Strange crash that I cannot diagnose...

Akucaen

Limp Gawd
Joined
Apr 11, 2004
Messages
269
Okay, so. I've been trying to track down a problem with my system since I got back. The problem is that my computer will just lock up hard. I get no bluescreen, no warning, nothing. The system locks up - OR - My open game crashes to the desktop with any error from a AppHang to AppCrash2 according to Vista. This is occuring in a number of games, and has shown itself in all of them I've reinstalled so far - GTA: San Andreas, Hegemonia, Hegemonia The Solon Heritage, Halo 2, Final Fantasy XI, Call of Duty 4, and I'm scared shitless to reinstall Crysis and see.

My specs are: OCZ StealthXStream 600Watt, Core2 Duo 2.66/1333, 2Gb DDR2 1066 (at speed), Radeon 3870, X-Fi Fatality, DFI BloodIron T2RL with the ICH9R southbridge, and two WD 4000 drives. - It's important to note the system is running stock speeds.

Now, I'm having a hard time figuring out where the fault may lie. I've been running a Sandra burn in for the last half hour with no problems whatsoever testing CPU & memory. The CCC testing of my Radeon shows no problems at stock speeds. (I'm not sure how 'valid' that is, but I know the utility looks for abnormal rendering, which seems good enough.)

My PSU lines read (at idle while typing this) as: +3.3 - 3.23v, +5 - 4.81v, & +12 - 11.98v. Looking at this SiSandra readout, an abnormal looking number to me is the -12v line, sitting at 10v. I don't know if that matters or not. (And I know the PSU is receiving (relatively) clean power since it's behind a UPS/Surge unit.)

The only thing I can think of is my ram, which may need a voltage boost. The packaging says DDR2-1066 at 2.25 volts, but the SPD2 column in CPUID reads it can be between 2.1-2.4 volts. I'm running it precisely at 2.25 in the BIOS, and it reads that from there as well. I'd point to my Radeon, and that might be the culprit, seeing as the latest Catalyst drivers actually toned down the hardlocks (mostly) into application crashes. (Is this part of the driver problems I read in an article a while back how nVidia drivers caused the most Vista problems, followed by MS or ATi?)
It's only when I'm playing a game. At the desktop it never happens, and this hardlock/crash is *oddly rare*. I'll note that it's in CPU & GPU intensive games, which is why I'm reluctant to blame the Radeon and call it a day.

For example: (And none of these are reproduced at will.)
In San Andreas - I flipped the camera to look behind me during a chase, I let go of the button and boom, maybe 10 frames later, crash.
In Final Fantasy XI - I went out to a colibri camp and the entire screen glitched. Everything turned into clear-cut boxes and no proper image could be made out. It looks like all the textures were vomited on the screen into 20x20 sized 'pixels'. It hasn't done this, or crashed to the desktop, in a few days now.
In CoD 4 - Just appcrashed to the desktop randomly.
In Hegemonia (and expansion) - Completely random. Sometimes I'd just be looking at the overall solar system and it's crash to the desktop, sometimes I would zoom in on a major battle, get about halfway through, and then bam, desktop.

I'm using SiSandra to 'burn in' my system looking for stresspoints, and the CPU & RAM appear to be passing. Can anyone recommend something else for stress-testing, say, the GPU in a controlled test? I'm on dialup at the moment for the next week, so 3dMark or the likes are a little... out there to get.

Any ideas? At all? (About to try Crysis)
 
CPU(s) are at 45C under load. Using the stock cooler at the moment. GPU peaks at 77.

Update - Installed the latest Intel Chipset drivers, then Crysis. Crysis defaults to 1024*768*Medium on my system, set it to 1280*1024*High and played for 30 minutes, no crashes. Trying other games again.
 
Haha, it's a great game, and my favorite space-RTS by far. I have my complaints, but damn does it get close to my idea of an ideal space RTS.

Anyway, Crysis was blind luck earlier. My computer insta-rebooted without a bluescreen during C&C3 Kane's Wrath earlier. After I restarted I played through that mission, the next, then did 2 skirmish games with no crash... and if anything, the later games were MORE intensive.

I did bump my ram up .25 volts, since the BIOS reported them being right under 2.25 (2.2) just in case, but that was since the reboot.

So, I rule out the Radeon being the cause. Graphic intensity doesn't seem to kill my PC, because usually after a crash, I try and make things *worse* GPU wise, and no dice on crashes, locks, or reboots. This is not producible at will. And that just confuses me more. CPU/memory/GPU benchmarks don't crash the system, but games do - and that has no relation to intensity of said graphics. My PSU lines are strong, and I'm not having any other abnormal behavior from it. (Cue Antec line of dying PSUs that won't shut off when you hit shutdown.)

...Any ideas? :\
 
excuse my laziness, I didn't want to fully read your post, but I do have a question, what are your memory timings, and did you adjust them manually instead of having SPD set to on?

hard-locks like that are usually due to bad memory timings or bios settings not working, or bad memory.
 
It's a legit question. My timings are: 5-5-5-18 2T and 533mhz (2.25v), and the EPP2 profile lists 533mhz profile as 5-5-5-18 2T with 2.25v.

Now, I'm reporting that via CPUID. I lost touch with RAM a looooong time ago when CAS/RAS-CAS/RAS precharge/tRAS stopped being the only numbers, and the memory timings page looked like a goddamn encyclopedia with blanks for the information.

Although... you know... that could be it. Before I left and traveled back home, I finally put 2 & 2 together and saw the dramatic voltage increase needed to run this ram at advertised speeds. (It'd do DDR800 at 1.8v all day long without a care, but DDR800 at 2.25v lets me lower the timings to 4-4-4-12 1T from 5-5-5-23 at 1.8v, go figure. I put the voltage at 2.25 and ran with it at speed. Now, I know that sounds ignorant, but! The bios still showed me the numbers I wanted to see: 5-5-5-18 2T.

The catch though? I was running my system fine without a single error or crash (granted that was in XP-x64)... and I ran it... here too... without any crashes until I upgraded to VISTA.

You think Vista is that bitchy about things? I remember I had an original, launch Creative X-Fi that had some strange IRQ allocation error. It would list itself 10 times in the IRQ listing during post, and XP32/64 never had a problem with it. When I installed Vista release, it would bluescreen immediately on booting windows setup, but if I took the card out, it was happy to continue. I even tried putting the card back in after Vista was done and installed, but it would bluescreen the moment that green loading-bar appeared. A new X-Fi fixed the problem and the old, bugged one sits in my closet.

I'll reboot and run my computer at 800/stock 1.8 and see how that goes tomorrow and post accordingly. But let me ask another question in the meantime: I only have those 4 numbers and the command rate to go on for 1066 speeds aside from autodetection. How do I fill in the other 8-12 numbers the ram page in my bios wants?

Don't say downgrade, please. I'm one of the (apparent) few that likes Vista (sans UAC,) but could that really be it?
 
what kind of memory do you have? brand and model number? if you load default values for your bios, getting into the memory section to turn spd off and set the timing manually, 5-5-5-18-2T should really be the only things your should change, the rest should be left alone for stability purposes.

oh and definitely do not go back to XP, vista is being picky for a reason. If your system is not stable in Vista, but is stable in XP, it's best to get that sucker working fine with vista, because xp is being friendly , while vista is telling you something is definitely not configured right.
 
It's a Mushkin paired set of 1Gb modules, part number 991578 (996578)

And I can manually time everything well beyond what I know about ram.

For example, this is my bios page, everything on auto right now:
CAS Latency (tCL): Auto [5]
RAS to CAS (tRCD): Auto [5]
RAD Precharge (tRP): Auto [5]
Precharge Delay (tRAS): Auto [18]
All Precharge to ACT: Auto
Ref to ACT Delay (tRFC): Auto (tRC numbers exist in JEDEC timings, but not in EPP1/2 timings, those are replaced with Command Rate 2T or 1T)
Performance Level (read delay): Auto
----------------
Read Delay Phase Adjust
--------------
MCH ODT Latency: Auto
Write to PRE Delay (tWR): Auto
Rank Write to Read (tWTR): Auto
ACT to ACT (tRRD): Auto
Read to Write Delay (tRDWR): Auto
Ranks write to write (tWRWR): Auto
Ranks Read to Read (tRDRD): Auto
Ranks Weite to Read (tWRRD): Auto
Read CASE# Recharge (tRTP): Auto
All Pre to Refresh: Auto

I filled in what I know in brackets. Are you saying I just need to fill that in and leave everything on auto? There's a ton of data that won't be filled in. Additionally, the stock numbers coming up now under JEDEC2 (400mhz, 5-5-5-18-(tRC23)). Either the bios made a lucky guess or it read it correctly, but it's set to standard DDR2 400 at 1.8v right now.

I'm not saying you're wrong at all, but my BIOS seems to be getting the timings right no matter which one I do. :\
-OR that CPUID is wrong.
 
I havn't seen a 4 gig kit of 5-5-5-18 ever. Almost everything now is capible of a JEDEC spec 1.8V DDR2 800 5-5-5-15. I suggest you revert your ram to that speed. 1.8V, 5-5-5-15 (unless spicifically otherwise mentioned) and 400MHz.

and yeah, I havn't had to work with a creative sound card in a while but their vista support is, um, lacking. I suggest you try using the onboard sound, an easy test right?

Also that StealthXStream is nastier then you think. I thought Jonny did a review, but I guess not as I cant find it. I coulda sworn I saw an O-scope shot of it pushing out 150mV. Anyways, heres the Hardware canucks review, but they still dont have a sunmoon deadloader, so they went with an SLI system as their gas-guzzler in this review. I hold it against those guys however, for reviewing a PSU when they arn't even able to draw half its rated power... All of this is to say that It could be that power supply is bad.

edit: eh, heres jonnys o-scope shot of the 12V on the GameXstream (one step UP from the StealthXstream) when loading the two of the four 12V rails to 11A, and the other two to 12A.
Test6-12V2-1.jpg

150mV. das sis nicht good yaa.....

When these games crash, just how unresponsive is the system? hit numlock for me during one of the crashes, does the numlock light on your keyboard turn on and off?

From what you've said so far it sounds like maybe you've got a bad API (from the sound card dealings perhalps), in which case I suggest re-formatting windows.

Download ATI tool and Rivatuner. Run ATI tool's "3d mode" or the "scan for artifacts" mode. Does it crash (or find artifacts? have you ever seen any artifacts?). If yes, run riva tuner and underlcock the card, how about now?
 
It's not a 4Gb kit, but a 2Gb kit. Two 1Gb paired modules. These modules will do JEDEC400 at 5-5-5-18 1.8v or EPP400 at 4-4-4-12 2.1v (EPP1 is 533 5-5-5-18 2.25v). They'll do superior to JEDEC spec numbers at 400 with a little voltage. Mind you, this is what the modules themselves report.

This creative card hasn't given me any issues at all, and oddly enough, aside from having to manually tune the drivers via an obscure applet deep in the control panel, it's been oddly lovely to use. If anything, the XP drivers gave me more of a hassle. Who would've thought a software soundstack would 'fix' everything, heh.

I bought this PSU over a year ago, and I did it based on a review. Where, I cannot remember, but I was on a $100~ budget and this seemed like the best of the lot.

As for the crashes, it depends. Sometimes they soft-crash to the desktop and everything is a-ok from there. Sometimes the computer hardlocks entirely and the screen glitches, resulting in multicolored vomit coming out of my LCD, and other times it just restarts with no warning. The soft-crashes are FAR more frequent. And with that said, the crashes are rare, once every few hours if that.

Also, it's not this sound card. The one I was talking about earlier hasn't even been in this motherboard, much less used in this installation.

I'll grab ATI tools right now and do that.
EDIT - 5 minutes, not one error via ATI tools
 
Update, the ram timings appear to be bad? I don't know. I've been playing C&C Kane's Wrath for a few hours now with no crashes, hardlocks, or reboots.
 
Update, the ram timings appear to be bad? I don't know. I've been playing C&C Kane's Wrath for a few hours now with no crashes, hardlocks, or reboots.

Nice.

Keep testing for stability, but I wouldn't be surprised if the memory timing settings, were the culprit to your problems.
 
But I'm back at DDR800 and I'd like to use my ram at spec. Should I loosen the timings a bit or give it a little more voltage? I'm not sure what to do, hence why I posted that entire bios page info in an earlier post.
 
I would leave your ram at no more than 2.2v, I've had many dead sticks of ddr2 memory after giving it more than 2.2v, so I would start there, then keep testing.
 
Back
Top