[SOLVED] Hardware freezes at high CPU load in multi-threaded applications

liubei · 03-21-2020, 12:12 PM

My computer freezes (requiring manual hardware reboot) whenever I run anything multi-threaded at high CPU loads using an AMD Ryzen 2600X on a MSI Tomahawk B450. The CPU reaches temperatures of around 80 deg C so this is clearly not a temperature issue. This issue only arises in multi-threaded applications, for example, if I play Grand Theft Auto: San Andreas (a game from 2005 on PC) in Wine, it only uses 1 core, but if I record the gameplay on OBS Studio, the machine freezes within 5 minutes since it's multi-threaded.

I've already posted this issue on Reddit but the only solution suggested to me was undervolting the CPU, I was hoping if there were any other solutions. https://www.reddit.com/r/techsupport...m_medium=web2x

Here's what it looks like: https://imgur.com/2U2WhzD

At the end of the video, I'm pressing Alt+SysRq+B to force the machine to reboot since SysRq functionality is enabled. It doesn't work so the kernel can't do anything. Also, that monotone BEEP sound is the result of having audio/music being played right before the freeze. Its intensity seems to be related to the volume at which the audio was played

Here are the specifications:

Code:

System:
  Host: archlinux Kernel: 5.5.10-arch1-1 x86_64 bits: 64
  Desktop: Cinnamon 4.4.8 Distro: Arch Linux
Machine:
  Type: Desktop Mobo: Micro-Star model: B450 TOMAHAWK (MS-7C02) v: 1.0
  serial: <root required> UEFI [Legacy]: American Megatrends v: 1.40
  date: 12/19/2018
CPU:
  Topology: 6-Core model: AMD Ryzen 5 2600X bits: 64 type: MT MCP
  L2 cache: 3072 KiB
  Speed: 3738 MHz min/max: N/A Core speeds (MHz): 1: 3738 2: 3949 3: 3795
  4: 3702 5: 4009 6: 3901 7: 3927 8: 3928 9: 3890 10: 4046 11: 3879 12: 3986
Graphics:
  Device-1: AMD Vega 10 XL/XT [Radeon RX Vega 56/64] driver: amdgpu
  v: kernel
  Display: x11 server: X.Org 1.20.7 driver: amdgpu
  unloaded: modesetting,vesa resolution: 1600x1200~60Hz, 1920x1200~60Hz
  OpenGL:
  renderer: Radeon RX Vega (VEGA10 DRM 3.36.0 5.5.10-arch1-1 LLVM 9.0.1)
  v: 4.5 Mesa 19.3.4
Audio:
  Device-1: AMD Vega 10 HDMI Audio [Radeon Vega 56/64] driver: snd_hda_intel
  Device-2: AMD Family 17h HD Audio driver: snd_hda_intel
  Sound Server: ALSA v: k5.5.10-arch1-1
Network:
  Device-1: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet
  driver: r8169
  IF: enp24s0 state: up speed: 1000 Mbps duplex: full mac: 00:d8:61:38:c5:ff
Drives:
  Local Storage: total: 938.73 GiB used: 724.23 GiB (77.2%)
  ID-1: /dev/nvme0n1 vendor: Samsung model: SSD 970 EVO Plus 500GB
  size: 465.76 GiB
  ID-2: /dev/sda vendor: Seagate model: ST9500420AS size: 465.76 GiB
  ID-3: /dev/sdb type: USB vendor: Kingston model: DataTraveler 3.0
  size: 7.20 GiB
Partition:
  ID-1: / size: 449.58 GiB used: 287.77 GiB (64.0%) fs: ext4
  dev: /dev/nvme0n1p2
  ID-2: swap-1 size: 8.00 GiB used: 0 KiB (0.0%) fs: swap
  dev: /dev/nvme0n1p3
Sensors:
  System Temperatures: cpu: 42.4 C mobo: N/A gpu: amdgpu temp: 40 C
  Fan Speeds (RPM): N/A gpu: amdgpu fan: 1233
Info:
  Processes: 294 Uptime: 13m Memory: 15.65 GiB used: 2.17 GiB (13.8%)
  Shell: bash inxi: 3.0.37

The PSU is a Corsair RM650x.

Going through journalctl nothing seems to be out of the blue.

systemd/journalctl

Code:

Mar 21 15:40:02 archlinux rtkit-daemon[966]: Successfully made thread 2707 of process 2503 owned by '1000' RT at priority 10.
Mar 21 15:40:02 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:40:02 archlinux pulseaudio[869]: E: [alsa-sink-ALC892 Analog] alsa-sink.c: ALSA woke us up to write new data to the device, but there was>
Mar 21 15:40:02 archlinux pulseaudio[869]: E: [alsa-sink-ALC892 Analog] alsa-sink.c: Most likely this is a bug in the ALSA driver 'snd_hda_intel'. >
Mar 21 15:40:02 archlinux pulseaudio[869]: E: [alsa-sink-ALC892 Analog] alsa-sink.c: We were woken up with POLLOUT set -- however a subsequent snd_>
Mar 21 15:43:17 archlinux systemd[1]: Starting Cleanup of Temporary Directories...
Mar 21 15:43:17 archlinux systemd[1]: systemd-tmpfiles-clean.service: Succeeded.
Mar 21 15:43:17 archlinux systemd[1]: Finished Cleanup of Temporary Directories.
Mar 21 15:43:17 archlinux audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-tmpfiles-clean comm="systemd" exe="/>
Mar 21 15:43:17 archlinux audit[1]: SERVICE_STOP pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-tmpfiles-clean comm="systemd" exe="/u>
Mar 21 15:43:17 archlinux kernel: audit: type=1130 audit(1584801797.664:76): pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-tmpfiles->
Mar 21 15:43:17 archlinux kernel: audit: type=1131 audit(1584801797.664:77): pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-tmpfiles->
Mar 21 15:44:47 archlinux dbus-daemon[746]: [session uid=1000 pid=746] Activating service name='org.gnome.evince.Daemon' requested by ':1.78' (uid=>
Mar 21 15:44:47 archlinux dbus-daemon[746]: [session uid=1000 pid=746] Successfully activated service 'org.gnome.evince.Daemon'
Mar 21 15:44:58 archlinux dbus-daemon[746]: [session uid=1000 pid=746] Activating service name='org.gnome.evince.Daemon' requested by ':1.80' (uid=>
Mar 21 15:44:58 archlinux dbus-daemon[746]: [session uid=1000 pid=746] Successfully activated service 'org.gnome.evince.Daemon'
Mar 21 15:46:22 archlinux vlc[3246]: QObject::~QObject: Timers cannot be stopped from another thread
Mar 21 15:46:39 archlinux vlc[3368]: QObject::~QObject: Timers cannot be stopped from another thread
Mar 21 15:47:23 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:47:23 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:47:25 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:47:25 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:50:25 archlinux systemd[736]: Started GnuPG cryptographic agent and passphrase cache.
Mar 21 15:50:25 archlinux gpg-agent[3759]: gpg-agent (GnuPG) 2.2.19 starting in supervised mode.
Mar 21 15:50:25 archlinux gpg-agent[3759]: using fd 3 for ssh socket (/run/user/1000/gnupg/S.gpg-agent.ssh)
Mar 21 15:50:25 archlinux gpg-agent[3759]: using fd 4 for extra socket (/run/user/1000/gnupg/S.gpg-agent.extra)
Mar 21 15:50:25 archlinux gpg-agent[3759]: using fd 5 for std socket (/run/user/1000/gnupg/S.gpg-agent)
Mar 21 15:50:25 archlinux gpg-agent[3759]: using fd 6 for browser socket (/run/user/1000/gnupg/S.gpg-agent.browser)
Mar 21 15:50:25 archlinux gpg-agent[3759]: listening on: std=5 extra=4 browser=6 ssh=3
Mar 21 15:50:25 archlinux gpg-agent[3761]: Failed to lookup password for key n/77C961597E9DA8CCB38627CA79C0FF539F2E0043 with secret service: The na>
Mar 21 15:50:38 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:50:38 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:50:38 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:50:38 archlinux rtkit-daemon[966]: Supervising 8 threads of 3 processes of 1 users.
Mar 21 15:50:38 archlinux rtkit-daemon[966]: Successfully made thread 3784 of process 2285 owned by '1000' RT at priority 10.
Mar 21 15:50:38 archlinux rtkit-daemon[966]: Supervising 9 threads of 4 processes of 1 users.
Mar 21 15:51:04 archlinux rtkit-daemon[966]: Supervising 9 threads of 4 processes of 1 users.
Mar 21 15:51:04 archlinux rtkit-daemon[966]: Supervising 9 threads of 4 processes of 1 users.

pan64 · 03-23-2020, 09:06 AM

looks like overclocked. Did you run ramcheck already?

liubei · 03-24-2020, 09:44 AM

Quote:

Originally Posted by pan64

looks like overclocked. Did you run ramcheck already?

It's not overclocked.

Unfortunately, I'm not familiar with ramcheck; searching for it, I don't see how the RAM can affect this freeze. Could there be an issue with the RAM sticks? Can you elaborate?

pan64 · 03-24-2020, 10:09 AM

[if you have for example ubuntu] there is a grub menu entry called ram test (or something similar).
you can check for example this: https://www.techwalla.com/articles/h...e-ram-on-linux too

liubei · 03-24-2020, 02:47 PM

Quote:

Originally Posted by pan64

[if you have for example ubuntu] there is a grub menu entry called ram test (or something similar).
you can check for example this: https://www.techwalla.com/articles/h...e-ram-on-linux too

I don't have any issue when running 5 loops on memtester, however, if I run a blend torture test on mprime, my machine crashes within 10 seconds.

jefro · 03-24-2020, 08:41 PM

Try another prime stress test maybe to double check. Since it fails in real use and it seems to fail on this test then we can start to believe some issue.

Is there any bios setting for default or failsafe or such to rerun failed tests?

pan64 · 03-25-2020, 02:20 AM

you [may] also need to check if the problem occurred on the same memory location. I sill would try to underclock that memory in bios.

liubei · 03-30-2020, 10:15 AM

Quote:

Originally Posted by pan64

looks like overclocked. Did you run ramcheck already?

It turns out it was overclocked. The MSI B450 Tomahawk has a "Game Boost" feature that increases the CPU frequency by 450 MHz, essentially an overclock.

https://www.reddit.com/r/MSI_Gaming/...boost_in_bios/