Discussion: the ati/radeon open source drivers & radeon repository

Radioactiveman · 2011-06-06 19:29:34

Cdh wrote:

venky80 wrote:
does anyone know if r600g will provide hardware acceleration for blue ray rips in MKV container ? I have a mobility radeon HD 3670
the HD 3670 (as every pre-4xxx card) lacks the UVD, the dedicated video decoding hardware thingy.

That's wrong, every card >= HD2000 (except HD2900) has UVD support.

Cdh · 2011-06-06 20:17:56

* UVD2

And does xvba->vaapi work with UVD?

venky80 · 2011-06-06 21:56:38

Cdh wrote:

venky80 wrote:
does anyone know if r600g will provide hardware acceleration for blue ray rips in MKV container ? I have a mobility radeon HD 3670
Probably, at some time,
Maybe after this GSOC:
http://xorg.freedesktop.org/wiki/SummerOfCodeIdeas
"Gallium H.264 decoding"
They want it to do on the shaders - good for you, because the HD 3670 (as every pre-4xxx card) lacks the UVD, the dedicated video decoding hardware thingy.
But it will take some time and we can be lucky if it indeed is ready after this summer.

Was this accepted , also I heard VADPU for r600g in phoronix a while back, is that merged?

dr-kart · 2011-06-08 15:07:29

dejavu wrote:

...small question:
Does it still (or now) make sense to use this repo for an hd 3470?...

I have HD3650 and I got same question. Would I get any benefits of using this repo with latest Gallium and drm-radeon-testing kernel? I do need maximum video acceleration by GPU.

venky80 · 2011-06-08 16:42:54

venky80 wrote:

Cdh wrote:
venky80 wrote:
does anyone know if r600g will provide hardware acceleration for blue ray rips in MKV container ? I have a mobility radeon HD 3670
Probably, at some time,
Maybe after this GSOC:
http://xorg.freedesktop.org/wiki/SummerOfCodeIdeas
"Gallium H.264 decoding"
They want it to do on the shaders - good for you, because the HD 3670 (as every pre-4xxx card) lacks the UVD, the dedicated video decoding hardware thingy.
But it will take some time and we can be lucky if it indeed is ready after this summer.
Was this accepted , also I heard VADPU for r600g in phoronix a while back, is that merged?

oh just read it in Phoronix today. the VDPAU is still developed in a branch...so nothing there in terms of video acceleration for this year i would imagine.

ChemBro · 2011-06-08 20:27:00

dr-kart wrote:

dejavu wrote:
...small question:
Does it still (or now) make sense to use this repo for an hd 3470?...
I have HD3650 and I got same question. Would I get any benefits of using this repo with latest Gallium and drm-radeon-testing kernel? I do need maximum video acceleration by GPU.

The gallium driver for this series (HD 2xxx -6xxx) is not so fast, as you might think. The series below does benefit from the gallium driver.

Perry3D · 2011-06-08 21:49:06

ChemBro wrote:

The gallium driver for this series (HD 2xxx -6xxx) is not so fast, as you might think. The series below does benefit from the gallium driver.

Its also a good idea to use gallium for HD 2xxx+ cards. It is faster, you should read some phoronix benchmarks: http://www.phoronix.com/scan.php?page=a … sa_mai2011

Vavooon · 2011-06-09 05:37:29

Hi to all. I have Acer AO522 with HD6250M. I has been changed power_profile and power_method, but VC is very hot (70 C).
Do you can help me with this problem? I can give an remote access to my netbook.
Thank you.

ChemBro · 2011-06-09 10:22:51

Perry3D wrote:

ChemBro wrote:
The gallium driver for this series (HD 2xxx -6xxx) is not so fast, as you might think. The series below does benefit from the gallium driver.
Its also a good idea to use gallium for HD 2xxx+ cards. It is faster, you should read some phoronix benchmarks: http://www.phoronix.com/scan.php?page=a … sa_mai2011

No, I shouldn't read phoronix benchmarks. Phoronix has sometimes completely other results than I do. And last time I checked for example SuperTuxKart and compared the "normal" driver with the "gallium" driver, there wasn't a noticable difference (both had framerate issue and the ingame "framerate counter" didn't show a noticable increase of framerates).

But I believe, there will be a difference in the future with the gallium driver.

handy · 2011-06-10 02:10:39

My HD2600 Pro, is about 250 fps faster on Gallium over classic Mesa.

The last 2 Gallium upgrades from Perry's repo have given me 3D speed increases, before that there hadn't been any for over roughly 2 months.

sumski · 2011-06-10 13:19:06

ijanos wrote:

That sounds like a bug, report it at bug.freedesktop.org

I reported this at kernel bugzilla -
With Debian/2.6.39 kernel i get:

radeon 0000:01:00.0: GPU lockup CP stall for more than 10004msec
Jun  9 16:30:06 sumarija kernel: [ 2818.056175] ------------[ cut here ]------------
Jun  9 16:30:06 sumarija kernel: [ 2818.056213] WARNING: at /build/buildd-linux-2.6_2.6.39-1-i386-RRBuT6/linux-2.6-2.6.39/debian/build/source_i386_none/drivers/gpu/drm/radeon/radeon_fence.c:246 radeon_fence_wait+0x257/0x2e8 [radeon]()
Jun  9 16:30:06 sumarija kernel: [ 2818.056220] Hardware name: Extensa 5630                   
Jun  9 16:30:06 sumarija kernel: [ 2818.056224] GPU lockup (waiting for 0x000BF53A last fence id 0x000BF535)
Jun  9 16:30:06 sumarija kernel: [ 2818.056228] Modules linked in: cryptd aes_i586 aes_generic coretemp loop snd_hda_codec_hdmi sg sr_mod arc4 cdrom ata_generic ecb ata_piix joydev pcmcia libata snd_hda_codec_realtek radeon iwlagn sdhci_pci ttm sdhci yenta_socket mac80211 snd_hda_intel drm_kms_helper pcmcia_rsrc mmc_core snd_hda_codec drm cfg80211 snd_hwdep snd_pcm snd_timer ac i2c_i801 acer_wmi snd sparse_keymap soundcore uvcvideo videodev media rfkill psmouse evdev serio_raw pcspkr pcmcia_core processor battery i2c_algo_bit i2c_core tg3 power_supply video snd_page_alloc button wmi uhci_hcd libphy ext4 mbcache jbd2 crc16 sd_mod crc_t10dif usb_storage uas scsi_mod ehci_hcd usbcore thermal thermal_sys
Jun  9 16:30:06 sumarija kernel: [ 2818.056340] Pid: 1639, comm: kwin Not tainted 2.6.39-1-686-pae #1
Jun  9 16:30:06 sumarija kernel: [ 2818.056344] Call Trace:
Jun  9 16:30:06 sumarija kernel: [ 2818.056356]  [<c1036b3d>] ? warn_slowpath_common+0x6a/0x7b
Jun  9 16:30:06 sumarija kernel: [ 2818.056385]  [<f86af8f7>] ? radeon_fence_wait+0x257/0x2e8 [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056393]  [<c1036bb4>] ? warn_slowpath_fmt+0x28/0x2c
Jun  9 16:30:06 sumarija kernel: [ 2818.056420]  [<f86af8f7>] ? radeon_fence_wait+0x257/0x2e8 [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056430]  [<c104c83e>] ? wake_up_bit+0x56/0x56
Jun  9 16:30:06 sumarija kernel: [ 2818.056441]  [<f849f523>] ? ttm_bo_wait+0xad/0x135 [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056452]  [<f84a190a>] ? ttm_bo_move_accel_cleanup+0x7c/0x227 [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056479]  [<f86b0014>] ? radeon_move_blit+0x14b/0x16b [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056507]  [<f86b02ca>] ? radeon_bo_move+0x1a1/0x1fe [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056519]  [<f849fec9>] ? ttm_bo_reserve_locked+0xbf/0xe6 [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056546]  [<f86b0129>] ? radeon_move_vram_ram+0xf5/0xf5 [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056556]  [<f849f90f>] ? ttm_bo_handle_move_mem+0x1da/0x290 [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056567]  [<f84a0f0d>] ? ttm_bo_move_buffer+0xc0/0xe5 [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056577]  [<f849fec9>] ? ttm_bo_reserve_locked+0xbf/0xe6 [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056587]  [<f84a0fdb>] ? ttm_bo_validate+0xa9/0xec [ttm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056615]  [<f86b0c6c>] ? radeon_bo_list_validate+0x68/0xbb [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056646]  [<f86c44c4>] ? radeon_cs_ioctl+0x9c/0x168 [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056663]  [<f84f8078>] ? drm_ioctl+0x224/0x2d5 [drm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056694]  [<f86c4428>] ? radeon_cs_parser_init+0x21d/0x21d [radeon]
Jun  9 16:30:06 sumarija kernel: [ 2818.056703]  [<c10d61f2>] ? user_path_at+0x42/0x61
Jun  9 16:30:06 sumarija kernel: [ 2818.056711]  [<c115b3fa>] ? copy_to_user+0x2c/0x100
Jun  9 16:30:06 sumarija kernel: [ 2818.056719]  [<c10cf310>] ? cp_new_stat64+0xee/0x100
Jun  9 16:30:06 sumarija kernel: [ 2818.056735]  [<f84f7e54>] ? drm_version+0x84/0x84 [drm]
Jun  9 16:30:06 sumarija kernel: [ 2818.056742]  [<c10d7733>] ? do_vfs_ioctl+0x457/0x4a2
Jun  9 16:30:06 sumarija kernel: [ 2818.056749]  [<c10cf880>] ? sys_stat64+0x1e/0x23
Jun  9 16:30:06 sumarija kernel: [ 2818.056756]  [<c10d77c2>] ? sys_ioctl+0x44/0x64
Jun  9 16:30:06 sumarija kernel: [ 2818.056764]  [<c12b6fdf>] ? sysenter_do_call+0x12/0x28
Jun  9 16:30:06 sumarija kernel: [ 2818.056771]  [<c12b0000>] ? panic+0xb5/0x141
Jun  9 16:30:06 sumarija kernel: [ 2818.056776] ---[ end trace 1b863a9f05671293 ]---

With Fedora/2.6.38.7:

radeon 0000:01:00.0: GPU lockup CP stall for more than 10004msec
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700217] ------------[ cut here ]------------
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700292] WARNING: at drivers/gpu/drm/radeon/radeon_fence.c:248 radeon_fence_wait+0x2a8/0x32b [radeon]()
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700300] Hardware name: Extensa 5630                   
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700306] GPU lockup (waiting for 0x000017BD last fence id 0x000017B7)
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700312] Modules linked in: fuse cpufreq_ondemand 8021q acpi_cpufreq freq_table mperf garp stp llc ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_filter ip6_tables snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_hwdep snd_seq arc4 snd_seq_device snd_pcm snd_timer acer_wmi iwlagn sparse_keymap snd uvcvideo microcode i2c_i801 joydev iTCO_wdt soundcore iwlcore serio_raw mac80211 videodev v4l2_compat_ioctl32 iTCO_vendor_support cfg80211 wmi tg3 rfkill snd_page_alloc ipv6 sdhci_pci sdhci mmc_core uas usb_storage yenta_socket video radeon ttm drm_kms_helper drm i2c_algo_bit i2c_core [last unloaded: scsi_wait_scan]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700437] Pid: 1091, comm: Xorg Not tainted 2.6.38.7-30.fc15.x86_64 #1
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700443] Call Trace:
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700461]  [<ffffffff8105511a>] warn_slowpath_common+0x83/0x9b
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700471]  [<ffffffff810551d5>] warn_slowpath_fmt+0x46/0x48
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700533]  [<ffffffffa00c4e74>] ? r600_gpu_is_lockup+0xbd/0xc6 [radeon]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700582]  [<ffffffffa009fc74>] radeon_fence_wait+0x2a8/0x32b [radeon]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700593]  [<ffffffff8106f26e>] ? autoremove_wake_function+0x0/0x3d
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700643]  [<ffffffffa00a0239>] radeon_sync_obj_wait+0x11/0x13 [radeon]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700662]  [<ffffffffa00655da>] ttm_bo_wait+0xbf/0x17a [ttm]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700680]  [<ffffffffa0066002>] ? ttm_bo_list_ref_sub+0x29/0x2b [ttm]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700736]  [<ffffffffa00afe3b>] radeon_bo_wait+0x7b/0x9f [radeon]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700794]  [<ffffffffa00b03ad>] radeon_gem_wait_idle_ioctl+0x3d/0x74 [radeon]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700822]  [<ffffffffa0019861>] drm_ioctl+0x29e/0x37b [drm]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700833]  [<ffffffff811eb273>] ? inode_has_perm+0x76/0x8c
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700889]  [<ffffffffa00b0370>] ? radeon_gem_wait_idle_ioctl+0x0/0x74 [radeon]
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700901]  [<ffffffff81474228>] ? _cond_resched+0xe/0x22
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700911]  [<ffffffff811eb32d>] ? file_has_perm+0xa4/0xc6
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700921]  [<ffffffff8112f51c>] do_vfs_ioctl+0x47e/0x4bf
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700931]  [<ffffffff8109fcb9>] ? audit_syscall_exit+0x12d/0x148
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700940]  [<ffffffff8112f5b3>] sys_ioctl+0x56/0x7b
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700949]  [<ffffffff81009e75>] ? int_check_syscall_exit_work+0x34/0x3d
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700959]  [<ffffffff81009bc2>] system_call_fastpath+0x16/0x1b
Jun 10 14:42:40 sumski-Extensa-5630 kernel: [  332.700966] ---[ end trace 4cbf591f7911c9de ]---

I'm using Fedora and Debian for this, because its easier to provide debug info

ChemBro · 2011-06-10 17:54:56

handy wrote:

My HD2600 Pro, is about 250 fps faster on Gallium over classic Mesa.

With what? Warsow?

handy · 2011-06-10 23:46:54

ChemBro wrote:

handy wrote:
My HD2600 Pro, is about 250 fps faster on Gallium over classic Mesa.
With what? Warsow?

No, just the gears. It gives me an indication. I couldn't be bothered using the pts each time I upgrade.

Last edited by handy (2011-06-11 21:01:47)

Perry3D · 2011-06-11 09:41:57

Some new gallium benchmarks: http://www.phoronix.com/vr.php?view=16110

ChemBro · 2011-06-11 19:59:50

handy wrote:

No, just the the gears. It gives me an indication. I couldn't be bothered using the pts each time I upgrade.

Don't use glxgears for that. I had, for example, a big decrease in fps with glxgears, while the real performance of the fglrx-driver was better than before. It's better to test it with a real benchmark, like Unigine Engine Heaven or so (if the gallium driver can run this benchmark).

bwat47 · 2011-06-11 20:11:32

Cdh wrote:

* UVD2
And does xvba->vaapi work with UVD?

afiak only for uvd2 right now.

Last edited by bwat47 (2011-06-11 20:12:53)

handy · 2011-06-11 21:01:14

ChemBro wrote:

handy wrote:
No, just the gears. It gives me an indication. I couldn't be bothered using the pts each time I upgrade.
Don't use glxgears for that. I had, for example, a big decrease in fps with glxgears, while the real performance of the fglrx-driver was better than before. It's better to test it with a real benchmark, like Unigine Engine Heaven or so (if the gallium driver can run this benchmark).

I know its not a true benchmark, & I'm sure that there are exceptions either way. I'm happy with the way the open stack are working so I don't personally really care that much.

There are some games that I would like to be able to play using the open driver stack, but I'm in no hurry. I've got another machine running Arch that has an nVidia XFX 7950GT /512MB card in it, that I use to do things that I can't on this machine, & a PS3 to do things that I can't do on that.

sumski · 2011-06-12 04:31:43

sumski wrote:

Has anybody tried .39 , or 3.0 kernel? Did powersaving improved ? Main reason why i use catalyst is because i can't get no satisfaction with OSS drivers, regarding powersaving. (Tried every method!) It's a mobility 3650 HD, if anyone has positive experiences , please share
With catalyst my GPU clock is at 110MHz 90% of the time ,memory clock is at 400MHz , but with radeon they are both at the highest values all the time. I know it's a known issue , but i must ask.

I got lockups , because kernel was changing memory frequency too low - to 300MHz (with catalyst , lowest frequency is 400MHz).
Changing one line in:

drivers/gpu/drm/radeon/radeon_pm.c

fixes the issue - memory clock stays untouched, but i can change GPU clock successfully with quieter fan and lower temperatures as results

sovok_kpss · 2011-06-13 09:50:02

@Perry3D: Could you please update the git repo (git://gitorious.org/radeon/radeon.git) ?

splashy · 2011-06-13 17:57:47

Do the amnesia's light artefacts disappear with the latest update of the repository and the last update of libdtxn?
Do s3tc have a better support?

Perry3D · 2011-06-13 18:45:09

sovok_kpss wrote:

@Perry3D: Could you please update the git repo (git://gitorious.org/radeon/radeon.git) ?

Done!

Perry3D · 2011-06-14 17:55:12

After my HD4850 got broken at the weekend (i have no idea why), i ordered a new MSI 6870 hawk. Is someone having experience with this card/chip and the open source drivers? As far as i know it should as good as my old card.

Last edited by Perry3D (2011-06-14 17:56:53)

Radioactiveman · 2011-06-15 11:40:38

If it's already dead you could try to bake it in the oven -> click (I know you are German )

Perry3D · 2011-06-15 13:06:55

Radioactiveman wrote:

If it's already dead you could try to bake it in the oven -> click (I know you are German )

Lol, that's one of the weirdest things i've ever read about hardware. Maybe i'll give it a try .
Does someone know why 100°C can repair a gpu-chip?

@all: the latest update should give a nice performance increase. Color tiling is now enabled by default.

Varg · 2011-06-15 13:24:00

Where can I find the patches and options that where applied to the radeon testing kernel ? I'm trying to strip down my kernel but it would be nice to squeeze some more performance out of my card. 2D games are still lagging way to much and performance with the testing kernel was better but it's useless for me without the brcmsmac module.

Arch Linux

#2051 2011-06-06 19:29:34

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2052 2011-06-06 20:17:56

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2053 2011-06-06 21:56:38

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2054 2011-06-08 15:07:29

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2055 2011-06-08 16:42:54

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2056 2011-06-08 20:27:00

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2057 2011-06-08 21:49:06

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2058 2011-06-09 05:37:29

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2059 2011-06-09 10:22:51

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2060 2011-06-10 02:10:39

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2061 2011-06-10 13:19:06

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2062 2011-06-10 17:54:56

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2063 2011-06-10 23:46:54

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2064 2011-06-11 09:41:57

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2065 2011-06-11 19:59:50

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2066 2011-06-11 20:11:32

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2067 2011-06-11 21:01:14

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2068 2011-06-12 04:31:43

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2069 2011-06-13 09:50:02

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2070 2011-06-13 17:57:47

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2071 2011-06-13 18:45:09

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2072 2011-06-14 17:55:12

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2073 2011-06-15 11:40:38

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2074 2011-06-15 13:06:55

Re: Discussion: the ati/radeon open source drivers & radeon repository

#2075 2011-06-15 13:24:00

Re: Discussion: the ati/radeon open source drivers & radeon repository

Board footer