Computer freezes while playing CM

[foamy]

Hi good people - anyone got an answer for my recent problem:

While "Flying" around during CMAK and CMBO battles the screen sometimes just freezes, the computer locks up (Ctl/Alt/Del will not work) and I get a single Beep from my box, this single beep sometimes repeats itself.

Only way out is to re-boot.

I got windows 2000, Asus K8V SE MB,GE force Ti 4200 graphics card, Athlon 64, 500 mb RAM. No changes to the system for over a year and it has worked perfectly up till now ???

Any clues or suggestions will be much appreciated.

[JasonC]

Yet another case of a Athlon system giving a freeze and crash while running 3D graphics...

[JasonC]

Pulled off an old website help board, buyer beware, for information purposes only etc -

"An issue has been identified that could result in the corruption of video data shared between AGP graphics adapters and AMD Athlon or AMD Duron family processors when running Microsoft Windows..."

Another item - "I remember from a while back that nVidia and AMD blamed each other for this bug. It seems both pieces of hardware tend to fight over a certain part of the AGP texture memory, hence the hard-lockups."

[foamy]

Thanks JasonC - this sounds serious, but I wonder why this would show up after a long, long time with no problems.

Also there must be a gizzilion people with AMD and NVidia systems, I am assuming this is a "Rare" situation, any ideas on what kind of circumstances might trigger this conflict ??

[JasonC]

http://www.gamecrashes.com/

Scroll down to the section under "AGP Bus Drivers"...

[Redwolf]

An Athlon 64 is not the same as an Athlon, actually they have few in common. In particular, the chipsets are different and hence they don't have the same AGP bugs.

Does the error disappear when you run your computer with the case side open?

Do you play any other 3D games?

[JasonC]

No the error does not disappear when I run with the case open.

Errors occur in Total War series and Capitalism II as well as in CM. Typically when the cursor or view is moving rapidly, sometimes in TW when large units with many polygons cross into each other.

AMD had architecture problems with their Athlon optimizations specifically conflicting with AGP memory allocation and caching, back in Windows 2000. I think it is likely everything since has just been patch up, the real issue never addressed.

That real issue was that dynamic re-allocation of cache space sometimes overwrote portions that the AGP protocol expects to be fixed, as dedicated to video pages. When they wrote the original optimization stuff they were only thinking of PCI. And just turning all of it off isn't viable for them, because without software juggling techniques they are snails compared to intel chips.

Why doesn't everyone see it? Not everyone is running a 19 inch or larger LCD display, with the same response times, has the same memory, etc. And it is intermittent - specific dynamic reallocations have to go wrong at the same time high demands are being made on video memory.

All sorts of nonsense answers get peddled for it, from drivers to heat to power. But look around the web and you will find hundreds of people reporting it, and precious little in the way of actual answers. And the companies have a poor record on the subject, blaming each other and telling people to just get the latest drivers - which I have, doesn't solve it.

[JasonC]

System specifics -

Athlon XP 3200+

1 gig RAM

GeForce 6600 256 meg card

Sceptre X9G Naga II 19 inch LCD (8 ms vertical response)

Windows XP Home with SP2

[Schrullenhaft]

JasonC - What motherboard are you using and what BIOS version is flashed to it ? What "latest drivers" do you have for the chipset ? Does your GeForce 6600 have a power connection on it and is that plugged in ? How many amps does your power supply have on the 12+V line ?

While there have been inconsistencies in AGP implementation and design over the years, chipsets (revisions & newer designs), BIOS updates and drivers have helped to alleviate some issues.

foamy - that single beep sounds like it may be an overheating warning. Is the fan on your GeForce Ti 4200 spinning at a good rate ? You may want to check to see if it is oscillating in speed (speeding up, slowing down, speeding up, etc.), which can reduce its cooling capacity significantly (you may need to direct a flashlight/strong light at the fan to notice this). As Redwolf suggested, does taking off the case sides help at all with this problem ? For most computer systems that have been stable and all of a sudden start experiencing problems, it is overheating that is the most common culprit (or static-discharge damage of some sort - which is usually just fatal for the chips and not an intermittent problem).

[ May 28, 2006, 12:10 PM: Message edited by: Schrullenhaft ]

[JasonC]

MB - can't possibly be bothered to look it up. BIOS - OEM, ditto. Both from an emachines T3265, maybe 18 months old. Drivers - latest on nVidia website as of oh about a week ago. Power to the video card plugged in - of course, you think I'm daft or something. People offering useless tech advice about problems they know nothing about always do.

[foamy]

Hello Schrullenhaft, thanks for acknowledging my original question

The problem is so infrequent (2-3 times a week for the past month) that I don't have any other observations to help with the diagnosis.

However, the fan speed on the CPU does sometimes run very, very fast and gets noisy, on these occaisions I have not had any issues of freezing up.

I am going to do a simple strip-down and check for things like dust, heat sink condition connectors etc, etc just to be sure nothing simple is the problem.

For discussions sake - supposing it is an over-heating problem, would overheating of the video card cause the computer to freeze and the single beep ??

[Jack Carr]

My system specs are almost identical to JasonC's. I just recently experienced a problem where my machine would simply shut off. I re-seated all of the cards, memory, processor and the problem went away. I found a RAM module was loose, the clip was not fully clamped down.

Just thought I would throw this in. Sounds like this problem is a bit different, but you might want to check this out.

Good luck!

[Schrullenhaft]

foamy - yes, it's possible that the videocard could be locking up resulting in a single beep error. However this error wouldn't be eminating from the motherboard for a videocard overheating (unless the overheating problem affected the motherboard to the point it produced an error condition of its own - which is possible).

The AMD64 has 'Cool 'n' Quiet' Technology (which appears in the Power Options control panel as the 'Power scheme' of "Minimal Power Management" - a somewhat confusing designation). Typically this technology adjusts the CPU frequencies and voltage levels depending on the processor load. It can also adjust the fan speed of AMD-supplied heatsink fans. For this feature to work you would need 'Cool 'n' Quiet' enabled in the BIOS and AMD Athlon 64 utilities installed (Cool 'n' Quiet for Windows 2000 is near the bottom).

Are you running any of the ASUS utilities for the motherboard such as PC Probe (which monitors temperatures, fan speeds, etc.) ?

Back to the most likely culprit, an overheating GeForce Ti4200 videocard. None of the motherboard monitoring utilities will really tell you anything about the videocard, unless the videocard's heatsink fan was plugged into a fan header on the motherboard (which is most likely not the case here). Your intention to go in and clean out the system and check the connections is probably the best place to start. Beyond that it involves a bit more work when it comes to the videocard. The videocard's heatsink fan is the most likely culprit since it is common for them to fail (and it's possible that this is the sound you're hearing regarding 'fan noise'). With a failing fan, you'd probably have to replace the whole videocard heatsink. There are after-market heatsink models that fit the Ti4200's (usually available for less than US$25 online).

So the best course of action would be to do the cleanup and check as you intended to do and then install some monitoring software (such as the ASUS PC Probe) and see if it reports anything unusual regarding temps or fan speeds (for everything other than the videocard). One word of warning regarding monitoring software (and BIOS reports of the same info), they can be off by several degrees. Typically BIOSes are 'calibrated' to the particular signal voltage levels put out by the monitoring chips, but even then they can still over- or under-report temperatures and voltages.

JasonC - I don't insult people on this board, for any reason. None of my questions are intended to insult or belittle anyone who is seeking help here. In the real world many details can be forgotten by even the most savvy technical people and a simple reminder of some details may be all that is necessary to solve the issue. At the minimum it is a starting point to go over the littany of possibilities regarding the troubleshooting of a problem. I have no idea what people have gone through in their quests to solve their problems, so the only thing I can do is ask questions and make suggestions - starting from the beginning with the mundane and simple.

If you were insulted, then I apologize.

[ May 30, 2006, 09:16 AM: Message edited by: Schrullenhaft ]

[Redwolf]

If the error goes away with the case open then the problem is that your machine is overheating, plain and simple.

Check the temps of CPU, chipset and video card. The temperature of most modern NVidia cards is displayed in the system preferences for the video settings, or countless downloadable utilities.

CM does not have problems on any common CPU, that Athon problem rumor is bull****. I spent countless hours on Athlon XPs and 64s in CM. Even if there was such a problem, it wouldn't go away with the case open.

[JasonC]

Redwolf - with what video hook up, PCI or AGP? With whose card? Driving what monitor, how many pixels at what speed?

The temp of the video card reads 56C to 62C.

Case off does not eliminate crashes.

Reseating memory etc does not eliminate crashes.

I have an Athlon notebook too, CM is entirely stable on it. But then it isn't connected to a 19 inch LCD with 8 ms refresh through an AGP slot, is it?

If you don't know anything about the problem, admit that you don't know anything about the problem, do not pretend that it cannot exist. It exists. Your guesses about its supposedly necessary cause are false, empirically.

[Schrullenhaft]

I have an Athlon notebook too, CM is entirely stable on it. But then it isn't connected to a 19 inch LCD with 8 ms refresh through an AGP slot, is it?

I can almost guarantee that your Athlon powered notebook is communicating to the video (whether it is a northbridge-integrated or dedicated ASIC) via an AGP bus. Your laptop may not have an AGP slot, but that doesn't mean it isn't utilizing AGP (the bus of which logically resides in the northbridge). Driving a 19 inch, 8ms refresh LCD display means almost nothing. The only issue this would have is the amount of data moving over the AGP bus for the resolution. The frequencies to drive higher resolutions and refresh rates would only be an issue to the video graphics core. Would your laptop have problems if it externally drove this LCD monitor (assuming you're driving it with an analog signal) or on the other hand would your desktop run in a more stable manner if the 3D resolutions were lowered ?

Your laptop's current stability would seem to poke some holes in the theory that a significant CPU bug is the primary culprit for your lockups (unless only certain core revisions have the problem). This PDF on the AMD Athlon XP Model 10 (which your XP3200+ most likely is) mentions most of the known errata on the CPU and nothing seems to pinpoint any AGP-related problems (though this doc may not be absolutely comprehensive).

[JasonC]

AMD uses various adaptative optimization methods, to among other things speed through cache bottlenecks. It is one of the principle ways they achieve better processor performance without a higher absolute clock speed. A side effect is that the data ranges used for different purposes (including caches) is not stable, but depends on the whole memory load. AGP has a different expectation.

It is therefore unsurprising that changing the specific profile of the memory load can change the stability of the system, and do so in on-again, off-again manner, not a deterministic "as soon as you do this, it will break" manner.

This problem was acknowledged by AMD soon after Athlons and AGP starting having to work together, specifically under Windows 2000 and Linux. Patches were made for both. The Linux one was specifically tagged as an immediate and partial solution, because it basically hacked around the problem (most dynamic re-allocations would not hit the video bits etc). If they fixed it permanently in a foolproof manner under XP, then bully for them. Since I have these symptoms however, and I am not remotely the only one reporting them, I sincerely doubt they did so.

[Hpt. Lisse]

Two more items, Foamy -

Same thing happened to me this fall (Barton 2500 OC'ed to 2.1 GHz, FX5900, WinXP Pro, 3x256mb RAM, Epox NForce2 mobo) All driver hunts were to no avail., etc. No heat problems.

2 things happened simultaneously - I ran MemTest and one of my 3 sticks of RAM failed right away (annoying, PC3500 Kingston HyperX, maybe 2 yrs. old.)

BUT

Right after that, my entire mobo crapped out. Yep, bad capacitors, as so many have from that particular era of manufacture. With the failures ontop of each other, it's impossible to say which might have been responsible for the lock-up/beep freeze. (If at all.)

So download MemTest (it's small) and give it a shot. Also, physically check your mobo for swollen/leaking capacitors. I'm sure Schrullenhaft has a link to MemTest here somewhere...

Worth a shot.

[ May 31, 2006, 05:53 AM: Message edited by: Hpt. Lisse ]

[Redwolf]

Originally posted by JasonC:

AMD uses various adaptative optimization methods, to among other things speed through cache bottlenecks. It is one of the principle ways they achieve better processor performance without a higher absolute clock speed. A side effect is that the data ranges used for different purposes (including caches) is not stable, but depends on the whole memory load.

Sorry Jason but this is a bunch of rubbish. I used Athlon XP and 64 system with no problems other than that the chipsets are sometimes spotty, XP chipsets in particular. I did that both for gaming at home and high-memory load computation at work.

Your supermarket computer is broken, that's that and it's all to it.

AGP has a different expectation.

Uh, what?

It is therefore unsurprising that changing the specific profile of the memory load can change the stability of the system, and do so in on-again, off-again manner, not a deterministic "as soon as you do this, it will break" manner.

This problem was acknowledged by AMD soon after Athlons and AGP starting having to work together, specifically under Windows 2000 and Linux. Patches were made for both. The Linux one was specifically tagged as an immediate and partial solution, because it basically hacked around the problem (most dynamic re-allocations would not hit the video bits etc). If they fixed it permanently in a foolproof manner under XP, then bully for them. Since I have these symptoms however, and I am not remotely the only one reporting them, I sincerely doubt they did so. [/QB]

Links, references? Surely you can point me to that Linux patch.

[Redwolf]

Originally posted by JasonC:

I have an Athlon notebook too, CM is entirely stable on it. But then it isn't connected to a 19 inch LCD with 8 ms refresh through an AGP slot, is it?

Actually, it is certain that you have an AGP video card in thaat notebook.

You don't seem to realize that the computer side of the video card does never seen and is not influenced by the 8ms reaction time of the LCD.

[Schrullenhaft]

Memtest86.

You can create either a bootable diskette or CDROM (via the ISO images). You'll probably want to download the pre-compiled one for Windows/DOS (if you get the diskette installer).

Depending on your system's speed and the amount of memory you have, it may take a little while to go through one run of the test. I'd let it sit and run for a least a few hours to see if any errors (especially heat-related ones) pop-up. Many hardware errors may show up pretty quick.

The ASUS K8V SE Deluxe probably has good capacitors since it isn't too old, but it may be worth checking them. What you want to look for are the electrolytic capacitors that look like little cans (often green, black or purple in color). Look at the little 'x-mark' on the top and see if it is raised beyond being completely flat. If it is raised, then it means that some electrolyte has leaked internally and compromised some of the capacitance (or more technically, increased the 'capacitive reactance'). A capacitor has really gone bad when it has leaked, often near the top, with the electrolyte being a tannish, crusty substance.

Bad electrolytic capacitors can be replaced, though care must be taken not to damage the traces of the motherboard when desoldering and resoldering the replacements (something you don't want to do unless you're familiar with electronics soldering). The replacement caps just have to be the same voltage, capacitance and thermal rating (which are almost invariably 105°C caps when it comes to motherboards and computer electronics - whereas most electronics stores typically stock the 85°C rated caps).

[ May 31, 2006, 08:49 AM: Message edited by: Schrullenhaft ]

[Redwolf]

"mprime" (sometimes called prime95) and "superpi" are also essential test programs.

superpi tests the memory/CPU interface much harder than memtest, but doesn't test all cells.

Mprime causes about the maximum imaginable CPU load and is good for detecting general stability problems around heat and PSU. But since it doesn't use the graphics card it is not good for overall game simulation.

Both are essential for testing computers that are either overclocked or cheap.

And the original poster needs to answer my question whether the problem disappear with the case open (my question wasn't directed at people who don't even know what mainboard they have).

[Jack Carr]

Probably stating the obvious here. Check your BIOS settings. I had stability problems when I raised a particular setting from Aggresive to Turbo, the clueless overclocker's setting.

The only other thing I tried was re-seating the processor using arctic silver paste. It's supposed to help the heat a bit.

[foamy]

Thanks to everyone for the advice and tips.

I removed, cleaned, re-seated all the obvious components including memory.

I did not notice anything bad, there was however a lot of dust on the fans (Two case fans, one Power Suply fan, CPU fan and Vid card fan).

Got "probe" and reads CPU = 50C, MB = 36C with my case open. (I should do this again with case closed).

Cool+Quite does not show anything other than CPU speed at 2400 Mhz, I can't get it to do anything else?

So far I have not had any lock-ups since the clean up, but I have also had the case open.

I'll button up now I think and see what happens over the next few days.

[foamy]

With the case closed CPU went up a bit 50 - 56C, and MB 36 - 38C. This seems very reasonable to me, Probe has logged a couple of days temp at 54 - 56C.

The freeze-up problem has not occured at all since I cleaned out the dust etc. I doubt this was the cause but I will never know ?? Intermitant faults are a real nuisance.

Anyway, all is well so no more posts from me - Thanks again to you all for the advice and comments.