the thingie that controls the memory stuff, does it control the video stuff too?

[iroc409]

just curious. i've been getting some new errors lately with video. i reinstalled bf1942 the other day for a change of pace. it is having a _ton_ of issues. sometimes i'm getting "artifacts" or what-not where terrain and items just dissappear at certain angles. and it dies a lot.

finally the other day the game crashed and a little ati box came up and said "vpu reset because it no longer responded blah blah". never had that before, but just a few minutes ago i was playing some dod, and dod crashed again (as per usual). anyways, i got the ati error box thingie, which has never happened with dod. it usually just crashes (this time it didn't).

i'm starting to wonder if maybe my video card isn't at fault due to all these video errors. but, then i think maybe if the northbridge (that's the thingie that runs the memory.. right?) is faulty, it could cause the card not to respond through the mainboard, and the video drivers think the card is at fault.

i guess the main question is, is it possible to get memory errors in windows (page fault errors, that kinda thing) from the video card (or the video memory), or is that specific to the system memory?

i'm assuming the northbridge controls the system bus, but i figured i'd ask you folks.

[airspirit]

If you're using a board with a non-locked PCI bus (IE: NForce2) and you are overclocking then your AGP bus is twice the PCI bus. This can cause you grief with many ATi cards. The AGP bus hangs off of the Northbridge, so any errors with it could potentially cause AGP bus errors, though you would typically see more widespread errors than just AGP related ones. I would test your card in another computer to see what happens.

[iroc409]

i'm also getting bsod's, which mostly seem to have memory-related errors, like page file faults, that sorta thing (i'm assuming those are all related to system memory).

i've used memtest to test my system memory, and it came up with no errors.

all of my drivers, including the 4in1's and the bios are fresh (again) as of yesterday.

i talked to a local computer store guy i kinda know, and he seems to think it's a problem with the video card getting "out of synch" with the system, so the system burps to re-synch. i was thinking about manually setting the card down to 4x or something, but i'd rather not do that.

oh yeah, it's not an nforce, it's a via kt400/vt8235 abit (at7-max2).

i've tried it not-overclocked before, but i'll turn it back down again see what happens.

i know i just need to test everything seperately, but i don't have another machine at the moment to try it in . i will probably be buying a new one soon anyways, so if i don't get it resolved by then, of course i'll start swapping things out.

[airspirit]

Find a friend that will let you crack their case open and try the card in there. That will give you the confirmation one way or another.

It sounds like a motherboard instability issue to me.

[Tempus]

The VPU recovery thing is an issue with the new ATI drivers (3.8s and 3.9s) you can enable/disable in the video options but the default is enabled.

Basically, if there is an error that should have resulting in a lockup from video, this is supposed to recover to desktop.


I (and others) have had much better luck using the older Cat 3.5s. You might want to try that.

You didn't give details on your Overclock so I'll just assuming you pumped the FSB and left the videocard gpu and memory timings alone.

If you changed the mobo FSB and didn't have the AGB frequency fixed at 66 then you will have some issues.

You didn't mention your specific video card. Is it a 9600? I've got a 9600 that acts up badly under the 3.7, 3.8 and 3.9 drivers, even at stock speeds (both video card and mobo stock.) It works fine under the Cat 3.5s, so again, drivers are key.

I can try to help more if you get some specifics down. What exact OC are you doing. What voltage settings do you have. What cooling are you using and what temps are reported.

[iroc409]

the video card is a 9700 pro 128mb, abit at7-max2 board, 2400+ processor.

the bios i just flashed to 'EB', via 4in1 4.49. i'm pretty sure i am using catalyst 3.9, although i was using older drives (just upgraded the other day), with the same problem.

the overclock i was running was just upping the multiplier, at the 2600's settings, stock voltage.

i've set it back to stock on more than one occasion, with no luck. i'm running stock now, and have been since yesterday.

the fsb:agpci settings are 4:2:1.

i did play with the overclock some, got it up to 2400mhz. but it seems it wants 2v to stay happy to run windows (and i still get a few bsods on startup), 1.9v posts fine but xp won't boot.

at stock voltages, it idles around 37deg C, under 2v it idles at 50. i'm using an all-copper zalman flower with the stock 90mm fan. i was using a 120mm nidec, but that was WAY too noisy. i've got an extra 80mm fan on the fan bracket to cool the video.

after i exit a game, like a hl mod, i've checked winbond and it's around 45deg C, running at the 2600 settings. i have no idea load speed at 2.4ghz, but i've only tried desert combat, which it usually crashes after a while anyways. that i'm assuming is either overheat or just the usual crashing.

using anything faster than 2.13ghz, i get a tendency to get a lot of "bad_pool_caller" errors on bootup. it's kindof hit-and-miss, but seems more pronounced if i do bump the fsb (i haven't bumped the fsb more than to like 136, but ran the multiplier up to 18).

i've pretty much eliminated the secondary storage hard drives and controllers being the problem, but haven't really had a way to test the primary controller. it seems to work fine, the only issue is when i try and move the primary hard drive to ata100 mode5 udma, it doesn't seem to want to do it. after restart it just goes back to mode4 ata66. not sure what's up with that, it's the only drive on that particular ide channel (using the via ide controller program thingie).

i ran this vid card in my old system, it was a 2000+ with a biostar m7vkb (i think), and i never had this issue. i was also running cheap pny memory as well. no problems there.

i tried disabling the ati vpu recover. without that, it just crashes anyway



hey airspirit, i notice on the fsb : agp : pci settings, it appears i can change the fsb ratios, between 3-5, but it doesn't look like i can change the agp : pci, it sticks at 2:1. think that's my problem??

[airspirit]

Ugh. PCI = FSB/PCI Multiplier. A 166 FSB at a 5 Multiplier gives you a PCI bus of 33 (stock). AGP always is 2*PCI on those boards. If you ran 200 FSB at a 5 Multiplier (max on your board), your PCI bus is 40 and your AGP is 80 (from 33/66 stock). Things will go wonky at these speeds if your hardware can't handle it. What you are looking at is the FSB/Mem Spd setting, so that you can, for instance, run your bus at 166 and your memory at (166*5)/3 = 276. Or, if you ran a 133 FSB, then (133*5)/3 = 221.

[iroc409]

Originally posted by airspirit
Ugh. PCI = FSB/PCI Multiplier. A 166 FSB at a 5 Multiplier gives you a PCI bus of 33 (stock). AGP always is 2*PCI on those boards. If you ran 200 FSB at a 5 Multiplier (max on your board), your PCI bus is 40 and your AGP is 80 (from 33/66 stock). Things will go wonky at these speeds if your hardware can't handle it. What you are looking at is the FSB/Mem Spd setting, so that you can, for instance, run your bus at 166 and your memory at (166*5)/3 = 276. Or, if you ran a 133 FSB, then (133*5)/3 = 221.

hrm. maybe i'm not following. i thought perhaps the 2:1 ratio of the agp spd vs. pci speed could possibly be causing the trouble. lol, i guess that really doesn't make sense anyways.

i haven't touched the memory speeds or fsb at all... only the multiplier, so i'm running at 133 fsb.

although on the bright side, it seems to be running reasonably stable now with the side cover off at 18x133, around 41 dec C... (doesn't fix the pause tho, lol).

[pHaestus]

The title of this thread made me laugh and I had NO idea what to expect when I entered.

[iroc409]

so i went over to the computer store this morning and picked up a gigabyte k7 triton with the kt600, because i was tired of this and thought i'd try to just swap the mainboard and be done with it.


so i get home, and decide i'm going to try it one more time. i start ripping and tearing stuff apart... and i think it actually works now. holy crap. i've gotten a couple bluescreens since, with the same usual error "IRQL_NOT_LESS_OR_EQUAL". but i'm not having the pauses.

i dunno what i did, i pulled everything out and put in just the vid card. not sure why that's working this time, but i'm gonna go with it i guess.

i'm running on the onboard sound now, but maybe it was the sound card. which is funny, because i pulled that out before and it still had problems. or i was hallucinating.

or, maybe i'm a doofus. who knows. i guess we'll see if it still works through tonight.

it also seems to be running a little smoother as well, and doesn't seem to want to freeze up in the os like usual, which got worse when i ran it to 2.4. but now it seems fine at 2.4, so i might even try to push up the bus again.


edit: well, i still get some _occasional_ bsods from desert combat (bf1942), and it hard locks when i try dod. but at least it seems the random pauses are gone, so maybe i just need a clean install.

[drummingpariah]

you almost definitely need a clean install:
your registry is still set for the old kt400 mobo, which is bad. some of it got updated to the kt600, some didn't, and that's going to cause all kinds of conflicts. don't install anything while you have this conflict, if you plan on keeping whatever you install after the fresh windows install. some of the core files of games you install at this point will become corrupt and give you problems in the future. your best bet is to do a fresh install of nearly everything on your system, starting with your OS, drivers, then i'd say BF1942, and try that out with nothing else going on at all. just try to run it stock, without any BF updates or anything, just right out of your fresh install. let us know how this goes, and if you have any more problems we can take it from there.

[iroc409]

i didn't install the gigabyte board, i just kept mine.

i did a clean install just to be safe, and installed bf1942 with pretty much nothing else installed.

the random pauses are gone... haven't noticed them at all.

but bf1942 (mostly desert combat) is still quite unstable. usually it just crashes to the desktop, but occasionally i get a hard lock, or sometimes a blank bluescreen. i've only gotten 1 or 2 hard locks in just windows, and while working, so while not playing games the system seems more reliable.

i just installed jane's fa18, and it bsod'd. i installed 3dmark03 a little bit ago just for fun, and i've tried running it several times and each time it's either hard locked or bsod'd... but never in the same spot (just "somewhere" during the run).

so, obviously something is still awry, and it makes it rather difficult to play games. desert combat seems to be much less reliable for online play, if i play an instant battle it seems to play (generally) longer.

i think i'm just going to have to buy new stuff and figure it out piece by piece, or something. i dunno. maybe it's a good excuse to get a 9800. bleh.

[Tempus]

Here's my check list:


1> Verify that the CPU, mobo, and ram are happy. Run PRIME95 for 12-24 hrs in torture test mode. If it hangs at all then you know you've got some issues there.

Given the errors you are posting, I'm willing to bet you will not pass PRIME95. Further, I'm guess its a Ram / ram timings issues.

Try pushing your ram voltage to 3.6 or 3.7. That might help alot.

2> Once you get PRIME95 stable (I call 24 hrs stable.) Then go and run 3DMark03 looping for a few hrs. If it crashes here then its either a bad video card, or a poorly cooled on. Try upping the volts to it, reducing the timings, and checking the heatsink on it. Maybe the thermal paste has dried or whatnot. Its also possible that you messed up the card (fried some transistors) when you OCed it before. Running the AGP bus at anything over 66Mhz (meaning if you overclocked your non-locked AGP when you upped the FSB) then it might be bad.

If it passes Prime95 AND 3DM03 then I'm willing to say that your hardware is solid and stable enough to game on.

At the point, if you are crashing in games (dod, DC, CoD) then you need to do a clean install and do it right. Get the OS patches, then directX, then mobo drivers, then videocard drivers. Reboot often and test it a few times during the process.

Then, once it is happy, make a Ghost boot disk and be safe.

[iroc409]

hrm, i was just looking through the bios some more, and just noticed something.

i'm pretty sure this memory is cas2, but i've got it set 'by spd' in the bios, and it seems to be picking it up as cas2.5. perhaps that might cause some of the problems, i think i'll try setting it manually to cas2 and see what happens.

the memory also looks to be 2.6v, and the bios is set to default at 2.55... it's not much of a difference, but maybe that's causing some issues as well.

the agp bus really hasn't been overclocked at all. for some reason the fsb on this system doesn't want to go up _at all_. i can only really get it to run at 133 or 166, that's pretty much it. i've normally been running the fsb at 133, but maybe i should be running it at 166 to match the memory. the cpu seems to run fine at either 2600 default settings, either 133 or 166, so that shouldn't be a problem.

the memory is 2x512mb kingston hyperx ddr 333 (pc2700).

i did just realize something tho, maybe this could be a part of the problem. my video card may have overheated this summer. i wasn't overclocking the card at all, but a couple months back my a/c broke, and i had an abient temp around 80 deg F. it shut down halflife a couple times before i realized it (but didn't crash). after that, i put a large fan on the system with the side off, and things were about normal. i had been having troubles before then, but mostly with the random pauses, from what i remember.

since then i doubt the vid card would overheat, i've since put an extra 80mm fan over the vid card, and it seems to be taking care of that fairly well.

i suppose something could have broke then. although unless i'm playing games, the system seems pretty stable now.

i'll try pushing up the mem voltage, but i guess it's kind of disconcerting running the memory at +1 volts like that. lol, i guess i'm just paranoid, but that seems like quite a bit, especially when i'm not really pushing the ram any.

[Tempus]

I wouldn't just go randomly tweaking bios setting/timings/voltages yet.

First off, figure out if there IS a memory or cpu issue by running a math torture test. I like prime95, as I said, but there are tons more out there.

You need to start eliminating causes and narrowing the problem field, THEN you start fixing it. Right now you are just shooting blind.

[iroc409]

yeah, i downloaded prime95 version 23.7. i started it up earlier, not sure if i'm running it right, lol.

it was apparently running, but according to the task manager said i was only using like 15% of the processor... but the system was running pretty slow, so it must have been using more. not sure how that goes, but i'll run it for a while and see what happens.

[airspirit]

Could it be that your windows install is borked? You may want to try a fresh install of that ... though I hate recommending it.

[Levish]

he did btw mention above that he did a clean install of windows and bf1942

And I'm assuming you have Directx 9.0b installed as well as all major patches and critical updates?

Have you tried rolling back the CAT drivers to a different version making sure to use ATI's CAT uninstaller?

Have you removed any other hardware from the system (modem's, sound cards, cd drives that aren't needed, hard disks) ...

Are you running a fan controller or are powering the fan's from the PSU or off the Mobo ...

how did the Prime testing go?

sorry for the long list of questions

[Tempus]

make sure you have prime doing a torture test and not just running. That will make a huge difference.

[iroc409]

yeah, it's a new install, and everything should be current as of last thurs. although i think there's a new 4in1 available that just got released like early this week, i haven't installed that yet.

well, now i'm thinking maybe it was some sort of kooky bios setting i was assuming to be harmless (and obviously setting it multiple times) that is causing the problems, or something.

i cleared the cmos again the other day, because i was setting the processor back to stock (for some reason after ocing, sometimes the bios won't accept resetting the processor ... says the cpu has been changed and won't boot).

i tried 3dmark03 again, 'just in case', and it ran. this is after running it a dozen times at different clock speeds (including stock), when it didn't run.

now i've run it anywhere from stock to 2332, using a 166fsb (i haven't tried it overclocking on 133fsb yet, but i'll get to that). it's run through about a dozen times.

so, i guess i dunno if that means it's fixed or not.. lol.

dc still crashes, but it plays much longer, and i haven't had a hard lock. it usually just crashes to desktop.

dx9 is installed, and i ran all the updates when i installed the other day. i haven't tried rolling back the video driver tho, still running 3.9.

before i reinstalled i pulled the sound card and the ata100 controller (running 2 extra hard drives), and moved the hard drives to the hpt374 controller on board. i was having the issue before i installed the ata card, and that's been tested. the pauses were removed with the soundcard (although i get some crackling in dc with the onboard sound, but i'm just guessing that's because the onboard chip isn't exactly stellar).

the cpu fan is running off the mainboard, i do have an inline fan controller on it, but it's turned all the way up.

and, i haven't really done much of the prime testing yet (i did start it as torture test from the file menu), i forgot to run it last i slept. but 3dmark03 is promising.

i guess what i need to do is just go through the bios and start things one at a time, and see what makes it break.