aos

Actually it manifests immediately upon starting the game. As I was re-playing the game, I was usually playing long stretches at the time, but I never had the problem just suddenly show up - it's there from the start, or it isn't happening. Although I didn't have problems for many days and they suddenly started to happen quite often, so they could be related to some particular software or hardware installation. I do run multiple programs even while the game is running - that's why I got X2 after all - so one of them may not be playing along nicely.

It happens to me with Far Cry quite often. The game seems to speed up immensely (you press a key and suddenly you're running all over the place), or there's audio stuttering. Alt-tab and setting cpu affinity on the game's process seems to solve the problem. I don't have x2 drivers installed (c&q is on "auto" in bios but it should make no difference since there are no drivers installed in windows) and the machine isn't overclocked. I'm not so sure I'd blame this on X2, but rather on incompatibility of OS, drivers and multi-cpu systems. Did anyone hear if users of regular dual cpu machines (e.g. dual opteron) or better yet users of dual-core P4's have these same issues?

Well, I've spent a week's worth of reinstalling and trying out various things. I now install only the XP with SP2 and just the chipset drivers - no ethernet, audio or ide. No windows updates of any kind. After a day, I dared to install drivers for my audio card, and that worked. I unpluged all USB devices. Today I plugged iPod in firewire, hopefully that is more stable than USB. I've had one install get bad after only a few hours - I wasn't able to open windows explorer, or any related application such as control panel, due to DPE or DEP or whatever is that executable protection called. I've never seen that before. Perhaps that was related to installing nvidia audio drivers - I needed *some* audio and thought onboard is safer than my pci soundcard, but I was wrong. Or it was related to installing some other programs, though those were all plain applications, nothing hardware related (except nero and plextools, later might be problematic with X2, I don't know anything any more). I thought it may be a virus but AVG and a few other intrusion detectors found nothing. Next install after first reboot complained that pci.sys is missing (!!). I copied it from another partition and had no more problems, and that's what I'm running now. I tried a few more installs previously using that nvlite and rvm update pack, only to end up with totally corrupted installation, in one case it wouldn't even finish installing windows. That driver and hotfix slipstreaming thing is nothing but trouble. Then of course the motherboard refuses to boot from the nero's partition backup CD (Which does the same thing as Ghost and runs caldera opendos). If I want to restore backup, I'll have to use another machine (luckily my old P4 runs everything I throw at it). I see some people have trouble with Maxtor DiamondMax 10 drives, but it always got recognized for me, and both of my drives have firmwares that are newer than anything I've seen on the board. And they worked fine with single core. There's still a possibility they are causing trouble though. It seems just about *everything* can cause trouble. So at the moment I don't use onboard networking or audio (the latter I really need as my professional soundcard doesn't have mic in, as for network I suppose I can live with using pci card). I don't dare use USB for anything but keyboard and mouse. I'm going to see if it remains stable. If not, I'll really have to get a new motherboard for my X2, perhaps this time one from the AMD's approved list, and leave this DAGF to run single core, as it has no problems with that.

Have you read through the AMD's readme file, supplied with the source code? I get the same message as you do, with 2.6.12-5 but I haven't actually conciously tried to get C&Q to work, nor I have verified that they're in the kernel. This tar file contains the Linux CPU frequency driver for AMD Athlon64 / Opteron processors. The driver obtains data as to supported frequencies and voltages for the processor from the BIOS. Version 1.00.xx of the driver uses a BIOS table called the PSB, which is not supported by all BIOSs. Included driver versions are: 1.00.09 - versions of the driver for both 2.4 and 2.6 kernels. The driver is dependent on the cpufreq driver, which is available as a patch for 2.4 kernels, and is built into 2.6 kernels. 1.00.12 - the latest version of the driver, for 2.6 only, that includes sample code to hardcode a PSB table to bypass BIOSs that do not offer the support. Version 1.39.xx of the driver can use either the PSB table or ACPI objects, dependent on whether the kernel is configured to include ACPI support or not. The driver version included is 1.39.04, for the 2.6 kernel only. This driver supports SMP frequency management and dual-core processors. It will not load on an SMP or dual-core system unless the ACPI objects are available. It will work on all 2.6 kernels, but the 1.50.03 driver (below) is preferred for kernels 2.6.10 and later. Version 1.50.xx of the driver can use either the PSB table of the ACPI objects. It supports Opteron, Athlon 64 processors, in both single and dual core versions, for all of the 754, 939, 940, and 1207 pin packages. This driver supports SMP frequency management but will not work on SMP or dual core systems unless the ACPI powerstate objects are available. It will only work on 2.6.10 and later kernels. This driver is distributed with the 2.6.13 and later kernels. The powernow-k8.c and powernow-k8.h files should be placed in the arch/i386/kernel/cpu/cpufreq directory. The kernel will then need to be rebuilt and the system rebooted. Builds of the 64-bit arch/x86_64 kernel use the same source files. For further documentation, see the linux/Documentation/cpu-freq directory.

I am pretty sure PCI is locked automatically on DAGF, or as long as you leave PciEx locked. I was as confused as you initially - last chip I overclocked was my P4 Northwood so I had to learn A64 from scratch. Anyhow, I am able to run at CPU 250 so if PCI wasn't locked, there's no way it would've worked. Some software may report it isn't locked, but some other reports it's locked. Same with temperatures and cpu core voltages - different software, different results. Can't trust anything.

So I installed a cheap gigabit ethernet card, disabled onboard lan in bios and then re-inserted dual core chip, stock speed. Removed xp limit of 10 simultaneous connections and started 5-6 random torrent sessions for a good measure. Later I played some Far Cry while stuff was downloading. No crashes for 3 hours. Then I turned on onboard lan and left the machine alone. 15 minutes later I saw it sitting on the login screen - it rebooted. So I turned onboard back off and left the machine for the night. Unfortunately, it was sitting at BSOD in the morning (Machine Check Exception, I usually get this one). So I installed another winxp on some spare space - it's always good to keep a few GB of non-partitioned space on every hdd for times like these. I bought this OEM winxp this year so it has SP2 on it already. I didn't install any drivers except for the pci ethernet (no usb, no video, no nothing). Left the machine running and I'll see what it will look like in the evening. There's still too many variables to make a conclusion. Overnight my iPod was plugged in, with iTunes started, and machine was also overclocked. Last day it would only lock up (or crash) when I'm not using it (which could imply problems with power savings mode, specifically video card as turning off the monitor is the only power saving that should kick in) but before it did lock up when in use too. Hopefully I'll be able to isolate the device or the driver that's causing this using this fresh xp install. It did seem more stable with onboard lan disabled, but not stable enough.

Reinstall of OS is very painful if you have a lot of applications and lots of peripherals like I do. It can take 10+ hours just to install software and all the updates, and then many more to configure it. Lots of USB perhipherals such as bluetooth and camera are very finicky. It took me most of the day last time I did it just to get the active sync working with my bluetooth and my pocket pc, and that's just one of several issues I had. So often it's easier and cheaper (if you believe time=money) to just buy a new piece of hardware rather than keep banging your head on the wall. So I think I will first disable onboard LAN and put in a PCI network card, then put back the X2. I have the suspicion that networking is not dual cpu ready due to either drivers or bios or the network subsystem of OS (in this case I will have to reinstall) but the fact that I had networking trouble in linux as well makes me think it's the chipset/bios thing. I'll know more in the evening.

Well, the single-core chip is running just fine (even though it's running on the multiprocessor HAL). I guess there are several culprits possible: - motherboard (or bios) defective - X2 cpu defective - OS or drivers defective - power supply insufficient However given that X2 doesn't crash when it's hot and loaded with work (which stresses cpu, memory, power supply and HT bus), but rather when it's mostly sitting idle (or doing networking), I really think the motherboard is the most likely culprit. Either that or some of the peripherals don't have thread safe drivers. Perhaps pairing a high end CPU and a budget motherboard (in fact the cheapest nf4 motherboard you can buy in Vancouver) was not such a good idea. The board works great with single core though so I'll probably leave it at that. And as I just got an unexpectedly high accounting and tax bill (read: I'm reluctant to plonk down cash on yet another motherboard), it looks my X2 may end up gathering dust for the next few months (argh!). Perhaps I'll try a fresh winxp install (since I have triple boot setup) in a few days as sort of a last resort.

Yeah, those two boards are one and the same. I really need a stable machine so I had to put back my "old" 3200+. So far the old cpu is running fine.

Unfortunately, no luck with bios upgrade (it's now 7/7/2005). Took battery out, cleared CMOS (had to disassemble half of the pc to do that), loaded defaults etc etc. Machine locked up witin 10 minutes again. The mouse did move this time but everything else stopped responding.

Thanks for the suggestion. Windows auto-updated the cpu driver and requested a reboot the first time I installed the X2. It's definitely showing two CPUs in task manager and they both get to work. According to the article you listed, it's the expected behaviour: "On Windows XP and later versions, the ACPI Uniprocessor HAL and the MPS Uniprocessor HAL recognize the existence of more than one processor and report the MP ID. Plug and Play detects that the computer devnode's hardware ID list has changed and moves the devnode back through the "found new hardware" detection process. Therefore, when you add a second processor, the MP files (HAL and kernels) are automatically installed, and you do not have to manually update the driver in Device Manager." One thing that does worry me is that during that first initial reinstall, the machine crashed as it was about to reboot because it was overclocked - not on purpose though. I just replaced the cpu and let it run with same settings (2.5G, default voltage) without realizing it. So there's a chance that the HAL multiprocessing files are corrupt. I reinstalled nForce drivers since as well as sound and video drivers. Perhaps I'll try to force reinstall of HAL before trying to reinstall the OS. Of course, it's Microsoft, so reinstall may not be avoidable. After all, even linux with its multiyear history of support for amd64 and dual cpus crashes with kernel panic when using dual core unless you run the very latest kernel, making it a major pain to get going - windows behaved much better here. The thing is, normally I don't post on support forums, but it's been a week and I'm trying to avoid having to reinstall (although I'm contemplating going RAID in which case I'd need to anyway, but then it's a really bad idea to try RAID now before I know that the machine is stable). Dealing with software compatibility issues on 64-bit linux gives me s more than enough tech problems, I'd rather not deal with hardware problems too if I can help it. Sorry for talking too much. I'm just waiting for the day to pass by so that I can go home and try new bios...

I'll try the beta bios, that's one thing I was planning to do (current bios has totally screwed up voltage selection for CPU, for example). Anyone knows when's the next official bios release scheduled for? By the way, memory is set to 2-3-3-7, at 2.5V, which is I believe according to specs (I made a mistake purchasing VX since I'd never add another fan that is needed for >2.9V, and if you don't use it at high voltages this RAM isn't worth the money as you can buy CAS 2 memory for a bit more than half of what I paid for this). When I overclocked it was set to 250MHz HTT (x4), 10x cpu multiplier, 0.8 mem divider, so memory wasn't overclocked at all.

I bought NF4 DAGF about a month ago - mostly for the fact that it had parallel port unlike lan party series - along with Athlon64 3200+ and a pair of OCZ Gold VX memory and new hdd and video card. My primary concern is quietness and stability of the machine but I certainly don't mind overclocking. I am not interested in extreme overclocking for that last MHz and bragging rights (such as pushing voltage to the limit, using high cfm fans or watercooling etc.) however, just for having very fast machine within practical and reasonable means. The machine was running perfectly with Athlon64 3200+ overclocked to 2.5GHz, on default voltage, in both Win XP and 64-bit Linux. I replayed HL2 and Doom 3, no problems with anything. Enter a week ago, when I decided to spend some hard-earned cash for X2 4400+. Apart from the fact that the new chip doesn't overclock well (second core won't work at even 2.4GHz without raising voltage), I started having random crashes. Initially I thought it's the overclocking of either cpu or memory, and during the initial few days of trying to find the limits of the CPU I did find instances where prime95 would fail, or where memory overclock would cause your typical crashes (during boot, gameplay etc.). These were fixed by which were fixed by raising voltage or dropping the speed. However, problems continued even after the machine would pass the whole night (or day) of dual prime95 testing, and they don't seem to be related to cpu or memory. Eventually, I found that they happen even at stock speeds and voltages and are not related to CPU being taxed or temperatures being high. What happens is that the machine randomly freezes. E.g. you leave it doing nothing overnight or during the day, and when youy get back to it, the monitor is off, the machine seems on but it doesn't react to any key press or mouse movements. Lately it would also just freeze randomly - screen stays on but the mouse is not moving and it's all frozen. Even worse, if I run linux, I see problems like I've never seen before - for example, machine freezes and even reset does not help! Machine comes back to splash screen but stops short of testing the memory, and subsequent resets only bring it back to that point, of initial splash screen. The only thing that helps is unplugging it from power - for more than 30 seconds! This is a seriously screwed up thing, in my 20 years as electrical engiener I haven't seen many things that would not come back after pressing the reset button! What I think I discovered is that freezes seem to be related to network activity. You can run prime for hours or run games for hours without problems but after 10 minutes of downloading something the machine could freeze, especially with multiple network streams. I had one case in Linux where network stopped responding, but machine would allow rebooting. I booted into windows and networking did not work there either - it behaved as if network adapter's cable is unplugged. I replaced the cable but to no avail (it was crushed by the desk leg so I thought it needed replacing anyway). The only thing that helped was to power off the machine completely for > 30s. That particular problem never came up again and I wrote it off to custom kernel but then problems continued in windows in their own way. I gave up on linux for the time since it's really difficult to get dual core working on the particular distribution I'm using (I had to build a custom kernel which works but with lots of issues), so don't contribute these problems to that os. But even just using windows is not giving up, it keeps freezing. Now, I haven't reinstalled XP since getting dual core but I'd really like to avoid doing that if I can. The machine did get a fresh install a month ago when I first got the mb and the A64 3200+ - the only change is dual core chip. Bios flash was done a month ago when I got the board (perhaps even before xp install) so I would think that any flash-related problems would've shown up on the original configuration as well. Since this whole setup was working fine with 3200+ - I only added second hard drive and the X2 chip - and I would seriously doubt my few-months old latest gen 500W from Seasonic is not up to the task - especially since the video card is only 660GT - the only thing I can contribute this to would be motherboard, or the OS drivers. But having problems in two OS'es makes me think that it's the MB that's the culprit. Any ideas? I'd rather not spend CDN$230 on a new motherboard (if I had to get a new one, it'd be Asus premium, which has passive cooling) and would try to reinstall windows first, but I'm not enjoying reinstallations every few weeks.