AMD Has Massive Radeon Patch Set - Power Management!

[d2kx]

OK.

Notebook: AMD APU E-350 (Radeon HD 6310): DPM works, but UVD fails
Desktop: AMD Radeon HD 6870: DPM works, but UVD fails aswell. Also, there seems to be a short GPU lockup during boot, then resumes normally

Also, DPM doesn't just work, it's works excellently. My desktop is usually loud as shit with the opensource drivers, so even if it was just slightly worse than Catalyst, I would notice instantly. But it's completely quiet.

So yeah, UVD fails on both but I am sure that'll get fixed soon enough and it's not *that* important to me. The opensource drivers are amazing compared to fglrx.

[Vincenzov]

i remove from rc.local
cho dynpm > /sys/class/drm/card0/device/power_method
# echo profile > /sys/class/drm/card0/device/power_method
# echo mid > /sys/class/drm/card0/device/power_profile
this command????


vincenzo@AthlonII-X4-640:/usr/src/linux$ git checkout -b drm-next-3.11-wip-5 remotes/origin/drm-next-3.11-wip-5
Checking out files: 100% (36879/36879), done.
Branch drm-next-3.11-wip-5 set up to track remote branch drm-next-3.11-wip-5 from origin.

[d2kx]

OK.

Notebook: AMD APU E-350 (Radeon HD 6310): DPM works, but UVD fails
Desktop: AMD Radeon HD 6870: DPM works, but UVD fails aswell. Also, there seems to be a short GPU lockup during boot, then resumes normally

Also, DPM doesn't just work, it's works excellently. My desktop is usually loud as shit with the opensource drivers, so even if it was just slightly worse than Catalyst, I would notice instantly. But it's completely quiet.

So yeah, UVD fails on both but I am sure that'll get fixed soon enough and it's not *that* important to me. The opensource drivers are amazing compared to fglrx.

OK, at least on my desktop (Radeon HD 6870), it does not increase the clocks in demanding 3D games, resulting in a bad performance in them. I read there are known issues with the 6xxx, is this one of them?

[GDJacobs]

i remove from rc.local
cho dynpm > /sys/class/drm/card0/device/power_method
# echo profile > /sys/class/drm/card0/device/power_method
# echo mid > /sys/class/drm/card0/device/power_profile
this command????

No, this isn't necessary.

vincenzo@AthlonII-X4-640:/usr/src/linux$ git checkout -b drm-next-3.11-wip-5 remotes/origin/drm-next-3.11-wip-5
Checking out files: 100% (36879/36879), done.
Branch drm-next-3.11-wip-5 set up to track remote branch drm-next-3.11-wip-5 from origin.

Right, now you have to build the kernel as you would from any other source tree.

I forgot to mention that you have to update the firmware in /lib/firmware/radeon as well as add the argument dpm=1 for the radeon module (or radeon.dpm=1 for grub).

[Vincenzov]

OK i have try.
With radeon 6450 bad screen , not start kde.
With radeon 5450 it's ok

vincenzo@AthlonII-X4-640:~$ sudo cat /sys/kernel/debug/dri/0/radeon_pm_info
[sudo] password for vincenzo:
cat: /sys/kernel/debug/dri/0/radeon_pm_info: File o directory non esistente
vincenzo@AthlonII-X4-640:~$ sudo cat /sys/class/drm/card0/device/power_method
dpm
vincenzo@AthlonII-X4-640:~$

But with vdpau crush a pc.

[Vim_User]

APUs don't need smc ucode. Only 7xx+ dGPUs need it.

Does that include the RS780? I can't find firmware for that.

[GDJacobs]

Does that include the RS780? I can't find firmware for that.

I just hoovered the contents of Alex's ucode directory. RS880 uses the same firmware as RS780.

HD5450 (CEDAR) runs with dpm, but uvd fails as well.

[dimsam]

OK.

Notebook: AMD APU E-350 (Radeon HD 6310): DPM works, but UVD fails
Desktop: AMD Radeon HD 6870: DPM works, but UVD fails aswell. Also, there seems to be a short GPU lockup during boot, then resumes normally

Also, DPM doesn't just work, it's works excellently. My desktop is usually loud as shit with the opensource drivers, so even if it was just slightly worse than Catalyst, I would notice instantly. But it's completely quiet.

So yeah, UVD fails on both but I am sure that'll get fixed soon enough and it's not *that* important to me. The opensource drivers are amazing compared to fglrx.

for the 6850, uvd works so propably u need to load the firmware.

dmesg | grep radeon
[ 7.267572] [drm] radeon kernel modesetting enabled.
[ 7.267724] bus: 'pci': add driver radeon
[ 7.267780] bus: 'pci': driver_probe_device: matched device 0000:03:00.0 with driver radeon
[ 7.267782] bus: 'pci': really_probe: probing driver radeon with device 0000:03:00.0
[ 7.270159] radeon 0000:03:00.0: VRAM: 1024M 0x0000000000000000 - 0x000000003FFFFFFF (1024M used)
[ 7.270385] radeon 0000:03:00.0: GTT: 512M 0x0000000040000000 - 0x000000005FFFFFFF
[ 7.273345] [drm] radeon: 1024M of VRAM memory ready
[ 7.273478] [drm] radeon: 512M of GTT memory ready.
[ 7.273630] Registering platform device 'radeon_uvd.0'. Parent at platform
[ 7.273633] device: 'radeon_uvd.0': device_add
[ 7.273643] bus: 'platform': add device radeon_uvd.0
[ 7.273654] PM: Adding info for platform:radeon_uvd.0
[ 7.273684] __allocate_fw_buf: fw-radeon/SUMO_uvd.bin buf=f034ccc0
[ 7.355901] platform radeon_uvd.0: firmware: direct-loading firmware radeon/SUMO_uvd.bin
[ 7.355910] fw_set_page_data: fw-radeon/SUMO_uvd.bin buf=f034ccc0 data=f8c9c000 size=205080
[ 7.355930] bus: 'platform': remove device radeon_uvd.0
[ 7.355932] PM: Removing info for platform:radeon_uvd.0
[ 7.355953] fw_name_devm_release: fw_name-radeon/SUMO_uvd.bin devm-f034cf94 released
[ 7.355972] radeon 0000:03:00.0: f01a6800 unpin not necessary
[ 8.026534] radeon 0000:03:00.0: fence driver on ring 5 use gpu addr 0x0000000000072118 and cpu addr 0xf933a118
[ 8.027745] [drm] enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0
[ 8.028039] Registering platform device 'radeon_cp.0'. Parent at platform
[ 8.028041] device: 'radeon_cp.0': device_add
[ 8.028050] bus: 'platform': add device radeon_cp.0
[ 8.028060] PM: Adding info for platform:radeon_cp.0
[ 8.028210] __allocate_fw_buf: fw-radeon/BARTS_pfp.bin buf=f0219660
[ 8.161396] platform radeon_cp.0: firmware: direct-loading firmware radeon/BARTS_pfp.bin
[ 8.161403] fw_set_page_data: fw-radeon/BARTS_pfp.bin buf=f0219660 data=f96be000 size=4480
[ 8.161406] __allocate_fw_buf: fw-radeon/BARTS_me.bin buf=f034cc00
[ 8.238772] platform radeon_cp.0: firmware: direct-loading firmware radeon/BARTS_me.bin
[ 8.238777] fw_set_page_data: fw-radeon/BARTS_me.bin buf=f034cc00 data=f96c4000 size=5504
[ 8.238780] __allocate_fw_buf: fw-radeon/BTC_rlc.bin buf=f028ca40
[ 8.253776] platform radeon_cp.0: firmware: direct-loading firmware radeon/BTC_rlc.bin
[ 8.253780] fw_set_page_data: fw-radeon/BTC_rlc.bin buf=f028ca40 data=f96c7000 size=3072
[ 8.253783] __allocate_fw_buf: fw-radeon/BARTS_mc.bin buf=f028cb40
[ 8.269503] platform radeon_cp.0: firmware: direct-loading firmware radeon/BARTS_mc.bin
[ 8.269509] fw_set_page_data: fw-radeon/BARTS_mc.bin buf=f028cb40 data=f96c9000 size=24096
[ 8.269513] __allocate_fw_buf: fw-radeon/BARTS_smc.bin buf=f07a06c0
[ 8.378062] platform radeon_cp.0: firmware: direct-loading firmware radeon/BARTS_smc.bin
[ 8.378066] fw_set_page_data: fw-radeon/BARTS_smc.bin buf=f07a06c0 data=f96d0000 size=24840
[ 8.378086] bus: 'platform': remove device radeon_cp.0
[ 8.378088] PM: Removing info for platform:radeon_cp.0
[ 8.378107] fw_name_devm_release: fw_name-radeon/BARTS_smc.bin devm-f028cd94 released
[ 8.378109] fw_name_devm_release: fw_name-radeon/BARTS_mc.bin devm-ef3513d4 released
[ 8.378111] fw_name_devm_release: fw_name-radeon/BTC_rlc.bin devm-f028cb14 released
[ 8.378112] fw_name_devm_release: fw_name-radeon/BARTS_me.bin devm-f028ca14 released
[ 8.378114] fw_name_devm_release: fw_name-radeon/BARTS_pfp.bin devm-f034cbd4 released
[ 8.390240] radeon 0000:03:00.0: WB enabled
[ 8.390363] radeon 0000:03:00.0: fence driver on ring 0 use gpu addr 0x0000000040000c00 and cpu addr 0xffc5bc00
[ 8.390566] radeon 0000:03:00.0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0xffc5bc0c
[ 8.440100] radeon 0000:03:00.0: fence driver on ring 5 use gpu addr 0x0000000000174118 and cpu addr 0xf9b32118
[ 8.440588] radeon 0000:03:00.0: irq 43 for MSI/MSI-X
[ 8.440602] radeon 0000:03:00.0: radeon: using MSI.
[ 8.440763] [drm] radeon: irq initialized.
[ 8.816737] [drm] radeon: dpm initialized
[ 8.910326] fbcon: radeondrmfb (fb0) is primary device
[ 8.945093] radeon 0000:03:00.0: fb0: radeondrmfb frame buffer device
[ 8.945149] radeon 0000:03:00.0: registered panic notifier
[ 8.945200] [drm] Initialized radeon 2.34.0 20080528 for 0000:03:00.0 on minor 0
[ 8.945264] driver: '0000:03:00.0': driver_bound: bound to device 'radeon'
[ 8.945272] bus: 'pci': really_probe: bound device 0000:03:00.0 to driver radeon


update

with mplayer it crashed...

[Vim_User]

I just hoovered the contents of Alex's ucode directory. RS880 uses the same firmware as RS780.

Can't see firmware for that, either.

[Bushido_root]

It doesn't look like it worked for me:

x@Akaigo-Arc:~$ dmesg | grep dpm
[ 0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-3.10.0-996-generic root=UUID=c569a25a-4c5c-4731-8836-2bdfd7ce3d7e ro quiet splash radeon.audio=1 radeon.dpm=1 vt.handoff=7
[ 0.000000] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-3.10.0-996-generic root=UUID=c569a25a-4c5c-4731-8836-2bdfd7ce3d7e ro quiet splash radeon.audio=1 radeon.dpm=1 vt.handoff=7
[ 29.492599] [drm:cypress_dpm_enable] *ERROR* rv770_upload_firmware failed
[ 29.492608] [drm:radeon_pm_init_dpm] *ERROR* radeon: dpm initialization failed

I used this kernel. I have a HD5400 Radeon card.

This is what the GRUB line looks like:
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash radeon.audio=1 radeon.dpm=1"