Announcement

Collapse
No announcement yet.

Motherboards With Broken ASPM On Linux

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

  • Motherboards With Broken ASPM On Linux

    Phoronix: Motherboards With Broken ASPM On Linux

    One of the many OpenBenchmarking.org features that haven't yet been fully taken advantage is the opportunities presented by the vast collection of system hardware/software information and logs that have been submitted to this collaborative testing platform from Phoronix Media. OpenBenchmarking.org is much more than just being a storage place for benchmark results. After writing a simple plug-in this morning, here's a list of many motherboards that have broken PCI-E Active State Power Management support from their BIOS, which can lead to greatly increased power consumption under Linux...

    http://www.phoronix.com/vr.php?view=OTk4NQ

  • #2
    My exact board isn't there but the Gigabyte P55A-UD3 is. I have the UD6 so it seems likely that I'm affected. Just as well I've been running with pcie_aspm=force for a while.

    Comment


    • #3
      My Gigabyte EX58-UD3R is affected too

      Running with pcie_aspm=force seems to sort it out, though. (No sign of instability...)

      Comment


      • #4
        If your motherboard isn't on the list and you're affected, that means you should be running more benchmarks via the Phoronix Test Suite and opting to submit data to OpenBenchmarking.org
        Michael Larabel
        http://www.michaellarabel.com/

        Comment


        • #5
          " To ensure some quality, each motherboard in the list had at least two unique submissions where ASPM could be detected as not working due to the ACPI table. "

          What if they fix it in a BIOS update AFTER people go and make their submissions? The board would still be counted even though it shouldn't? I'd think you should be looking for cases where NO submissions say that the ASPM works on that board. Finding just two different submissions that say it doesn't work really isn't that helpful, IMO as mobo manufs. can claim they fixed it later.

          Comment


          • #6
            I know this was written more in a "let's show what pts can do" way, but:
            Toshiba Portable PC
            Sony VAIO
            are too vague.

            Comment


            • #7
              Originally posted by [Knuckles] View Post
              I know this was written more in a "let's show what pts can do" way, but:
              Toshiba Portable PC
              Sony VAIO
              are too vague.
              Unfortunately it's limited to what the motherboard vendor exposes from their BIOS as their vendor and product/model strings...
              Michael Larabel
              http://www.michaellarabel.com/

              Comment


              • #8
                My Lenovo W520 (4270cto) is affected. The "pcie_aspm=force" workaround works fine though.

                Comment


                • #9
                  What factors are you using to indicate that ASPM is broken on a given motherboard?

                  Comment


                  • #10
                    Not surprised that there are a crapload of Asus boards on the list given my experience dealing with their bios engineers who despite have all the proof in the world that something is broken won't listen to anyone until the issue becomes an epidemic.

                    Comment


                    • #11
                      Originally posted by deanjo View Post
                      Not surprised that there are a crapload of Asus boards on the list given my experience dealing with their bios engineers who despite have all the proof in the world that something is broken won't listen to anyone until the issue becomes an epidemic.
                      Considering the large range of boards that they sell, I wouldn't consider that a "crapload"..
                      Their engineers are Chinese and they pretty much only speak Chinese.. So they're not browsing the USA message board the same way they browse the Chinese message board.. You need to get Asus to relay your concerns over to their Chinese engineers (or maybe, learn to speak Chinese?).


                      I had complaints about my 5 year old Asus M2N-E (Socket AM2) BIOS not officially supporting AM3 CPUs. After showing proof that I (and many other people) had these brand new AM3 CPUs running in 5-year-old AM2 motherboards as an "Unknown CPU", ASUS eventually released a BIOS update that supported these CPUs.. The overwhelming majority of other mobo manufs. never released a BIOS to support an AM3 CPUs in an AM2 socket. In order for Asus to make the AM3 CPUs work in the AM2 Socket motherboard they had to delete a few features from the BIOS to make room for the AM3 CPU microcode. ASUS did provide what their customers were asking for, even if it meant axing their own "features" to cram it in there. So I think that was pretty darn cool.. No other Mobo manufs. would go that far in order to cram additional microcode into their BIOS just because their customers wanted more CPU upgrade-ability in their 5+ year old motherboard..


                      So yea, I don't agree that Asus BIOS engineers won't listen to anyone.. They most definitely do listen... How many people have gone to Asus and asked about this ASPM problem? I'd bet no more than one or two, if any.. Heck, nobody even knew the cause of the power management problems until Phoronix pointed it out..
                      Last edited by Sidicas; 10-10-2011, 12:45 PM.

                      Comment


                      • #12
                        My desktop is an Abit KN9 that is also affected by this bug. The chipset in this board is nVidia nForce4 (CK804) and pcie_aspm=force switch works so far and no instability. Are there any known quirks with this chipset and I have the latest BIOS for this board dated 4/28/2008

                        Comment


                        • #13
                          Originally posted by Sidicas View Post
                          Considering the large range of boards that they sell, I wouldn't consider that a "crapload"..
                          Their engineers are Chinese and they pretty much only speak Chinese.. So they're not browsing the USA message board the same way they browse the Chinese message board.. You need to get Asus to relay your concerns over to their Chinese engineers (or maybe, learn to speak Chinese?).
                          There are several more, those are the only ones tested. Asus has an excellent record of carrying over BIOS issues though a complete line. (PS, their engineers speak Hokkien and english).

                          I had complaints about my 5 year old Asus M2N-E (Socket AM2) BIOS not officially supporting AM3+ CPUs. After showing proof that I (and many other people) had these brand new AM3+ CPUs running in 5-year-old AM2 motherboards as an "Unknown CPU", ASUS eventually released a BIOS update that supported these CPUs.. The overwhelming majority of other mobo manufs. never released a BIOS to support an AM3+ CPUs in an AM2 socket. In order for Asus to make the AM3+ CPUs work in the AM2 Socket motherboard they had to delete a few features from the BIOS to make room for the AM3+ CPU microcode. ASUS did provide what their customers were asking for, even if it meant axing their own "features" to cram it in there. So I think that was pretty darn cool.. No other Mobo manufs. would go that far in order to cram additional microcode into their BIOS just because their customers wanted it..
                          Sorry but companies like Abit (RIP) and Tyan have a better record when it comes to items like that (in fact Gigabyte does a lot of that as well). Asus doesn't stand out at all in those reguards.

                          So yea, I don't agree that Asus BIOS engineers won't listen to anyone.. They most definitely do listen... How many people have gone to Asus and asked about this ASPM problem? I'd bet no more than one or two, if any.. Heck, nobody even knew the cause of the power management problems until Phoronix pointed it out..
                          Might want to search the forums, have posted transcripts of Asus's uselessness from past BIOS bugs.
                          Last edited by deanjo; 10-10-2011, 12:48 PM.

                          Comment


                          • #14
                            Originally posted by DeepDayze View Post
                            My desktop is an Abit KN9 that is also affected by this bug. The chipset in this board is nVidia nForce4 (CK804) and pcie_aspm=force switch works so far and no instability. Are there any known quirks with this chipset and I have the latest BIOS for this board dated 4/28/2008
                            You can be sure there's no hope of that ever getting fixed... ABit went out of business..

                            Comment


                            • #15
                              Originally posted by Sidicas View Post
                              You can be sure there's no hope of that ever getting fixed... ABit went out of business..
                              I see, but it should be possible for the kernel ASPM code to make use of any quirks in the NF4 chipset though, or maybe editing tables and such in the BIOS

                              Comment

                              Working...
                              X