You can sign up to get a daily email of our articles, see the Mailing List page.
[Rant]: RX 5700... a frustrating experience
Page: «18/21»
  Go to:
tuubi Mar 28, 2020
Quoting: TuxeeHow many displays do you use? Seems as if some issues stem from multi-monitor setups.
Only one. A TV screen to be exact. This is a sofa gaming setup.

You might be right that this only happens if you use multiple display outputs (or a specific single output). I'm using the HDMI only.
dvd Apr 1, 2020
Quoting: Tuxee
Quoting: dvdWhat kind of motherboard are you using?
ASRock X570M Pro4.

I have an ASRock one too. Have you read the smallprint before downloading and upgrading to the latest BIOS? For my model they don't recommend upgrading the bios past a certain version for some AMD CPUs.
Tuxee Apr 3, 2020
Well, upgrading made sense: With the most recent upgrade I can finally run my RAM at 3600MHz. Before that I could only go up to 3200. And I can't detect any problems with the CPU.
m2mg2 Apr 3, 2020
I've got 3 different Radeon cards that work with amdgpu, not one of them could generate a functional mutli monitor setup without a massive hassle. None of them work ootb with a vga monitor I've got. I just got a 5700 and it took me hours and hours to get a working multi monitor setup. I had to hook my Acer monitor up to my NVidia system and export an EDID file and then figure out the horribly documented kernel switches needed to get the EDID file read so the monitor would work at over 800x600. I also had to include the edid file in the initrd (not 100% sure this was necessary due to the aforementioned bad documentation but what I read seemed to indicate that amdgpu uses early KMS and requires the EDID in the initrd). The end result is great but a massive amount of work needs to be done to get EDID reading and multi monitor setups to work. I shouldn't have to have a NVidia box around to generate edid files so that my AMD video card can display to a monitor properly.

I saw some edid progams available in Ubuntu (read-edid, parse-edid???) but couldn't get them working in Fedora. Getting properly functioning monitors shouldn't be this complicated.

Last edited by m2mg2 on 4 April 2020 at 3:04 am UTC
Tuxee Apr 15, 2020
For completeness' sake: Not always but every now and then I end up with this

[    6.954609] amdgpu: [powerplay] failed send message: PowerDownJpeg (46)  param: 0x00000000 response 0xffffffc2
[    6.954653] [drm:amdgpu_dpm_enable_jpeg [amdgpu]] *ERROR* Dpm disable jpeg failed, ret = -62. 
[    9.171323] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[    9.171364] [drm:amdgpu_dpm_enable_uvd [amdgpu]] *ERROR* Dpm disable uvd failed, ret = -62. 


"PowerDownJpeg"?

[   13.262325] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   15.730846] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   17.949442] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   20.427404] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   22.908628] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   25.392032] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   27.875584] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   30.355106] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   32.571936] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   35.052594] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   37.531160] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   40.013514] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   42.117738] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[   42.117740] amdgpu: [powerplay] SMU11 attempt to set divider for DCEFCLK Failed!


...and then it boots as if nothing ever has happened.


Well, the problem has been known for months now:

https://gitlab.freedesktop.org/drm/amd/issues/929
m2mg2 Apr 15, 2020
I have a 5700 XT using one HDMI at 24hz 1080p and one DP at 144hz 1080p. I don't have that issue. Have you tried generating edid files for both monitors and forcing them to be used instead of the built in edid detection? Are you using xorg.conf file/files? I had to create an xorg.conf and force an edid file for things to work right. The built in edid detection is garbage, especially with multi monitor. The thread seemed to indicate it could be related to metamodes, which should get resolved with the correct forced edid.

Last edited by m2mg2 on 15 April 2020 at 5:03 pm UTC
nolddor Apr 15, 2020
Just installed RX 5700 TI on my computer (Debian testing) all seems to be fine.

uname -r
5.5.0-1-amd64


glxinfo | grep "OpenGL renderer\|OpenGL version"
OpenGL renderer string: AMD Radeon RX 5700 XT (NAVI10, DRM 3.36.0, 5.5.0-1-amd64, LLVM 9.0.1)
OpenGL version string: 4.5 (Compatibility Profile) Mesa 19.3.3


ls /lib/firmware/amdgpu/ | grep navi10
navi10_asd.bin
navi10_ce.bin
navi10_gpu_info.bin
navi10_me.bin
navi10_mec2.bin
navi10_mec.bin
navi10_pfp.bin
navi10_rlc.bin
navi10_sdma1.bin
navi10_sdma.bin
navi10_smc.bin
navi10_sos.bin
navi10_ta.bin
navi10_vcn.bin


Last edited by nolddor on 15 April 2020 at 9:57 pm UTC
Tuxee Apr 17, 2020
Quoting: nolddorJust installed RX 5700 TI on my computer (Debian testing) all seems to be fine.

uname -r
5.5.0-1-amd64


glxinfo | grep "OpenGL renderer\|OpenGL version"
OpenGL renderer string: AMD Radeon RX 5700 XT (NAVI10, DRM 3.36.0, 5.5.0-1-amd64, LLVM 9.0.1)
OpenGL version string: 4.5 (Compatibility Profile) Mesa 19.3.3


ls /lib/firmware/amdgpu/ | grep navi10
navi10_asd.bin
navi10_ce.bin
navi10_gpu_info.bin
navi10_me.bin
navi10_mec2.bin
navi10_mec.bin
navi10_pfp.bin
navi10_rlc.bin
navi10_sdma1.bin
navi10_sdma.bin
navi10_smc.bin
navi10_sos.bin
navi10_ta.bin
navi10_vcn.bin

What does dmesg say? No powerplay in red? As far as firmware goes - hashes are more interesting, since the files don't vary in name and hardly vary in size. Next point: How many displays attached? Finally: It works most of the time, which means I can get 10 boots in a row without any hiccups, then 2 boots with a plethora of errors during boot, then back to normal.
Tuxee Apr 17, 2020
Quoting: m2mg2I have a 5700 XT using one HDMI at 24hz 1080p and one DP at 144hz 1080p. I don't have that issue. Have you tried generating edid files for both monitors and forcing them to be used instead of the built in edid detection? Are you using xorg.conf file/files? I had to create an xorg.conf and force an edid file for things to work right. The built in edid detection is garbage, especially with multi monitor. The thread seemed to indicate it could be related to metamodes, which should get resolved with the correct forced edid.

No I haven't generated any edid files. And for f*ck's sake I don't want to. Because...
  • I have had this displays for about two years now and

  • They never had any issues with my NVidia setup

  • They don't have any issues with kernel 5.3/Kisak's mesa (apart from that the displays are swapped when compared to my kernel 5.6 on Ubuntu's 20.04 mesa)

  • boot is pretty smooth most of the time, which would make testing of my edid setup just a PITA


I am completely aware that maybe a custom edid file could solve my problems. But this was the intention of this thread all along: That you don't get a "mildly pleasant" OOTB experience with an RX 5700. Sure, I can setup edid files, try these and that mesa versions, throw in some BIOS files, pick a mainline kernel, do some distro hopping, or draw a pentagram and sacrifice my hamster (redheads are way too scarce nowadays). It's just not something I expect to do in 2020.
m2mg2 Apr 17, 2020
Quoting: Tuxee
Quoting: m2mg2I have a 5700 XT using one HDMI at 24hz 1080p and one DP at 144hz 1080p. I don't have that issue. Have you tried generating edid files for both monitors and forcing them to be used instead of the built in edid detection? Are you using xorg.conf file/files? I had to create an xorg.conf and force an edid file for things to work right. The built in edid detection is garbage, especially with multi monitor. The thread seemed to indicate it could be related to metamodes, which should get resolved with the correct forced edid.

No I haven't generated any edid files. And for f*ck's sake I don't want to. Because...
  • I have had this displays for about two years now and

  • They never had any issues with my NVidia setup

  • They don't have any issues with kernel 5.3/Kisak's mesa (apart from that the displays are swapped when compared to my kernel 5.6 on Ubuntu's 20.04 mesa)

  • boot is pretty smooth most of the time, which would make testing of my edid setup just a PITA


I am completely aware that maybe a custom edid file could solve my problems. But this was the intention of this thread all along: That you don't get a "mildly pleasant" OOTB experience with an RX 5700. Sure, I can setup edid files, try these and that mesa versions, throw in some BIOS files, pick a mainline kernel, do some distro hopping, or draw a pentagram and sacrifice my hamster (redheads are way too scarce nowadays). It's just not something I expect to do in 2020.

If you look at my previous post on this thread before the one quoted, I had many of those same complaints. The OOTB experience with amdgpu is garbage, unless you're really lucky and have the right monitor on the right display connector. This is deeper than just the 5700. I had these issues on all 3 of my amdgu compatible cards with three different monitors. How can amdgpu not do something as basic as read edid settings? On my primary gaming set up with 2 monitors neither one had a functional display OOTB.

I didn't have to mess with any BIOS versions, jump distro's, deal with kernel or mesa versions. The only things I had to do were generate the xorg.conf, create and force the edid settings (I put the edid in the initrd but I'm not sure that is needed). BUT, it took me way too long to get there and that is way too much!!!

Overall I'm happy with my functionality and that I don't have to deal with a proprietary module every kernel update but Linux can do better. At the very least we need a utility that can read the edid settings properly, put a file in the firmware folder and add the kernel parameters to force them to be used.
While you're here, please consider supporting GamingOnLinux on:

Reward Tiers: Patreon. Plain Donations: PayPal.

This ensures all of our main content remains totally free for everyone! Patreon supporters can also remove all adverts and sponsors! Supporting us helps bring good, fresh content. Without your continued support, we simply could not continue!

You can find even more ways to support us on this dedicated page any time. If you already are, thank you!
Login / Register


Or login with...
Sign in with Steam Sign in with Google
Social logins require cookies to stay logged in.