, as soon as I tried logging in, Xorg went into an endless spin-cycle with 100% usage of one CPU. I was eventually able to ssh in from a remote box and swap the driver back out again, but although I managed to "neutralize" the rogue process, I couldn't remove it from the process table completely without rebooting!

Am I missing a dependency somewhere? E.g. does the DDX driver in git need Fedora 10's version of Xorg?

Remove the EXAVSync option if you have it enabled in your config. It doesn't seem to work reliably on all r3xx cards at the moment.

[hobbes]

Wow, I'm just extreamely happy! This brought some tears in my eyes.

Recently Debian moved this to the experimental repos and only now have I been able to fully test it.

It freaking does work!

Thank you, guys!!!

I've just updated to very latest ubuntu jaunty packages and the problem (tearing) persists. :(

xserver-xorg-video-ati (1:6.9.0.91-1ubuntu1) jaunty; urgency=low

* Merge from debian experimental, remaining changes:
- Add 104_use_exa.patch: Switches to EXA acceleration by default.

xserver-xorg-video-ati (1:6.9.0.91-1) experimental; urgency=low

* New upstream release candidate.

xserver-xorg-video-ati (1:6.9.0+git20081129.783cdb73-1) experimental; urgency=low

* Pull upstream snapshot, up to commit 783cdb73.
+ Add AGPMode quirk table, closes: #461144, #462590, #467460.

xserver-xorg-video-ati (1:6.9.0+git20081012.c0e6cb6d-1) experimental; urgency=low

* Pull upstream snapshot, up to commit c0e6cb6d, closes: 500903.

Date: Fri, 02 Jan 2009 13:27:30 +0200
Changed-By: Timo Aaltonen <tjaalton at ubuntu.com>
Maintainer: Ubuntu X-SWAT <ubuntu-x at lists.debian.org>
https://launchpad.net/ubuntu/jaunty/+source/xserver-xorg-video-ati/1:6.9.0.91-1ubuntu1

$ egrep EXA* /var/log/Xorg.0.log
(**) RADEON(0): Option "EXAVSync" "1"
(==) RADEON(0): Using EXA acceleration architecture
(==) RADEON(0): Not using accelerated EXA DownloadFromScreen hook
(II) RADEON(0): Setting EXA maxPitchBytes
(II) RADEON(0): EXA VSync enabled

$ egrep "AGP 8x" /var/log/Xorg.0.log
(**) RADEON(0): Using AGP 8x

(--) RADEON(0): Chipset: "ATI Radeon X1600" (ChipID = 0x71c2)
My Hardware:

Sapphire AGP x1600pro 512MB

Apparently, we have the very same packages by now and no tear-free xv for me.

Any help would be appreciated.

Thanks For Reading.

[TechMage89]

You don't need EXAVsync for Xv vsync.

It works for me, with an x1600xt and the 6.9.1 RC driver (no tearing on Xv at all)

What media player are you using? Are you sure it's using Xv for video out? OpenGL still isn't vsynced.

[hobbes]

You don't need EXAVsync for Xv vsync.

It works for me, with an x1600xt and the 6.9.1 RC driver (no tearing on Xv at all)

What media player are you using? Are you sure it's using Xv for video out? OpenGL still isn't vsynced.

Are you using Jaunty Alpha 2 with the very latest packages?

Totem is my main player.

Compiz is on by the way and it gets some tearing (horizontal lines) when I zoom in/zoom out on a gnome-terminal window or move around nautilus.

yes, Totem is set to use Xv (as default)

[bridgman]

Tear free won't work if you have Compiz running; it will just deliver tear-free video to Compiz then Compiz will copy it in a tear-y manner to the screen.

I am not aware of a tear-free solution for Compiz.

[hobbes]

Tear free won't work if you have Compiz running; it will just deliver tear-free video to Compiz then Compiz will copy it in a tear-y manner to the screen.

I am not aware of a tear-free solution for Compiz.


Hum.. now I see.

So, GEM/DRI2 Won't actually solve this issue in the future?

[bridgman]

So, GEM/DRI2 Won't actually solve this issue in the future?

Not really; DRI2 (actually RDR, built on DRI2) will help to solve the flicker problems when direct-rendering 3D apps run under Compiz, but flickering and tearing are two different problems.

What is really needed is for Compiz to synchronize *it's* drawing operations to vertical retrace (rather than the individual driver stacks like Xv) and for it to have a flow control mechanism to tell the individual apps when to start and stop drawing. My guess is that most of what is needed is in place but nobody is looking at the end-to-end solution.

[calica]

My guess is that most of what is needed is in place but nobody is looking at the end-to-end solution.

And it needs to be standardized across compositing managers. It would be horrible for Compiz and kwin to do it differently.

[hobbes]

Not really; DRI2 (actually RDR, built on DRI2) will help to solve the flicker problems when direct-rendering 3D apps run under Compiz, but flickering and tearing are really two completely different problems.

What is really needed is for Compiz to synchronize *it's* drawing operations to vertical retrace (rather than the individual driver stacks like Xv) and for it to have a flow control mechanism to tell the individual apps when to start and stop drawing. My guess is that most of what is needed is in place but nobody is looking at the end-to-end solution.

I just tested without using compiz and the video playback is AWESOME. Really AWESOME. :)

Thanks to everyone involved!

Oh man, I hope the compiz devs (anyone here?) get into it ASAP, because a tear-free experience running compiz is a MUST. :(

I can't live without compiz and switching on/off compiz is not a solution

Just visualize you or somebody else trying to explain that to all new users who tests Ubuntu every new release.

Think about them!

[bridgman]

In the short term, *if* you are playing video full screen then the "don't redirect full screen windows" option in Compiz will help. It also relieves the heartbreak of flickering 3d (again, only when fullscreen) so is a Good Thing to turn on.

It's only a workaround, of course, but it's a lot easier to remind someone to "make it fullscreen" than to explain that they have to turn visual effects on and off.

[gordboy]

x1650 with xorg 1.5.3/radeon 6.10 and all the trimmings. The video playback is nice, and so is 2D/3D. I don't use AIGLX or Composite, and all the VT switching etc is fine.

Relevant options in xorg.conf -

################################################## #

Section "ServerFlags"
Option "AIGLX" "off"
Option "AllowEmptyInput" "off"
Option "AutoAddDevices" "off"
Option "AutoEnableDevices" "off"
EndSection

Section "Extensions"
Option "Composite" "disable"
EndSection

Section "Device"
Identifier "Videocard0"
Driver "radeon"
VendorName "ATI"
BoardName "ATI Radeon X1650"
Option "EnablePageFlip" "on"
Option "AccelMethod" "EXA"
Option "AccelDFS" "on"
Option "EXAVSync" "on"
EndSection

###############################################

Of course, all the children absolutely *need* compiz and fancy see-thru X terminals, but us old-timers are more than happy with the progress thus far. Keep up the good work.