I bought a Ryzen 9 7950X and a X670E Steel Legend mobo. I got everything hooked up, downloaded the latest version of arch, plugged the USB in and now I get this immediately when I try to boot arch. I never even got to the install screen.
There is something wrong with the ACPI setting in the Bios. I haven’t figured out what yet. @wendell Did you run into this?
Updates:
This issue can be bypassed by using a GPU. It only appears when using an APU on a linux install. Updating the APU drivers in linux may also resolve. I haven’t tried yet.
Bugs:
When using HDMI cable to APU. I can access the UEFI but not Linux. If I connect the HDMI cable to the GPU I can use linux, but the UEFI screen freezes after it comes up. This issue doesn’t happen with DisplayPort.
If you install NVIDIA drivers in linux, and are using HDMI you will have a bad time. Soon as I installed the Nvidia Drivers the Tty login would freeze and both keyboard and screen wouldn’t work. I could still ssh in. This issue was completely resolved when switching to display port.
There is something funky with the stock drivers for the X670E Steel Legend’s wifi card in both Linux and Windows. After using an ethernet connection install NetworkManager in Linux, and update the drivers in windows I had no more issues. But on clean install I got weird issues, so strongly recommend setting up on wired.
I just found the issue. I was trying to install arch via the APU and got that failure message. I just plugged in my 3070ti GPU and installer is working fine now.
Thanks for the help! And hopefully this helpful to the next person.
Interesting. I was using fedora, but I immediately hopped to a 6.0 kernel release candidate, and no issues. in fact I was kinda blown away by how well everything seemed to be working.
I will probably take home the steel legend for my personal system with the 7950x I just ordered from amd.com if I can resolve this weird issue where it doesnt like my older power supply (pretty sure I have, am testing now)
For mine I am running a Seasonic PRIME Ultra Titanium 1000W, with an Ice Giant for a cooler. I’d like to let you know if I had any issues with this combo, but I am only installing Arch now. I will let you know in a few hours =)
Previously I was using an HDMI, and switching to GPU prevented me from viewing UEFI. I had to move the HDMI to the mobo, and back to the GPU each time to get them to work right depending on what I wanted to look at.
Plugging into my main monitor using display port resolved this weird issue.
I am still getting the original issue soon as I boot linux, but the GPU lets me progress further. With the GPU I was able to read more of the error message. Looks like some issue with the amd/gpu drivers. I did not spend effort to resolve that one bug yet.
Due to other issues with my Arch install, I have been booting off my USB to make fixes.
Suddenly I can not longer detect my wireless adaptor. Wlan0 no longer exists, nor the wireless card on lspci -k.
I tried installing win10 just to double check the card still works. I got stuck in a BSOD boot loop after install.
Refreshed my Bios, and trying a new USB stick… really hoping I dont have a dead mobo…
That would be horrible.
Thank you for your service. Hopefully all of the bugs are worked out before I can afford such a setup. My 6850U laptop has been taking up the daily system duties until I can replace my poor-Dozer desktop setup. 10 Years is a long time to be on PCIe2 and DDR 3 for my use cases.
UPDATE: Was able to fix the boot issue by using the 9/13 linux-firmware release - so now can boot with just the iGPU…
However, I do still see that same error message on boot (after I exit Grub menu)…it just doesn’t seem to create any issues…
I hope your motherboard is OK…
Original post:
I’m seeing exactly the same error as you - with the same hardware setup, and with only the iGPU (i.e. no discrete GPU installed) - but booting an already installed Ubuntu 22.04 with Kernel 6.0 rc7 (update: and the latest v1.07 BIOS) …
I installed Ubuntu 22.04 on a different AMD system this morning, and updated to kernel 6.0 rc7… but just realized that I didn’t grab the latest firmware from kernel git repo…
FWIW, I’ve booted a fully-up-to-date install of Windows 10, and it booted fine… I had -pre-installed the latest Radon Adrenalin drivers (v 22.9.2) from 9/22, and 3DMark (from Steam) is now happily generating very unbalanced scores using the iGPU…
I’ll add a discrete GPU - I have an RX 6700 XT to test with - and see if it fixes the Ubuntu boot issues, and also download the latest firmware to see if that helps too…and then report back on the outcomes.
I’m using two 32 GB sticks of Corsair Vengeance [5200] @ 5200 (using the Expo timings).
Definitely was the Mobo =( Luckily it failed with in the first 24 hours. At Microcenter let me swap for a new one without issue.
To rule out myself from braking things in arch. I did a clean install of windows 10. I was able to boot into windows after installing it this time. My wifi also worked, so definitely the board.
I did experience 3 BSOD. Each time I got an error about my TMP not found with 1 second later I had the critical error and BSOD. After installing windows updates, I haven’t got them again. I also wonder if that would have happened had I turned on TMP in the BIOS. (But then windows would try to upgrade me to 11…)
When I removed my cooler to return my first board, I noticed I was too aggressive in applying thermal paste. The AM5 spider leg design makes it VERY easy to drop thermal paste on top of resisters! They do have a little bit of silicon on top of them, but the coverage is not really enough. Using a tooth brush and isopropyl alcohol I was able to get everything nice to clean. Also due to the spider leg design it is also easy jam thermal paste under the lid!!! FML but THE toothbrush does work very well to solve that issue without delidding… Anyways I was fearful that the BSOD was from the thermal paste and I had ruined a 7950x =(
@Mastic_Warrior this pc is also replacing my last 2013 pc I built. It is SO FAST! However my 2013 is still perfectly useable today, so the value add of the new parts are questionable.
@edge-case Definitely some issue with the AMD apu and the latest linux kernel =(
@wendell Disabling everything ACPI related didn’t help me. I didn’t try the frame buffer yet.
I suspect your system has the same error and you are by passing it by using a gpu. The error should be the very first thing in journalctl everytime you boot. If you don’t have an error then I wonder what Fedora has that Arch and Ubuntu doesn’t. I plan to check to see if a bug report for it was filed, but as I only have a working system now. It may be a little while
My next steps are to find a game or two to stress test with. See if I get any BSOD. I am really scared that I FUBAR’ed my cpu with the thermal paste. My 2013 build had a bad cpu (not due to anything on my part). I replace every part before the cpu and wasted about 2-3 months of trouble shooting. It would boot windows and run for 20-45sec before giving a random BSOD. I am having PTSD again. I never want to real live that nightmare.
Glad you were able to get replacement motherboard (I bought my from MicroCentre too) and everything seems to be good for you…
In terms of your CPU, as long as you used “normal” thermal paste then you should be fine with the over-application as most are electrically non-conductive; Liquid Metal on the other-hand would be a big problem…
My build seems very stable so far - using a Corsair H150i PRO 360 mm AIO and Seasonic Platinum 1200W PSU; both in Ubuntu and Windows 10.
As I stated above, system seems to working well with in Ubuntu with just the iGPU (I’ve run Vulkan demos, and RPCS3 emulator and dual 4K+ displays), but I do still get the ACPI error on boot.
I booted from poweroff this morning, and here’s my log (journalctl --no-pager | grep ACPI):
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: Added _OSI(Linux-HPI-Hybrid-Graphics)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.PCI0.GPP0._PRW], AE_ALREADY_EXISTS (20220331/dswload2-326)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20220331/psobject-220)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI BIOS Error (bug): Failure creating named object [\_GPE._L08], AE_ALREADY_EXISTS (20220331/dswload2-326)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20220331/psobject-220)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: Skipping parse of AML opcode: Method (0x0014)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: 14 ACPI AML tables successfully acquired and loaded
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: Interpreter enabled
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: PM: (supports S0 S3 S4 S5)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: Using IOAPIC for interrupt routing
Sep 29 06:38:38 ubuntu-Linux6 kernel: PCI: Using host bridge windows from ACPI; if necessary, use "pci=nocrs" and report a bug
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: Enabled 5 GPEs in block 00 to 1F
and, a bit further down:
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: video: Video Device [VGA] (multi-head: yes rom: no post: no)
Sep 29 06:38:38 ubuntu-Linux6 kernel: ACPI: bus type drm_connector registered
Sep 29 06:38:38 ubuntu-Linux6 kernel: amdgpu: Ignoring ACPI CRAT on non-APU system
I’m much more familiar/comfortable with Debian/Ubuntu than Red Hat based distros, but I may give Fedora a try this evening and see if that’s any different…
I tracked down my BSOD issues. I was installing a crusty enterprise version of windows 10 instead of win10 pro commercial. Once I installed the correct iso all my issues were resolved and I didn’t get a single BSOD.
I did use ethernet to download and update all my drivers on win10, and on my arch installation. Something definitely weird with the wifi card on a clean install, so I recommend just using ethernet for initial setup.
@R-Savage Ice giant has me mid 40C at desktop and high of 65. Loading up RDR2 and doing the stress test I was upper 60s high of 79.5. When my CPU was cooler I was around 5.5 GHz. I drop to 2.9 GHzs underload.
I will have to do more testing, but it definitely seems like a perfectly fine cooler
For something to test it against I did some RTM cpu mining. Around 11am I kicked it off where it did some self tuning to find the highest hash rate . You can see where it came it came online around 2pm.
The first rig is my Dell 720. They are kind of power hungry servers, so I have it mine on the side of hosting my home lab stuff. (Is cheaper to pay the power diff of hosting + mining that is just to host on the dell, as the mining pays for the over all power cost)
The dell mines around 300-600 H/s. I think the desktop I replaced was around those same numbers. By comparison 3.5Kh/s is just insane and probably pulls the same amount of power as that dell!
I have included HWMonitor of the 7950x during the testing phase. This probably when it must have got to 95C, but it looks like dropped to mid 70s when it actually started mining, and the current core clocks look nice considering its under load. (But there are heavier loads out there)
I am using a Lian-li mesh II performance case with 2 Noctua NF-A12x25] mounted at the top. I definitely feel a nice warm breeze emanating from the case. Current fan speeds sounds like the same noise level as the compressor on a refrigerator. Listening with music or playing a game you can easily tune it out.
I definitely say this the Ice Giant for a 7950x is more than enough of a cooler. The funny thing was the Microcenter rep was trying so hard to upsell me a liquid cooler. I told him nah, my Ice Giant will work just fine.
If you were going for 100% max overclock then you’d probably would want to go liquid, but otherwise this air cooled beast is still king in my book.
I believe I may be having a similar/related issue, but I wasn’t entirely sure after reading some of the replies here; If you’d prefer I make another topic please let me know!
I’m on the X670E Steel Legend (v 1.07) and 7900X, and I am NOT using a dedicated GPU for now. I’ve installed Fedora beta, installed the latest firmware and amdgpu and confirmed that this driver is in use, but it seems the GPU clock frequency is always set at 600 Mhz and cannot be changed. Correspondingly many low requirement/basic games (CSGO) run at ~20 FPS on low settings, on very low resolutions. Can you try running the following and let me know if you have (had?) the same issue?
These look ridiculous to me. I think this might be a driver issue or motherboard issue causing the gpu clock to be ‘stuck’ at this low speed. The issue seems consistent on both 5.19 and 6.0 rawhide kernels. I would say “it’s a small igpu, what should I expect”, but the performance in several benchmarks and a video review on youtube seem to point to my performance being much much lower than it should be.
UPDATE:
I can confirm that my processor cores can all hit >=5ghz for an extended period, but when I launch something graphically intensive, there is basically no uptick in my fans/cooling – presumably because the gpu clock boost just isn’t happening.
It should go up to like 2.2Ghz on the GPU, there might be a manual setting for GPU clock in bios, I know there is a curve optimizer for GPU but I wouldn’t touch that until you get your frequency normal
@wendell fyi it looks like the newer ASRock bios versions 1.07 and 1.08 beta break the iommu groups (goes from 41 groups to just 6 in the initial bios)
I cannot see any options for ACS so (unless I’m blind) no options to fix this in bios.
Also annoyingly no way to force the boot GPU to be other iGPU
(what bios you on)
