Announcement

Collapse
No announcement yet.

A 20 Year Old Chipset Workaround Has Been Hurting Modern AMD Linux Systems

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • ryao
    replied
    Originally posted by yump View Post

    Yeah, I would really like to see details of how this was found.
    He likely had the idea to check for this kind of issue and devised a way to check for it. I suggest emailing him to ask. I would be interested in hearing what he says.

    Leave a comment:


  • Chewi
    replied
    Some of you may recall the early Ryzen CPUs had an issue where they would just randomly freeze up after hours or days under Linux. It only seemed to happen when the system was very idle. I said half-jokingly that this wasn't seen on Windows because it never gets that idle. I then heard through the grapevine that people inside AMD thought that may indeed be the case.

    I'm no expert, but I now reckon this kernel bug was causing it to happen more often than it otherwise would have. Maybe it's just as well. If it hadn't happened so frequently, AMD may have taken much longer to fix it.​

    Leave a comment:


  • Radtraveller
    replied
    Originally posted by coder View Post
    No, IBS refers to Instruction-Based Sampling, which is a hardware-assisted profiling mechanism.

    https://www.phoronix.com/news/Linux-...IBS-Perf-Tools
    yeah, but I couldn’t resist..
    it does seem, according to the article, linux suffered an irritable bowel with amd..
    someone dug into the “bowels” and found out why and performed some minor surgery to fix it.

    shrug. Was a joke fer chrissakes.

    Leave a comment:


  • cyring
    replied
    This is how I have been fixing that performance issue within CoreFreq
    1. Boot kernel blacklisting any cpu idle driver
    2. Start my kernel module "corefreqk.ko" with parameters to register my own assembly idle loop. See Q&A
    3. Default Idle Route is for Zen, an I/O-Wait; AMD families 16h & prior, with HLT; MONITOR/MWAIT for Intel
    4. ​​​​​Can change Route at any time from the CLI "corefreq-cli" while monitoring freq, C-state entries, voltage, power impacts
    github.com/cyring/CoreFreq
    CoreFreq is a CPU monitoring software designed for the 64-bits Processors. - GitHub - cyring/CoreFreq: CoreFreq is a CPU monitoring software designed for the 64-bits Processors.
    Last edited by cyring; 27 September 2022, 09:56 AM.

    Leave a comment:


  • Mahboi
    replied
    Originally posted by ATLief View Post

    Is no one else going to say it? Ok.

    ...They should rewrite the Linux kernel in Rust.
    Rust really has become the new "I use arch btw"...

    Leave a comment:


  • Linuxxx
    replied
    Originally posted by RealNC View Post
    I hope this gets backported to LTS. It would be interesting to test whether or not this is the source of the performance issues I have with some emulators ever since I upgraded from my old Intel system to a 3700X one. The emulators will run full-on for like 2ms or so to emulate one frame, then idle the rest of the time until the next emulated vblank. If I don't use GameMode to switch from the schedutil to the performance governor, I get sound dropouts and animation hiccups.
    No, if anything, looks like this will make the situation worse in combination with the schedutil governor, because another user reported the following with this fix applied:

    After this patch i have a lot of slowdowns on Yuzu that I didn't have before.

    amd-pstate + schedutil + 5950x​
    https://www.phoronix.com/forums/foru...16#post1348316

    Leave a comment:


  • CochainComplex
    replied
    Originally posted by ATLief View Post

    Is no one else going to say it? Ok.

    ...They should rewrite the Linux kernel in Rust.
    Since the Kernel has some experimtal support for Rust written drivers, it might be possible in the far future. I dont know how fissible it is, but maybe rewriting parts/modiles peu a peu could transform the kernel in the longrun. And rewriting those parts would force people to completly reassess existing code. Albeit not breaking compatibly at once. On the other hand maybe some new concepts wouldn't be incorporated to the code base due to obvious incompatiblities.

    Leave a comment:


  • yump
    replied
    Originally posted by ATLief View Post
    Does anyone know if the "performance" governor has this problem? The way it's described it sounds like a schedutil thing (or whatever the older version is called).
    This is unrelated. Schedutil is (one of) the CPU frequency governor(s). The governor this is confusing is the CPU idle governor. cpufreq says how to work slower. cpuidle says what to do when there is literally nothing to do.

    Leave a comment:


  • ATLief
    replied
    Does anyone know if the "performance" governor has this problem? The way it's described it sounds like a schedutil thing (or whatever the older version is called).

    Leave a comment:


  • ATLief
    replied
    Originally posted by Mahboi View Post
    Horrid thought because of the necessary workload, but I wonder how much code clarity, speed and maintenance efficiency we'd have if we canned Linux and rebuilt a new OS from scratch with the experience Linux has given us. OpenVPN vs Wireguard style.

    Yes yes, I'm just dreaming.
    Is no one else going to say it? Ok.

    ...They should rewrite the Linux kernel in Rust.

    Leave a comment:

Working...
X