AMD Radeon R7 360 presents poor performance after hibernation

MasterGeek.MX · 2019-09-17 01:00:49

Hi there.

I just configured my system to hibernate, following all the necesary steps in the proper wiki page. It works without issues...

...except that after reasumming the system the GPU presents degraded performance. For example Minecraft usualy runs at ~140 FPS with qute maxed out settings (16 chunks of render), but after reasumming the framerate drops to 30-40, even on 2 chunks of distance render. I'm not sure if it only affects OpenGL apps or 3D programs in general becasue I don't have anything to tesk Vulkan, but for example I use KDE Plasma with the compositing in either OpenGL 2.0 or 3.1, and plasmashell works fine before and after hibernation.

Info about my card:

$ lspci -vvv

...

25:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tobago PRO [Radeon R7 360 / R9 360 OEM] (rev 81) (prog-if 00 [VGA controller])
        Subsystem: PC Partner Limited / Sapphire Technology Tobago PRO [Radeon R7 360 / R9 360 OEM]
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+
        Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0, Cache Line Size: 64 bytes
        Interrupt: pin A routed to IRQ 41
        Region 0: Memory at e0000000 (64-bit, prefetchable) [size=256M]
        Region 2: Memory at f0000000 (64-bit, prefetchable) [size=8M]
        Region 4: I/O ports at e000 [size=256]
        Region 5: Memory at fe900000 (32-bit, non-prefetchable) [size=256K]
        Expansion ROM at 000c0000 [disabled] [size=128K]
        Capabilities: <access denied>
        Kernel driver in use: radeon
        Kernel modules: radeon, amdgpu

...

A thing to notice is that dmesg shows some errors after reasumming.

This is the output before hibernating:

$ dmesg | grep -i radeon

[    0.859439] [drm] radeon kernel modesetting enabled.
[    0.859502] radeon 0000:25:00.0: remove_conflicting_pci_framebuffers: bar 0: 0xe0000000 -> 0xefffffff
[    0.859504] radeon 0000:25:00.0: remove_conflicting_pci_framebuffers: bar 2: 0xf0000000 -> 0xf07fffff
[    0.859505] radeon 0000:25:00.0: remove_conflicting_pci_framebuffers: bar 5: 0xfe900000 -> 0xfe93ffff
[    0.859507] fb0: switching to radeondrmfb from VESA VGA
[    0.859641] radeon 0000:25:00.0: vgaarb: deactivate vga console
[    0.859888] radeon 0000:25:00.0: No more image in the PCI ROM
[    0.859971] radeon 0000:25:00.0: VRAM: 2048M 0x0000000000000000 - 0x000000007FFFFFFF (2048M used)
[    0.859972] radeon 0000:25:00.0: GTT: 2048M 0x0000000080000000 - 0x00000000FFFFFFFF
[    0.860050] [drm] radeon: 2048M of VRAM memory ready
[    0.860050] [drm] radeon: 2048M of GTT memory ready.
[    0.867450] [drm] radeon: dpm initialized
[    0.892871] radeon 0000:25:00.0: WB enabled
[    0.892879] radeon 0000:25:00.0: fence driver on ring 0 use gpu addr 0x0000000080000c00 and cpu addr 0x000000009a982ae9
[    0.892880] radeon 0000:25:00.0: fence driver on ring 1 use gpu addr 0x0000000080000c04 and cpu addr 0x0000000031e24a14
[    0.892881] radeon 0000:25:00.0: fence driver on ring 2 use gpu addr 0x0000000080000c08 and cpu addr 0x0000000049caa016
[    0.892882] radeon 0000:25:00.0: fence driver on ring 3 use gpu addr 0x0000000080000c0c and cpu addr 0x00000000b16bc5e7
[    0.892883] radeon 0000:25:00.0: fence driver on ring 4 use gpu addr 0x0000000080000c10 and cpu addr 0x00000000b696b17a
[    0.893268] radeon 0000:25:00.0: fence driver on ring 5 use gpu addr 0x0000000000078d30 and cpu addr 0x000000007c766633
[    0.893419] radeon 0000:25:00.0: fence driver on ring 6 use gpu addr 0x0000000080000c18 and cpu addr 0x00000000778f2112
[    0.893420] radeon 0000:25:00.0: fence driver on ring 7 use gpu addr 0x0000000080000c1c and cpu addr 0x000000009b8e1689
[    0.893472] radeon 0000:25:00.0: radeon: using MSI.
[    0.893505] [drm] radeon: irq initialized.
[    1.619261] [drm] Radeon Display Connectors
[    1.758456] fbcon: radeondrmfb (fb0) is primary device
[    1.816320] radeon 0000:25:00.0: fb0: radeondrmfb frame buffer device
[    1.852610] [drm] Initialized radeon 2.50.0 20080528 for 0000:25:00.0 on minor 0
[   35.163978] radeon_dp_aux_transfer_native: 116 callbacks suppressed
[  660.238784] radeon_dp_aux_transfer_native: 74 callbacks suppressed
[  713.669707] radeon_dp_aux_transfer_native: 32 callbacks suppressed

and after resumming:

$ dmesg | grep -i radeon

[    0.859439] [drm] radeon kernel modesetting enabled.
[    0.859502] radeon 0000:25:00.0: remove_conflicting_pci_framebuffers: bar 0: 0xe0000000 -> 0xefffffff
[    0.859504] radeon 0000:25:00.0: remove_conflicting_pci_framebuffers: bar 2: 0xf0000000 -> 0xf07fffff
[    0.859505] radeon 0000:25:00.0: remove_conflicting_pci_framebuffers: bar 5: 0xfe900000 -> 0xfe93ffff
[    0.859507] fb0: switching to radeondrmfb from VESA VGA
[    0.859641] radeon 0000:25:00.0: vgaarb: deactivate vga console
[    0.859888] radeon 0000:25:00.0: No more image in the PCI ROM
[    0.859971] radeon 0000:25:00.0: VRAM: 2048M 0x0000000000000000 - 0x000000007FFFFFFF (2048M used)
[    0.859972] radeon 0000:25:00.0: GTT: 2048M 0x0000000080000000 - 0x00000000FFFFFFFF
[    0.860050] [drm] radeon: 2048M of VRAM memory ready
[    0.860050] [drm] radeon: 2048M of GTT memory ready.
[    0.867450] [drm] radeon: dpm initialized
[    0.892871] radeon 0000:25:00.0: WB enabled
[    0.892879] radeon 0000:25:00.0: fence driver on ring 0 use gpu addr 0x0000000080000c00 and cpu addr 0x000000009a982ae9
[    0.892880] radeon 0000:25:00.0: fence driver on ring 1 use gpu addr 0x0000000080000c04 and cpu addr 0x0000000031e24a14
[    0.892881] radeon 0000:25:00.0: fence driver on ring 2 use gpu addr 0x0000000080000c08 and cpu addr 0x0000000049caa016
[    0.892882] radeon 0000:25:00.0: fence driver on ring 3 use gpu addr 0x0000000080000c0c and cpu addr 0x00000000b16bc5e7
[    0.892883] radeon 0000:25:00.0: fence driver on ring 4 use gpu addr 0x0000000080000c10 and cpu addr 0x00000000b696b17a
[    0.893268] radeon 0000:25:00.0: fence driver on ring 5 use gpu addr 0x0000000000078d30 and cpu addr 0x000000007c766633
[    0.893419] radeon 0000:25:00.0: fence driver on ring 6 use gpu addr 0x0000000080000c18 and cpu addr 0x00000000778f2112
[    0.893420] radeon 0000:25:00.0: fence driver on ring 7 use gpu addr 0x0000000080000c1c and cpu addr 0x000000009b8e1689
[    0.893472] radeon 0000:25:00.0: radeon: using MSI.
[    0.893505] [drm] radeon: irq initialized.
[    1.619261] [drm] Radeon Display Connectors
[    1.758456] fbcon: radeondrmfb (fb0) is primary device
[    1.816320] radeon 0000:25:00.0: fb0: radeondrmfb frame buffer device
[    1.852610] [drm] Initialized radeon 2.50.0 20080528 for 0000:25:00.0 on minor 0
[   35.163978] radeon_dp_aux_transfer_native: 116 callbacks suppressed
[  660.238784] radeon_dp_aux_transfer_native: 74 callbacks suppressed
[  713.669707] radeon_dp_aux_transfer_native: 32 callbacks suppressed
[21956.851774] radeon_dp_aux_transfer_native: 32 callbacks suppressed
[21966.978358] radeon 0000:25:00.0: GPU pci config reset
[21967.723927] [drm:ci_dpm_enable [radeon]] *ERROR* ci_start_dpm failed
[21967.723945] [drm:radeon_pm_resume [radeon]] *ERROR* radeon: dpm resume failed
[21967.731429] radeon 0000:25:00.0: WB enabled
[21967.731435] radeon 0000:25:00.0: fence driver on ring 0 use gpu addr 0x0000000080000c00 and cpu addr 0x000000009a982ae9
[21967.731436] radeon 0000:25:00.0: fence driver on ring 1 use gpu addr 0x0000000080000c04 and cpu addr 0x0000000031e24a14
[21967.731437] radeon 0000:25:00.0: fence driver on ring 2 use gpu addr 0x0000000080000c08 and cpu addr 0x0000000049caa016
[21967.731438] radeon 0000:25:00.0: fence driver on ring 3 use gpu addr 0x0000000080000c0c and cpu addr 0x00000000b16bc5e7
[21967.731439] radeon 0000:25:00.0: fence driver on ring 4 use gpu addr 0x0000000080000c10 and cpu addr 0x00000000b696b17a
[21967.731824] radeon 0000:25:00.0: fence driver on ring 5 use gpu addr 0x0000000000078d30 and cpu addr 0x000000007c766633
[21967.731977] radeon 0000:25:00.0: fence driver on ring 6 use gpu addr 0x0000000080000c18 and cpu addr 0x00000000778f2112
[21967.731978] radeon 0000:25:00.0: fence driver on ring 7 use gpu addr 0x0000000080000c1c and cpu addr 0x000000009b8e1689
[21968.434896] [drm:radeon_pm_resume [radeon]] *ERROR* radeon: dpm resume failed
[21971.876645] radeon_dp_aux_transfer_native: 32 callbacks suppressed

Zod · 2019-09-17 01:38:17

https://bbs.archlinux.org/viewtopic.php?id=235709

MasterGeek.MX · 2019-09-17 05:22:52

Zod wrote:

https://bbs.archlinux.org/viewtopic.php?id=235709

Didn't worked. It even worsened the situation because now the GPU works bad in a fresh boot.

V1del · 2019-09-17 07:10:01

That's not an error message, nor log output which of the things outlined in that thread did you actually do? FWIW since this is a new enough GPU you could also opt to switch to using AMDGPU

MasterGeek.MX · 2019-09-17 23:07:29

I set the kernel parameters cited in the post.

Also I'm already using AMDGPU.

MasterGeek.MX · 2019-09-17 23:12:27

My bad. I checked two times and im in fact using the radeon module. Gonna anable amdgpu and will come back with results

MasterGeek.MX · 2019-09-18 23:40:08

I'm back. I listed `amdgpu` as the only module in the mkinitcpio config file, but the system keeps using the `radeon` module. I then blacklisted it on /etc/modprobe.d, but that left the system broke. I already fixed it by removing the blacklist file, but I'm right back at square one.

If it works, I installed radeon-profile, and the thing is that after resuming the program does not show memory anc clock speeds anymore.

loqs · 2019-09-18 23:53:51

What is you add both modules to mkinitcpio.conf ensuring amdgpu is first

MODULES=(amdgpu radeon)

ensure modconf is in the HOOKS array and have generated both /etc/modprobe.d/amdgpu.conf and /etc/modprobe.d/radeon.conf ?

MasterGeek.MX · 2019-09-19 23:22:01

I tried both (radeon first on the list and also amdgpu firt on the list), only radeon and only amdgpu. Still nothing. lspci and radeon profile keep reporting me that radeon is in use.

loqs · 2019-09-19 23:26:31

Did you rebuild the initrd after making the changes?
Edit:
If yes please also post the journal from the boot where the amdgpu module was listed first (without piping the output through grep).

Last edited by loqs (2019-09-19 23:51:46)

MasterGeek.MX · 2019-09-20 02:04:26

Yes, I ran mkintcpio -p linux after setting up the hooks at /etc/mkinitcpio.conf

And here is the journalctl log, unfiltered as asked

V1del · 2019-09-20 08:37:12

Post all the contents of all the files in /etc/modprobe.d

MasterGeek.MX · 2019-09-21 01:32:06

/etc/modprobe.d is empty. I once tried to force amdgpu by creating a blacklist for radeon, but obviously it disabled my GPU entierly so I had to remove it using a livecd.

V1del · 2019-09-21 03:28:58

So you did not actually read this link that I posted? https://wiki.archlinux.org/index.php/AM … K)_support ? follow that to properly switch to AMDGPU, no one said anything about a blacklist.

MasterGeek.MX · 2019-09-21 06:49:58

I read it 3 times. It says that the linux package automatically enables amdgpu for Southern Islands (SI) and Sea Islands (CIK), but I have a Tobago card. The other thing is to enable some flags at custom building the kernel, which is no the case becasue I have the repos kernel.

and yes, I have amdgpu listed as the first module on the hooks list.

V1del · 2019-09-21 10:45:47

I don't really care* (or know wtf they are thinking) to go into the specifics of AMDGPUs naming scheme, every evidence points that your card needs these options to properly switch to utilising amdgpu, so you should apply them.

As far as I'm concerned the equation goes if the kernel uses radeon but offers amdgpu as a possibility you need to use these.

* I did end up caring, tobago is a carribean island, so in terms of amdgpus naming scheme likely needing the CIK switch.

Last edited by V1del (2019-09-21 10:46:04)

MasterGeek.MX · 2019-09-21 16:57:11

My bad for being that stubborn. Apologies for the anger caused by being a cloth-eared.

Indeed I had to set the cik and si support, both in /etc/modprobe.d and in the kernel parameters at /etc/default/grub. (setted both SI and CIK support on in amdgpu and off in radeon) Now it uses amdgpu!...

but we are still at square one. I still see low performance after hibernating.

Dunno if this is releated, but now dmesg is full of a spam of

[...] amdgpu: [powerplay]

after a reboot.

MasterGeek.MX · 2019-10-02 04:01:52

So... Any clues?

*Sorry for begging*

Arch Linux

#1 2019-09-17 01:00:49

AMD Radeon R7 360 presents poor performance after hibernation

#2 2019-09-17 01:38:17

Re: AMD Radeon R7 360 presents poor performance after hibernation

#3 2019-09-17 05:22:52

Re: AMD Radeon R7 360 presents poor performance after hibernation

#4 2019-09-17 07:10:01

Re: AMD Radeon R7 360 presents poor performance after hibernation

#5 2019-09-17 23:07:29

Re: AMD Radeon R7 360 presents poor performance after hibernation

#6 2019-09-17 23:12:27

Re: AMD Radeon R7 360 presents poor performance after hibernation

#7 2019-09-18 23:40:08

Re: AMD Radeon R7 360 presents poor performance after hibernation

#8 2019-09-18 23:53:51

Re: AMD Radeon R7 360 presents poor performance after hibernation

#9 2019-09-19 23:22:01

Re: AMD Radeon R7 360 presents poor performance after hibernation

#10 2019-09-19 23:26:31

Re: AMD Radeon R7 360 presents poor performance after hibernation

#11 2019-09-20 02:04:26

Re: AMD Radeon R7 360 presents poor performance after hibernation

#12 2019-09-20 08:37:12

Re: AMD Radeon R7 360 presents poor performance after hibernation

#13 2019-09-21 01:32:06

Re: AMD Radeon R7 360 presents poor performance after hibernation

#14 2019-09-21 03:28:58

Re: AMD Radeon R7 360 presents poor performance after hibernation

#15 2019-09-21 06:49:58

Re: AMD Radeon R7 360 presents poor performance after hibernation

#16 2019-09-21 10:45:47

Re: AMD Radeon R7 360 presents poor performance after hibernation

#17 2019-09-21 16:57:11

Re: AMD Radeon R7 360 presents poor performance after hibernation

#18 2019-10-02 04:01:52

Re: AMD Radeon R7 360 presents poor performance after hibernation

Board footer