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

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

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

    AMD engineer K Prateek Nayak recently uncovered that a ~20 year old chipset workaround in the Linux kernel still being applied to modern AMD systems is responsible in some cases for hurting performance on modern Zen hardware. Fortunately, a fix is on the way for limiting that workaround to old systems and in turn helping with performance for modern systems...

    https://www.phoronix.com/news/Linux-AMD-Old-Chipset-WA

  • #2
    Great find by the AMD engineer, i hope this lands in 6.0!

    Comment


    • #3
      That also tells us that AMD has not bothered to fix this for 20 years. On the other hand, I welcome their stepped-up efforts to improve the Linux experience for their end-users. There are still some areas where more participating could yield some benefits though, e.g. glibc or better upstream compiler support/tuning.

      Comment


      • #4
        Originally posted by ms178 View Post
        That also tells us that AMD has not bothered to fix this for 20 years. On the other hand, I welcome their stepped-up efforts to improve the Linux experience for their end-users. There are still some areas where more participating could yield some benefits though, e.g. glibc or better upstream compiler support/tuning.
        If they knew about the issue they would have fixed it.

        Comment


        • #5
          Originally posted by ms178 View Post
          That also tells us that AMD has not bothered to fix this for 20 years. On the other hand, I welcome their stepped-up efforts to improve the Linux experience for their end-users. There are still some areas where more participating could yield some benefits though, e.g. glibc or better upstream compiler support/tuning.
          I think it mostly tells us that the kernel is filled to the brim with old cruft and we can't keep track of everything.
          Over time, information gets lost.
          All it really needed was somebody to look close enough into that dark, funky smelling, corner. :P

          Comment


          • #6
            Yet we still have people always asking "why do they have to remove old unused unmaintained [email protected] from the source tree?!" like with DT_HASH. Lol.

            Complexity. Learn about it, people.

            Comment


            • #7
              Originally posted by milkylainen View Post
              All it really needed was somebody to look close enough into that dark, funky smelling, corner. :P
              This was obviously found by someone doing workload profiling and then going on a hunt for the offending "dummy reads", as indicated in the patch:

              "Sampling certain workloads with IBS on AMD Zen3 system shows that a significant amount of time is spent in the dummy op"


              The way you phrased it makes it sound like the issue was discovered through code inspection.

              Comment


              • #8
                micheal
                I hope we will see some performacetests with and without that optimization?

                Comment


                • #9
                  Originally posted by anarki2 View Post
                  Yet we still have people always asking "why do they have to remove old unused unmaintained [email protected] from the source tree?!" like with DT_HASH. Lol.
                  Given that this wasn't correctly disabled on newer systems, it seems unlikely that simply de-supporting those old chipsets would've resulted in the offending code being removed.

                  Originally posted by anarki2 View Post
                  Complexity. Learn about it, people.
                  Hopefully, the kernel has better structures for managing such chipset-specific workarounds than it did 20 years ago.

                  Comment


                  • #10
                    Originally posted by ms178 View Post
                    That also tells us that AMD has not bothered to fix this for 20 years. On the other hand, I welcome their stepped-up efforts to improve the Linux experience for their end-users. There are still some areas where more participating could yield some benefits though, e.g. glibc or better upstream compiler support/tuning.
                    Just brief history
                    April 19th, 2006​ AMD kuschelt mit Linux

                    AMD will in Dresden offenbar eine Forschungsgruppe einrichten, die nicht nur Linux auf die eigenen Prozessoren optimieren, sondern auch die Anforderungen des freien Betriebssystems in die Entwicklung kommender Prozessorgenerationen einbringen soll​
                    https://www.heise.de/newsticker/meld...ux-118161.html
                    AMD Closes The Operating System Research Center
                    on 7 November 2012
                    https://www.phoronix.com/news/MTIyMzI
                    AMD Is Hiring More Linux Engineers For The Scheduler, Memory Management, Net I/O
                    on 7 June 2021
                    https://www.phoronix.com/news/AMD-Ke...eers-June-2021
                    A Prominent, Longtime Dell Linux Engineer Recently Joined AMD's Linux Team
                    AMD on 28 August 2021
                    https://www.phoronix.com/news/Dell-M...AMD-Linux-Team
                    AMD Working to Create A New Yocto Linux Platform For Xilinx SoCs
                    on 3 May 2022
                    https://www.phoronix.com/news/AMD-Hiring-Embedded-Yocto


                    AMD Hiring For Another Open-Source GPU Driver Developer With Multimedia Expertise
                    on 6 July 2022​
                    https://www.phoronix.com/news/AMD-20...timedia-Hiring

                    And even Xilinx Acquisition is mainly due to acquire highly skilled GNU/Linux software develloper workforce

                    Comment

                    Working...
                    X