AMD Bulldozer, can it get even worse?

A number of reviewers have reported problems with a Blue Screen Of Death on AMD’s Bulldozer, even with stock settings:

The reviewers reporting this problem include, but are not limited to:

http://www.bit-tech.net/hardware/cpus/2011/10/12/amd-fx-8150-review/9

http://www.sweclockers.com/recension/14579-amd-fx-8150-och-fx-8120-bulldozer/21

http://muropaketti.com/artikkelit/prosessorit/amd-fx-8150-zambezi,3

http://www.hardware.fr/articles/842-22/jeux-3d-total-war-shogun-2-starcraft-ii-anno-1404.html

As you can see, Total War: Shogun 2 triggers this bug. Which is rather ironic, since it is part of AMD’s Gaming Evolved program.

The hardware.fr review says that AMD has been able to reproduce the bug, and is working on a fix.

Now, what is happening here? Let’s look at Microsoft’s site for an explanation of this particular error code:

The CLOCK_WATCHDOG_TIMEOUT bug check has a value of 0x00000101. This indicates that an expected clock interrupt on a secondary processor, in a multi-processor system, was not received within the allocated interval.

..

Cause

The specified processor is not processing interrupts.  Typically, this occurs when the processor is nonresponsive or is deadlocked.

So apparently there is a problem with the multiprocessing in Bulldozer. The interrupt handling can get messed up at times.

Note also that the above review-sites have used different hardware, so it is unlikely that this is caused by a specific motherboard, BIOS, memory, videocard, PSU or anything like that. Also, as most overclockers will be able to tell you: if your system is just unstable because of problems with voltage, overheating or such, then although you may get BSOD’s, they are generally not reproducible. The actual reported error is quite random, it just depends on where the CPU fails first. So you generally get a different error code with every BSOD. In this case, it is always the same error. It is also an error that I have never seen myself, during all the years of overclocking, dealing with unstable memory, voltage and whatnot…

So Bulldozer is starting to look more and more like Barcelona: Not only is the performance underwhelming, but apparently the stability is not 100% either.

Update: AMD has finally acknowledged the issue in their knowledgebase: http://support.amd.com/us/kbarticles/Pages/STEAMGamesonAMDFXplatforms.aspx

They recommend installing the latest BIOS, which have updated microcode, as expected. This updated microcode appears to fix the BSOD problem, however people who tried this bios are complaining about poor performance. So I wonder what the microcode fixes exactly. After all, the fix for Barcelona’s TLB was just a microcode update as well, but it disabled the TLB cache altogether, since the hardware was just broken. As a result, the performance was severely compromised.

Update: Tomshardware is investigating the issue: http://www.tomshardware.com/news/STEAM-AMD-FX-Processors-BSOD-bios,15630.html

AMD’s KB article focuses solely on some boards with the 990FX chipset. However, there have been various other motherboards with 9xx, 8xx and 7xx chipsets, that have been sold as AMD FX-compatible, and some of them have never received the proper BIOS updates, and as such, still suffer from the BSOD.

If you have problems, I suggest you leave a message there. They might be able to get your motherboard vendor to release a fix for your board.

This entry was posted in Hardware news and tagged , , , , , , , , , , . Bookmark the permalink.

105 Responses to AMD Bulldozer, can it get even worse?

  1. Sune Marcher says:

    Will be interesting to see if this warrants a processor recall, or if they’re going to ship an “AMD Processor Optimization Service” (nice term for ‘bugfix/workaround’) like they did with the unsynced TSC bug of the early AMD64 CPUs.

    First time I’ve seen this STOP code, gotta give AMD credit for that :)

    • Scali says:

      Yea… but at least the TSC ‘fix’ did not impact performance.
      I’m more worried about a Barcelona-like fix: the ‘fix’ was to just turn off the TLB cache altogether, which resulted in a massive drop in performance.
      AMD had to respin the architecture to fix it properly.

      However, in Barcelona’s defense… although the bug was there, it was exceedingly rare in everyday use. So most people could get away with an unpatched CPU.
      This time the bug seems to be of a deterministic nature: it seems that everyone had the problem with Shogun 2, everytime. And who knows what other software is affected as well. So in that sense this bug is more serious. It has a much more direct effect on the end-user.

  2. Steve Brodeur says:

    Hi ive experienced this issue not only on Shogun 2 , but also on Deus-ex human revolution and dawn of war 2 retribution and its seems all amd games!

    • Scali says:

      Well yes, the issue reported by the BSOD is a very lowlevel one. So it is not a bug in the game itself. The bug occurs at a lower level, in the depths of the Windows kernel, it seems to be the clock timer that the scheduler runs on. Windows reports that one of the cores is no longer responding (probably deadlocked). There are likely many situations where the bug will pop up.

  3. Pingback: [Thread OFICIAL] AMD Bulldozer & Bobcat (Chipsets AMD 9xx Socket AM3+) - Página 25

  4. Pingback: AMD FX (Bulldozer) - Topic Ufficiale - Pagina 33 - Forum di Tom's Hardware

  5. Bonzai says:

    Hi face. Meet palm. In other news, Rage PC is the ugliest game I’ve seen in quite sometime. And I give the gameplay a bit up there with Crysis 2. Look up “Terribad”. To be fair…… well, actually I can’t. It’s pretty bad. The gameplay really isn’t as bad as Crysis 2 (To me.) but when I look at things, or climbe a ladder and see how ABSOLUTELY LOW RES AWFUL it is. Well, I lost interest. It’s like playing a game through mud and sometimes you find diamonds that sparkle on top of the slop. Goddamnit Carmack.

    I look forward to Battlefield 3, it will take away all this awful pain. And make me mad at players, instead of the game.

  6. Alan C Salt says:

    I was under the impression that there was some consistency in BSOD codes while overclocking.
    BSOD codes for overclocking 1366
    0x101 = increase vcore
    0x124 = increase/decrease vcore or QPI/VTT…have to test to see which one it is
    0x0A = unstable RAM/IMC, increase QPI first, if that doesn’t work increase vcore
    0x1E = increase vcore
    0x3B = increase vcore
    0x3D = increase vcore
    0xD1 = QPI/VTT, increase/decrease as necessary
    0x9C = QPI/VTT most likely, but increasing vcore has helped in some instances
    0x50 = RAM timings/Frequency or uncore multi unstable, increase RAM voltage or adjust QPI/VTT, or lower uncore if you’re higher than 2x
    0x109 = Not enough or too Much memory voltage
    0x116 = Low IOH (NB) voltage, GPU issue (most common when running multi-GPU/overclocking GPU)
    0x7E = Corrupted OS file, possibly from overclocking. Run sfc /scannow and chkdsk /r

    • Scali says:

      No. If your system is unstable because of voltage or such, the behaviour is non-deterministic. It will run okay until there’s a glitch. But you can never know where or when that glitch happens.
      In this case the BSOD occurs everytime you try to start the game, at the exact same spot. And the error is always the clock watchdog. Besides, there is no overclocking involved.

  7. Pingback: AMD Bulldozer Benchmarks Leaked - Page 101

      • Scali says:

        It’s not a Windows 7 bug.
        If it was, ALL systems would have problems with it ALL the time.

        My god, has it come to this? AMD apologists commenting on MY blog? If you don’t realize you’re in the wrong place, perhaps you should ask JFAMD why.

    • Scali says:

      Yes, it *can* happen on other CPUs as well (obviously… if it could never happen, Windows would never have had a watchdog that checks for and reports problems in the first place). But that means the system is faulty somehow.
      So yes, occasionally you may run into faulty systems with any kind of CPU. The problem here is different however: it seems to happen to ALL Bulldozers, ALL the time!
      Not only the listed reviewers have mentioned that various Steam games don’t work, there are also tons of people on forums reporting the same thing. This is a structural problem, not just one faulty system out of thousands.

      • Deuce says:

        What exactly makes you think it happens to ALL Bulldozers ALL the time? All I see is a random spray shit of posts on some forums and blogs. How many Bulldozers do you think are out there by now?

        Remember the TLB bug? That was acknowledged and fixed very publicly and quickly by AMD. If there is a hardware problem with the Bulldozer that causes this ALL the time on EVERY Bulldozer ever shipped, don’t you think AMD would have said something? Or is this one of those AMD sucks so much and Bulldozer is so bad they are covering it up because they are a bad company or something?

      • Scali says:

        Forums and blogs? I point out no less than 4 review sites that report the same issue. Is that not official enough for you?
        Why do I think it happens to ALL Bulldozers ALL the time? For the simple reason that there are too many reports of the same issue, with the exact same software in the exact same place.

        This is not the same as a TLB bug. The behaviour of this problem is very deterministic, where the TLB bug was not of a deterministic nature… It could crash, or it could not. You never knew exactly when and where.
        With this bug, the problem can be reproduced perfectly. The Steam-version of Shogun 2, Deus Ex and Portal 2 are reported to cause a BSOD by a lot of people… and the error is always the clock interrupt one I mentioned above.
        With the TLB bug something like that would never happen, it would be too much of a coincidence. Just like with ‘normal’ instability related to overclocking/undervolting/poor PSU/etc… you’d rarely get a BSOD in the exact same place twice, let alone that you’d get the exact same error.

        As for why AMD hasn’t said anything… I don’t know. The people who contacted AMD about the issue (reviewers and forum users alike), all say that AMD thinks it’s a problem with the CPU, and that the CPU should be RMA’ed. Hardware.fr even says that AMD has reproduced the crash themselves and are looking into it. An AMD rep said that they would require a minimum of 4 weeks to approve a BIOS update for release. Perhaps we should take that as an indication that AMD thinks the problem can be solved with a BIOS update.
        I suppose we will just have to wait.

  8. Pingback: NEWS - Procesoare - Page 184 - My Garage

  9. MakeItSo says:

    LOL at AMD for this selection of reviews at http://blogs.amd.com/news/2011/10/20/amd-fx-reviewed/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+amd%2Fall+%28All+of+AMD%29 Talk about between a rock and a hard place, AMD is having a hard time covering up Bulldozer fiasco.

    • Scali says:

      Indeed… I mean, come on… Some of the biggest and most reputable review sites are missing from that list (Anandtech? Tomshardware?).
      And I don’t even want to get into some of the sites that they DO list.

  10. Miles Marx says:

    Looks like I’m running into this as well…Just had it happen in minecraft, and also many times running y-cruncher stress test :(

  11. Scali says:

    In other news… the linux patch for Bulldozer’s cache aliasing issues does… nothing for performance:
    http://openbenchmarking.org/result/1110200-AR-BULLDOZER41

    Dang, another magic bullet for Bulldozer fails to deliver!

  12. Pingback: [Sammelthread] AMD Bulldozer "Zambezi" 32nm "New CPU Architecture" Sockel AM3+ [Part 3] - Seite 246 - Forum de Luxx

  13. Pingback: AMDs FX-8150 "Bulldozer" Processor - The reviews are out! - Page 2

  14. Pingback: AMD Bulldozer released and it a big time FAIL. - Windows Community Forum

  15. David Mannock says:

    As someone who has found bugs when pushing scientific software to its limits, I know that it can be frustrating. Gaming is one area in which the poor abused CPU must operate beyond its design specs with absolute perfection over a range of temperature, pressure and water content beyond anything imaginable anywhere on Earth or in space. As has been pointed out in this thread, other cpus produce BSOD in similar and not so similar conditions. Yes, AMD will have to work on some patches for these issues with the BD chips, but with a new architecture that is not unexpected. If everything new worked perfectly 100% of the time, these forums would be unnecessary and would be empty! The obituaries, however, would be far more extensive as humans everywhere failed to keep pace with the game on the machine. Only the lucky few who survived the game would be digitized and sent to the virtual realm inside the box, never to return. Lawns would go unmowed, bloodied windows unwashed, only those not playing the game would survive. OK, you can all carry on with your mental masturbation now!

    • Scali says:

      Nonsense, at stock speed, running code is *always* within design specifications. It’s not like game code is anything other than any other code, let alone that it would be ‘more abusive’.
      And no, anyone who ‘pointed out’ that BSODs happen on other CPUs doesn’t know the difference between deterministic and non-deterministic behaviour.

    • k1net1cs says:

      “As someone who has found bugs when pushing scientific software to its limits, … ”
      Is it wrong if I actually read that “As someone who OC’ed his CPU and made it unstable when running Folding@Home and/or BOINC projects, … ” after reading your entire comment?

      You really have no idea about how CPU or program works if you consider gaming as a cause of a CPU to ‘operate beyond its design specs’.
      Most games are GPU-bound, with the exception of Civilization games, especially the latest one.
      Also, those BD BSODs were replicable on different system configurations with the same applications (games) and the same conditions (when loading a level), regardless it’s on stock speed or OC’ed.

      Lay off those forum posts, Wikipedia and The Complete Idiot’s guides’ knowledge, and refer to real computer science books.
      Because when I read apologetic and ignorant trolling comments like yours, I masturbate.

    • SpladamB says:

      The BSOD mentioned above is not random and is not from pushing the processor. My FX-6100 has no problem crunching games and is pretty damn stable, hours of gaming without crash. But I can reproduce the above bsod just by launching Portal 2, Deus EX Hr, and Homefront in Steam. These are games I paid for and cannot play because my processor has a flaw that neither AMD or Steam or willing to say much about. What is your defense for that. Why do you think both parties do not answer my queries on the above mentioned. Do I not deserve some kind of information. Steam told me to write down the BSOD after I told them what it said. They already knew what it said. And your above paragraph, with your flurries of sarcasm and prose about something you know nothing about, is the only mental masturbation I observed when I came to this blog in the second week of searching for answers to my problem. You push scientific software to it’s limits huh. Has limits does it? You a scientist are you? Frustrates you when your “scientific software” crashes when your pushing it’s “limits”? You mean my problem is like when your PC crashes after 8 hours of prime95? I think that’s what you are saying.

  16. FX says:

    Here is a funny video of bulldozer that I found http://www.youtube.com/watch?v=0x4MHUgMlOI

  17. PixelOz says:

    Most games are GPU-bond with the exception of the Civilizations games? No, sorry but that statement is absolutely incorrect. Most games today – that are worth anything anyway use both the CPU and the GPU extensively.

    The 3D rendering part of many games runs on the GPU but many other things do not. Some other things like Phys-X physics computations also run on the GPU of Nvidia cards and maybe a few other things in very, very new games may also use some of the available GPU power but if that is the case this is barely staring to appear on the scene now.

    Games are very demanding on all the available hardware power to be able to perform well. Ai for example doesn’t run on the GPU it runs on the CPU and there are plenty of things that have to run on the CPU too, plenty.

    That statement is very, very incorrect.

  18. Old Dude says:

    I can reproduce the crash under Linux (Ubuntu 11.10, Linux 3.0.0-12), on a FX-6100. With 10 heavy duty math processes (100% utilization per core), the system crashes within about 2 hours. With 2 or 4 processes, the system does not crash (runs 10+ hours successfully). I’m testing 6 processes now to see where the threshold is.

    I’m guessing that, when they went from N cores (pre-Bulldozer) to N/2 modules (Bulldozer), they missed something in the microcode that synchronizes the hand off of interrupt system responsibility from one module to another. Hopefully just a race condition that can be fixed with by inserting a couple of idle microcode cycles during interrupt system hand off synchronization, with no real performance degradation. Hopefully nothing that requires a complete redesign of the interrupt hardware system hand off synchronization protocol.

    • Old Dude says:

      Just a follow up:

      With all 6 cores fully loaded with the above math process, the system crashed in less than 2 hours. With just 5 cores loaded, the system has run for over 8 hours, as of this morning. So, it looks like leaving one core lightly loaded, at least under Linux 3.0.0-12, prevents the crash.

      Older Linux systems, and Microsoft Windows systems, may show different results, due to differences in task scheduling between the different operating system versions.

      On the upside, for my purposes, the FX-6100 performs better than projections for a Phenom or Phenom II processor. In general, at the moment, Bulldozer is either a blessing or a curse, depending on your application and your operating system.

  19. NewImprovedJDWII says:

    I’m sorry but WoW! Amd really?

    What a peace of overpriced $hit.

    Is it even possible to fix this design? Please Amd just make a Phenom III x8 and add the new instruction sets add 256Bit FP on !EACH! Core Add a new memory controller, Improve the prediction unit, Speed up the cache and improve the clocks more and shrink to 32nm and Improve the Die space so their is not so much wasted space on your 32nm die like with BD.

    I don’t want to hear some Amd employee tell me it would’t make that much of a difference in most multithreaded apps the 8 core just Barley(5% or less) beats the 6 core Phenom a 8 core Phenom with all the improvements i listed would easily be faster and would even challenge the 2600K and 2500K more.
    This would be such a better product then a 8 core Faildozer!

    • Scali says:

      Since he refuses to say what was fixed (in fact, at this point, nobody even confirmed that it actually fixed anything), there is nothing to comment on at this point.
      In fact, from the video you can’t even tell whether it’s the Steam version of the game or not. Standalone versions of games didn’t suffer from the issues.

  20. Pingback: Cateva intrebari, doar pentru cunoscatori!!! - My Garage

  21. Mendel says:

    FYI: Sampsa Kurri (of Muropaketti) confirmed in the Muropaketti forums that he can now run both Shogun 2 and Deus Ex though he gets very low framerate with Deus Ex

    I´ll provide link even though it is in Finnish
    http://murobbs.plaza.fi/prosessorit-emolevyt-ja-muistit/818895-listaus-peleista-ja-ohjelmista-jotka-eivat-toimi-amd-fx-prosessorilla-korjattu-2.html#post1707576884

    • Scali says:

      Sounds like this ‘fix’ does the same as the Asus 0705 BIOS for the Sabertooth board: http://hardforum.com/showpost.php?p=1037960408&postcount=149
      Note that Asus released the 0810 update for better CPU support with Bulldozer.

      So it could be that there is a Bulldozer microcode update in the latest BIOS updates for AM3+ boards, which causes the problem with Steam (but probably fixes others). And it could be that this guy removed/disabled the microcode update in the BIOS image, so that the Steam issue goes away, but instead, issues that appeared with the 0705 BIOS are back, apparently.

      At any rate, it doesn’t sound like a proper fix.
      It really does look like we’re waiting for AMD to release a proper microcode fix.

  22. Jon says:

    It hqpped to me also with my Ubuntu Web Server. Its NOT just Windows. I wasnt even running a game. I was just installing simple apt-get applications and it all just keeps freezing. Also when I tried to set the ASUS bios to the performance level, it horked bad.

  23. Blowdozer Sucks says:

    lol to anyone who bought a Bulldozer. Glad i waited, Ivy Bridge here I come!

  24. Just wanted to note that this problem seems to be happening to all Bulldozer users who are trying to run the new Serious Sam game.

    I’ve been getting a mix of 101’s and 124’s. I’ve tried modifying core volts and setting CPU affinity and nothing.

  25. Pingback: Serious Sam 3 Crashes, Freezes, Errors, Performance, and Fixes | SegmentNext

  26. Ben says:

    I got my two new desktops last Friday and both suffer with bsod 101 at times, also prime is not stable even under stock settings – rounding and illegal sumout errors.

    This is the same for both pcs which are of identical spec.. Anyone have anything new to add to this situation? Or any possible fixes?

    Ben

  27. Old Dude says:

    On 11/16/11, ASUS put out a BIOS update (M5A99X EVO BIOS 0813). It is listed as improving system stability. I loaded this into my system a couple of hours ago.

    On my math tests, it effedtively slows system throughput down by about 10%. For me, that is still not bad, as I had calculated a performance increase for Bulldozer of 60%-70% higher (per GHz) than the performance gain achieved in the Phenom to Phenom II transition. Therefore, for me, Bulldozer delivered what AMD promised, as it still delivered about 30% performance gain (per GHz) over the previous generation.

    I will post a follow-up, once I know whether the BIOS update fixed the problem. I’m only about one hour into the test, so it is too soon to tell whether the update has fixed the problem.

    I realize that, even it the fix works, some gamers may not be happy with Bulldozer. However, if the games do not crash, that would certainly be some improvement.

    I guess we will all be waiting for Piledriver, to see what kind of actual performance gains can be obtained with the Bulldozer micro-architecture. Perhaps a full speed hardware fix for this problem can be added as part of the Piledriver development cycle.

    As an aside, has anyone tested an Opteron 6200 series chip to see if the problem can be reproduced in that CPU?

  28. Old Dude says:

    OK. The system used to crash within 2 hours. With this BIOS update (M5A99X EVO BIOS 0813) from ASUS, it has run the math test all night long, with all cores fully loaded.

    Looks like they have a fix, but that fix costs 10%-12% in system throughput. My programs all fit within the system caches, so I don’t think the slowdown has anything to do with the memory subsystem.

    This system is running Ubuntu 11.10 (Linux 3.0.0-12).

    If anyone has a friend at ASUS that could tell us exactly what the fix is, I would love to know.

    For my purposes, Bulldozer is still a success, even with the slowdown imposed by the fix.

    The test will run for another 4 hours or so, but it is Thanksgiving here in the US, and I have holiday functions to attend. I will post here if the test crashes before completion. However, it has run for over 12 hours, and is more than 80% complete. I do not expect it to crash in the home stretch.

    I’m glad someone found a f ix for this problem, and it didn’t require a CPU recall.

  29. killer19 says:

    Hey everyone, this issue is most definatly reproducable. Just as a test I loaded CoreTemp, and the started a stability test with Prime95, and everytime the machine would BSOD, with the same BSOD Code as mentioned above. The one thing I did notice was Cores 4, 6, or 3 and 5 would almost completely shut off or go idle immediatly causing the BSOD. I have test with stock clocks and was unable to produce the error, but when ever I OC’d the processor(8120) passed 4.0 Ghz and no matter how high the voltage it would bsod during Prime95, and when running games or other apps that tax the processor. Also on ocassion even at stock clocks when running most games from within steam it would BSOD. I am starting to wonder if AMD released a defective cpu. Just my thoughts.

  30. Pingback: Pc 1500ron - My Garage

  31. pettabytes says:

    Thank you for the info, good work

  32. Jon M Knight says:

    After reading this and the comments, I just got this processor and it is sounding like I should promptly return it…

    Yes, I do have the processor, and yes, I’m experiencing this same issue with Deus Ex, one of the few games I actually want to play (so I am majorly ticked off, since my old, still functioning duel core processor probably could have still played the game, but this newer quad core processor can’t? ).. My motherboard is a 970a-ud3 Gigabyte board, and I’m trying one last time to update the bios in hopes that the beta bios flash thingy will work (and that it wasn’t what was installed before).

    • Scali says:

      Well, yes, the saddest part of this whole affair is how this issue is being dealt with.
      Firstly, the issue is clearly there. However, no major hardware review site has written an article about this. AMD has made a token knowledgebase-article for the issue, but other than that, they have not made any communication about it either. So most potential buyers are unaware of the problem (my blog is one of the few places where the issue is being discussed, and I can only reach a modest audience).

      Secondly, AMD lists only three BIOS updates. There are tons of other motherboards out there with the same issues. Many of them don’t even seem to have an official BIOS update yet, with the fix included. Again, things don’t seem to be communicated very well. This only makes it worse for people who have bought a Bulldozer system and suffer from the issue.

      If you can return it and get your money back, I think you should. Firstly, to give AMD the message that they aren’t taking care of their customers well enough. Secondly, even if you get the issue fixed, the Bulldozer is just a poor buy anyway. Its performance is not that impressive, its powerdraw is actually quite bad, and the price is not low enough to compensate for either. You’re better off with a Phenom or an Intel system.

    • SpladamB says:

      Yes, the UD3 bios tool will just continue to install the latest revision even if you already have it. Use the manual method by downloading the file from gigabyte and installing from that, at least you will know if it’s a new revision.

  33. Dz0 says:

    Just wanted to say Thanks. I have been scratching my head looking for a fix for my problems for a few months now, and this has been the only site to give me light on the issue. The various games it happened with were of course no help, and AMD knowledge base/support is rather ineffectual as i think you will agree.
    Luckily/Unfortunately it seems i am stuck with the CPU now, but my motherboard has a new BIOS update that supposedly addresses it.(this update was of course not available when i first started looking, figures)
    I got the same CPU for my wife(mostly out of laziness for her PC build, just using whatever i was getting or close), so i may look into a return on that however.

    Thanks again. Almost 2 months of annoyance, cleared up on my lunch break. Cheers!

    • Scali says:

      Yes, it’s amazing how little media coverage there has been on this issue.
      As you see, a number of reviewers already reported the issue during the reviews done at launch time.
      It took a while for AMD to even put up an obligatory KB-article, including only 3 BIOS updates (while there are numerous other motherboards with the same issues). And none of the larger hardware sites bothered to run an article on it. So most people who have this problem with Bulldozer have no idea what causes it, let alone that there might be a BIOS update that fixes it for them.

      What makes it worse is that there are AMD fanboys going around posting that this is a Windows 7 issue. Which is obviously nonsense, as people with Intel CPUs or AMD Phenom/Athlon CPUs have no problem running the same games under Windows 7.

      Anyway, I hope the BIOS update fixes your problem.

  34. SpladamB says:

    Interesting, I’m currently going back and forth with Steam support about this. Does anyone have an FX-6100 or FX-8xxx that can run Portal 2, Deus Ex HR, or Homefront on steam. Let us know if you do so we can see if it is indeed all Bulldozer’s. It’s kind of ridiculous that AMD suggest I revert to an old Bios revision isn’t it? Not even gonna try. Fix it people, or tell us it can’t be done. That’s good damage control!!!!

    • SpladamB says:

      -side note: I can reproduce the BSOD no matter what my CPU and GPU clocks are, default or otherwise. Tried everything, that is except reverting my BIOS revision. That just opens the door for all the other things the new revisions fixed!!!! That’s not even a good patch.

    • SpladamB says:

      Well I did try, and both BIOS utilities will not let me revert? Any suggestions?

  35. peter says:

    I got the FX 6100 6 core Bulldozer, and I have been pulling my hair out ever since I got it. I use the Asus M5 A99 EVO motherboard and I didnt know who to blame, Asus or AMD. I got the latest BIOS update and the problems still happen:
    – Some games like DIRT and now Microsoft Flight cannot even be installed because unzipping .CAB files fails!
    – I constantly have Adobe Flash plug ins in the browsers crash when my kids do homework
    -random computer crashes
    I hate this! From what I read I guess its the CPU and not the motherboard that is doing this. Anyway, the latest BIOS upgrade from ASUS has NOT fixed my problems. I was really looking forward to laying Microsoft Flight with my son and now I can’t even install it – DARN!!

    I sure hope there will be a real fix one of these days !!!!!!

    • SpladamB says:

      Sounds like you have some other technical issues. My 6100 on a Gigabyte GA990 ud3 does everything flawlessly except play those three damn games. But other than that, it pushes win7 great, spreads out the core load now, and for the most part, runs really fast. No random BSOD’s, just the above mentioned.

      • peter ritter says:

        hi

        I think you are right. I think the problem lies with the Asus motherboard. I reinstalled the operating system from scratch and bought myself a Phenom X6 Black edition, and I still get occasional system crashes – although less than before. I think I will get myself a Gigabyte motherboard too, and try the bulldozer again. I quite liked it because its definitely very fast, and runs quite cool. The computer was definitely very responsive with it. I think part of the blame is also with Asus software, but I think there are definitely some hardware issues.
        I have a Crosshair IV motherboard ROG from Asus on another computer which has been ‘perfect’ – never an issue, but this M5 A99 EVO is awful – I should have gone with the crosshair – and spent more money.

        P.

      • Scali says:

        You should have gone with a Core i5 or i7 system :)

  36. Patrick says:

    I have BSOD with this cpu before, with the exact same error. After update the bios to version 0901, and *reset the bios to default settings*, and stop overclocking. After that, windows works okay again.

    Do not use the stock fan to overclock, it is a disaster. Use liquid cooling.

    GPU: 2x Nvidia 470
    CPU: FX8150
    RAM: 8gb 1600MhzDDR3
    Board: ASUS M5A97

    Good luck.

  37. MJ says:

    I have the latest version bios and I’m still getting this problem trying to run Deus Ex Human Revolution. It usually restarts too quickly for me to see the error screen, but at least now with the mobo bios update I can get into the game and play it for a little bit before it freezes up.

    It should be noted that I’m also having this same problem just running starcraft 2 on low graphics settings. I could run starcraft 2 on that setting on my 10yr old pentium 4 with no problem. What is going on AMD? Technology shouldn’t be like this if you want to gain customers.

  38. Sanjay Das says:

    I have AMD FX 4100 CPU and GA-LMT78-S2P and getting same error “a secondary processor interrupt —” i have updated BIOS but it nothing done…Please suggest…

    • Scali says:

      Return it and get a Phenom CPU or an Intel system.

      • SpladamB says:

        Scali, the architecture did not live up to the hype, but the fx-6100 is a smoking fast $150 budget build processor that runs SUPER STABLE, and got me up and playing Skyrim maxed out for dirt cheap. It’s definitely not as bad, or as close to as bad as you make it out to be. I was pissed at first, but this machine has run awesome at 3.9 on stock air on hot days in Louisiana without a problem, and has given me hours of killer gaming without crashing. Crashes less than my E5800 dual core, which is almost never. Seriously, in the last nine months maybe two, maybe, random crashes on any software, and one was Skyrim with something like 8 new mods turned on at once, and when restarted, ran great. Pushing about 23-26 mods on Skyrim WITHOUT A PROBLEM. The only time the limitations showed so far was Call of Pripyat, which for some dumb reason only uses one core on all chips. I had that game maxed out and ran core #1 at 100% almost the whole time. The first time I “seemed” to notice the reduction in IPC’s. That said, the game ran great, really. So far, even though at first I felt a little ripped off, I, as an owner of said processor, and a gamer, have come to like this cheap ass little chip that lets me do so much on my pc at one time without a problem. I mean I was impressed the first time I alt tabed out of skyrim and browsed the net for wiki stuff and watched videos and forgot I had a bad as game with a buttload of mods still running. That more often now, and I have gotten distracted before and launched other games and software without realizing that I had other stuff running, stuff that by itself, would push my last pc. And I’m always impressed when these things happen, cause for a while, people like you had me thinking my computer was crappy.

      • Scali says:

        Just because it (sorta) works doesn’t mean it’s a good product. You sound like someone I know… He bought a Pentium D shortly before Core2 Duo was introduced, even though I told him to wait for Core2 Duo.
        Sure, the Pentium D works, and after he overclocked it to 4.2 GHz, it could more or less keep up with my stock E6600, but that doesn’t make it a good system (and unlike Bulldozer, it didn’t have any stability issues/bugs either).
        He still paid too much for what he got, and his CPU is still horribly inefficient. Using almost three times as much power as mine, requiring a bigger and noisier PSU and all that. But even after all these years, he won’t admit that the Pentium D was a bad deal. The terms ‘cognitive dissonance’ and ‘rationalization’ come to mind…

        For $150 you can also get a Phenom II X6 1045T, Phenom II X4 980, which still are better deals than the FX6100. Or a Core i5 2300/2400 for just a few bucks more, but giving you so much more in terms of value for money.

      • SpladamB says:

        (Reply to Scali below) Yeah, the Pentium D was some hot garbage, and props to your buddy for getting it up to 4.2, if he had it stable at that clock, he must’ve had some monster cooling. I did consider the x6 1045t for a bit, and it was a little more at the time I believe, but I figured I’d get something brand new and upward scaling, and correct me if I’m wrong, but bulldozer owners should see performance increase under Windows 8 as Microsoft has stated that the task scheduler in 8 IS optimized to use Bulldozer “modules”, so in that sense it is upward scalable. But as much as I wanted to hate AMD for this bullshit, (and they in turn fired their entire marketing and PR divisions, which I believe was made up of mostly fanboys who stretched the facts on what the engineers told them, at the very least), I have come to enjoy my new “Scorpious Platform” PC (and that is the campiest, fanboy corny marketing crap ever) and have found that I have indeed gotten my value out of my purchase. Probably could have done better, but could not have done it for cheaper, and at the time, it was all about my budget. ($615 all together not counting OS). I could grab an x6 1100T right now on the cheap, but the bulldozer is stable and fast right now, and I would like to see what gains (if any) I get out of Win8. I’m putting money aside for an Ivy bridge i7 build (paired with a HD 79XX) at the end of the summer or early fall, but I WOULD recommend the FX-6100 to anyone putting together a budget build right now (as well as the x6 1100t)

      • Scali says:

        The worst Pentium Ds had a TDP of 130W, not that different from the 125W TDP of AMD’s Bulldozer. Later steppings were ‘only’ 95W TDP, so actually quite decent compared to today’s standards (or well, AMD’s standards… Intel is going down in TDP again, 77W for a quadcore now).
        And yes, you are (mostly) mistaken. Windows 8 includes a Bulldozer-friendly scheduler, but it doesn’t make a world of difference.
        Don’t be fooled by comparisons between Windows 7 and 8 on Bulldozer: Windows 8 includes various other improvements, which are not Bulldozer-specific. Phenom and Core CPUs also get a boost from running Windows 8.
        And even in these comparisons the difference is only a few %, not quite what I’d call ‘upward scalable’.
        And sadly, AMD did not fire their entire marketing and PR divisions. The biggest ‘fanboy marketer’ of them all, John Fruehe, still works at AMD as director of marketing.

        Also, why are you doing this to yourself again? Why would you pair an Ivy Bridge with a HD79xx when you can get a GTX6xx?

      • SpladamB says:

        That said, I would like to add that it seems the processor, or architecture, limitations of the bulldozer seem to compound or multiply themselves with the Bulldozer 8 core offering, and the FX-81XX does not seem to offer the same price to performance that the smaller bulldozers do. I would NOT recommend one of them to anyone, for that I would suggest Intel or x6 Phenom 2’s.

      • peter says:

        What I really like about my FX 6100 is that it runs really cool, even with a small heatsink. Its practically cold at idle, assuming you have a motherboard that manages power properly. My Phenom X4 955 was heating up the room – which was unpleasant – especially for non- gaming activities.
        But, yeah, after replacing a faulty disk that caused all sorts of problems, which I blamed on the CPU at first, I’m quite happy with the bulldozer now. I just ‘feels’ fast, – and I’m about to build another one.

    • SpladamB says:

      Make sure your Bios is the same revision as your mobo, ran into that problem before, some updates won’t work and can even brick your board. Ex= if your mobo has a 1.1 in the bottom corner than it’s a revision 1.1 and the bios software is specific to that revision.

    • Scali says:

      Tomshardware is investigating the issue: http://www.tomshardware.com/news/STEAM-AMD-FX-Processors-BSOD-bios,15630.html
      Your motherboard is not among the listed boards where a fix is available. And it is in fact being sold as AMD FX-compatible: http://www.gigabyte.com/support-downloads/cpu-support-popup.aspx?pid=3833
      If your board still has the issue with the latest BIOS, then apparently what AMD said to Tomshardware is not exactly true. It’s not just the boards they listed that needed the fix.

  39. brian says:

    This stupid bug is still happening (FX 6100 owner), even w/ an April 2012 BIOS/chipcode patch installed. I’ve found a way around the problem by limiting active cores to 4 in my BIOS, but that’s really a horrible workaround to have to settle for.. I wish I had just gotten the Phenom II. I just got this chip about 14 days ago, so I’m going to contact Newegg and see if they’ll let me trade it in.

    • brian says:

      Newegg is letting me return the chip, even w/ no box and it having been used for 14 days (outside of their return policy). I’m guessing a lot of other people have been complaining about this CPU – I don’t think Scali is exaggerating when he says this is a problem happening to ALL FX- owners, ALL the time. This bug is re-godddamn-dickulous.

      • Scali says:

        Well, it doesn’t happen to everyone anymore, because some boards have a BIOS update that fixes the issue.
        For some strange reason though, many boards sold as FX-compatible have not been updated. So even today, many months after the introduction, a lot of people still run into this problem with a brand new system.
        And for some reason no major hardware site has ever given any attention to the issue.

      • brian says:

        Posting this in hopes that another FX-6100 owner on an ASRock board might read this (stop error 0x00000101 when playing certain steam games or using VideoLAN). I found a way to get their patch to work. You need to first install the 1.2 BIOS, then the 1.3 BIOS, then the 1.4 BIOS. I have the Asrock 970 Extreme3 edition, so your mileage may vary depending on which edition you have – but ‘step’ updating the BIOS seems to have fixed this issue for me.

        And Scali, it is more than suspicious how little coverage this issue has been given. Thanks for speaking the facts on this and keeping this post active. Trying to find information about this on AMD’s website was a complete waste of time. Take care and thanks again!!!

  40. Pingback: Does Your AMD FX Platform BSOD with Steam? Read This. | Free Best Trends

  41. ECOTOX says:

    Interesting posts here, figured I’d leave my piece

    I got bulldozer FX8120 at the beginning of february and have yet to have this issue on anything I have done. I’ll make a list

    Deus Ex HR
    Vindictus
    HL1
    HL2 Ep1&2
    Team Fortress
    PCSX2
    Kernel Compilation
    Image/Video Editing
    DC Universe
    Crysis
    ….list goes on for games and such and dont feel like listing EVERY one

    I’m not sure why I’ve had no issues, and I’m even clocked to 4.3ghz right now. As I understand it may be a combination of un optimized task scheduling where windows/linux (i run both) dont know how to probly schedule tasks to the cores in BD. for instance you may have 2 threads that share some data thread 1a & 1b, running along with thread 2. Thread 1a is put on core 1 while thread 1b/2 are both put on core 3, causing issues due to the fact that those cores do not share fetch/decode and other resources. Now this is only really a guess, but it seems plausible as the way BD works would need a better optimized thread scheduler in order to get the maximum performance and to eliminate potential issues. Microsoft has yet to fix the scheduler from what i here but I believe it has been fixed in linux aswell as changes to the micro code for the procs to reduce/eliminate the occurrence of the issue as much as possible. This is may still be occurring for some but from what I know and have seen it isnt 100% in AMDs hands to fix as it requires modifications to task and thread scheduling aswell. Windows 8 from what i here will be better optimized for BD, and linux from what i here has already patched the scheduler.

    like i said though, I cant say this 100% for sure on anything, all i know is none of this has happened to me

    PS sorry to dig up old postings =3

    • ECOTOX says:

      also sorry for all the typos 0 .0

    • Scali says:

      Well, if your motherboard has a BIOS with the fix included (which has been available since October 2011), you would never see the issue.
      The problem was/is mainly that:
      1) When I originally wrote this blog, there was no official fix available yet. In fact, AMD had not even acknowledged the issue yet.
      2) When AMD eventually published the KB article, the news was not picked up by any major sites, so hardly any users knew about it. Aside from that, AMD originally linked to only 3 BIOSes (yay, there are a few more now).
      3) Motherboards shipped with outdated BIOSes for months after the issue became known and BIOS fixes became available. But since the issue was never made public by any major site, most people did not even know they had to update their BIOS to fix their problem.
      4) AMD concentrated purely on the motherboards with 990FX chipset. OEMs sell many boards with various other chipsets, with an “AMD FX-compatible” logo on the box. It appears that some of these boards go unpatched even to this day.

      The task scheduling is a different issue. It makes the AMD FX CPUs slightly less efficient than they could be, but it does not make them unstable.
      Microsoft is not going to update the scheduler in Windows 7 or earlier OSes as far as I know, so AMD FX owners will need to upgrade to Windows 8 to get the Bulldozer-friendly scheduler. I suppose it is just unfortunate timing. When Intel launched their HyperThreading technology, it more or less coincided with the launch of Windows XP, so a new OS with an optimized scheduler (and various other modifications to synchronization objects) was available at almost the same time.
      Bulldozer was probably too late for Microsoft to include it in development for Windows 7, and Bulldozer was released too early for Windows 8.

      Don’t get your hopes up though, the new scheduler only brings a few percent more. It is not going to make Bulldozer more competitive with Intel’s offerings.

  42. Pingback: Перезагружается компьютер при запуске игры call of duty mv3 - Страница 6 - Форумы otstrel.ru

  43. daggerbite says:

    I know this is a late post, but figured I’d add my situation too. I’ve just got my hands on a machine with a Crosshair IV Formula mobo, AMD-FX processor, and am blue-screening on the usual suspects of steam games (Shogun 2, Space Marine etc). This is a just built fresh system – not even 24hrs old at this point, with a fresh install of Win7 64bit etc.

    The fix pointed to from everywhere when I google is to update the bios as per http://support.amd.com/us/kbarticles/Pages/STEAMGamesonAMDFXplatforms.aspx

    But that only mentions Crosshair V, mines an IV. Furthermore, mines currently flashed to the latest IV version, 3027 (http://uk.asus.com/Motherboards/AMD_AM3/Crosshair_IV_Formula/#download)

    I’ve sent a support ticket into Asus, asking for help. Not sure what to do otherwise.

    • Scali says:

      Yes, I’m afraid that’s pretty much the story: AMD has only officially acknowledged the problem on 990FX-based boards. Many other boards have been sold as ‘AMD FX-compatible’, or at least they were suggested to be compatible since AMD claimed for a long time that AMD FX would be backward-compatible with AM3 boards.
      The problem is not related to the chipset however. The problem is that AMD FX CPUs ship with broken microcode. The BIOS needs to update the microcode in the CPU everytime the system boots.

      I notice in your case that Asus has listed the AMD FX CPUs on the CPU support list, but ‘Beta Support only’.
      That probably means that they know it doesn’t work, but there won’t be a BIOS update to fix that.
      I’ve heard the same story for various other boards.
      It’s a shame that everyone loves AMD so much. If it were Intel, people would probably have gone for a class action lawsuit long ago (then again, Intel would probably not have let it get that far. They also replaced the buggy Pentiums free of charge back in the day. AMD has *never* done anything like that).

      • daggerbite says:

        Yeah it doesn’t look good. However I heard about, then downloaded and ran Orochi-CEG, and it now runs the steam games fine with no issues. I have no idea what the file actually is or does, googling it just brings up links to the download, or links to forum posts that then recommend using it to fix the same issues..

      • Scali says:

        Yea, even the guy who made Orochi-CEG (The Stilt) doesn’t know why it should have any effect on the BSOD problem at all.
        He made the thing because even after the BSOD problem was solved, some systems ran the game incredibly slowly.

        From what I gathered, Orochi-CEG fixes some bad information in the tables used to control the power saving states of the CPU. Somehow it gets stuck in some slow powersaving mode.
        Perhaps as a side-effect of this fix, the cores no longer get out-of-sync anymore in the first place, which is what caused the BSOD.

  44. In many cases problem disappears if reduce CPU voltage. I dont know why, but works. And lets to finally get so desired overclocking. I think they can not patch microcode – this part of CPU has not microcode controller at all. But they can set optimal voltage by the BIOS.

    Try to follow CPU voltage on old BIOS versions, and open one small fact that the CPU exceeds voltage set by the BIOS when it jumps from reduced state to full load. Not very frequently, but enough to catch this BSOD.

  45. marktrinder8@hotmail.co.uk says:

    I have the GA-78LMT-S2P board rev 4 ive updated bios with the software that came with the cd but no luck.any drivers asus list for the board dont work with 64 bit win 7

    games that crash
    portal 2
    saints row 3
    mw3
    mw2
    and more

    btw running amd fx 4100

  46. teocamber99 says:

    I got motherboard asrock 870 Extreme3 R2.0…should i download this BIOS?

  47. Dullahan says:

    I have the Crosshair V Formula http://www.asus.com/Motherboards/AMD_AM3Plus/Crosshair_V_Formula/

    With AMD FX(tm)-8150 Eigth-Core Processor 3.97 GHz

    I installed the last drivers and BIOS stuff like you said but when i started a game (Darksiders 2) my pc crashed… again….

    I really dont know what else to do and i spent a lot of money on this new pc…

    Help me please!

  48. domdom40 says:

    After numerous months of playing with AMD fx-8150 overclocking and gaming included: seams that the problem described above happens only when CPU is significantly overclocked (above 4.2 Ghz) with FSB and NB voltage alterations even minor that without Fine tuning (CPU voltage ) or ram frequency / voltage timings.

    Never happened with my PC with CPU or Ram at regular clocks.

    it happens randomly during gaming but did not occur during SIS sandra stress test !
    I have never submitted any of my overclocking configuration to a regular stability test
    It seems that this issue is linked to a latent instability situation caused directly by the overclocking experiment.

  49. Rodrigo Descalzo says:

    Do these new piledrivers present the same problem?
    In our lab we have built a cluster of five FX-8150 to run quantum chemical calculations but we cannot finish any calculation without a random crash occuring. The only way to prevent that is to limit each process to use only one core — but that’s not what we bought these bulldozers for. Work is being seriously delayed as a result. Looking for viable alternatives now.

    • Scali says:

      Well, the problem is caused by broken code found in the BIOS of various motherboards. For some boards there are BIOS updates that fix the issues, for other boards, there is no fix/update.
      I haven’t heard any specifics about Piledriver (most people probably bought a new motherboard which no longer has a bad BIOS), but I wouldn’t be surprised if Piledriver CPUs would show the same issues as Bulldozer in a board with a broken BIOS.

  50. Jeannot61 says:

    I have a Crosshair IV Formula with bios 3027, FX -8350 and above , no problems with Windows 7. The installation was done with my 1090T before putting the FX.
    I also tried the FX -8150 on this Crosshair IV Formula, and such , no problem!

  51. Jeannot61 says:

    Never had a BSOD or anything like that!

    • k1net1cs says:

      Read the article thoroughly about what may cause the BSOD, and check the date of when the article posted.
      It’s MORE THAN A YEAR AGO.

      Please exercise a more logical thinking the next time around.

  52. Jeannot61 says:

    I made this comment because many users still encounter this problem on many forums. Otherwise, it is true that I could not pay attention to the date! ;)

  53. I solved … in the BIOS and restart the Default check. then just put the DOS BIOS updates. 1 – CPU VCore 1.225 Manual, the rest all automatic! 2 – Remove SV1 Virtualization. and ready solved the BSOD 0 × 101 and 0 × 124. :-) This is with the motherboard (ASUS M5A99X_EVO) Windows7_64bit.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s