Vcore flutter normal?

[sdrac]

Question on Vcore readings - should they remain rock solid or is it ok to see variation on it?

I have an older system: A8V deluxe, Fx-60 (@stock speeds), 1G Ram,
2 73GB Raptors in Raid0, x1950Pro 512MB (@stock speeds), 750W Thermaltake Toughpower PSU. Win XP Professional SP2.

Anyway, all the rails look rock solid, but I notice Vcore fluttering frequently between 1.39V and 1.374. Not a big difference but it happens all the time.
I'm having some other issues with the system and I'm wondering if this could be the cause (or, at least, if its anything to be concerned about).
I have Cool&Quiet shut off.

Thx,
Sdrac

 

[mitch074]

that's normal - sensors aren't THAT precise, and there is usually a 5% tolerance by design.

A variation by 0.016V is nothing to be worried about, except if it started recently (be aware though that a BIOS upgrade could make sensors show reeatic results while nothing has changed in reality)

Describe your other issues.

(note to self: 1024th post! Geez...)

 

[sdrac]

Mitch,
The problems have been with playing either Oblivion or Med TW 2.
They started about a month to 6 wks back after playing for many months with no issues. What first started happening was I'd get blue screened - claimed there was a infininte loop in the ATI driver. ATI didn't have any record of it and I tried reinstalling the driver and upgrading to the later version - to no avial.
Then I started Crashing to the Desktop from the game. Sometimes cleanly, sometimes with an error report being generated claiming there was an error in Oblivion. I tried removing and reinstalling the game and the latest patches - no joy.
I ran memtest-86 for 12 hours without error. I then ran Prime95 for nearly 24 hours with no errors. So I'm assuming my memory and cpu are ok.
I figured that left either the PSU (which originally was a 2.5 year old Ultra unit), the Mobo or the graphics card itself. I replaced the PSU with the Thermatake Toughpower and still no go.
The crashes to desktop were getting more and more frequent. My prime suspect was now the graphics card - but I noticed the Vcore flutter. Originally with the Bios setting to Vcore of 1.4V (default) I was seeing the flutter and the voltages shown were pretty low (like 1.342 sometimes). I up'd the Vcore in Bios to 1.425 - crashes haven't gone away completely but they have slowed down to where I can play almost normally.
So now I'm wondering if its the Mobo. There is one later version of Bios for this board - I coulid try that and see if it makes a differnce (but its a beta). But the viedo card is still a suspect too - its running hotter then it used to under load and as long as I'm not doing anything viedo intensive, the system seems fine.
But how do I figure this out? I don't have a spare video card to plug in - nor do I have a spare mobo. Any thoughts??

Thanks,
Sdrac

-- congrats no the 1024 posting! Wow.

 

[mitch074]

Humm... Looking at the mobo's specs...

Try restricting the AGP rate to 4X and play. I had similar reactions on an older system with AGP8X, and this fixed it.

reset your Vcore; you can however see if there is a way to set your AGP voltage (1.5V instead of 0.8V) if the above does indeed fix it; AGP 8X is able to work at either 0.8V or 1.5V, and raising the voltage may have gotten necessary if, say, your mobo's condensers are getting worn out.

 

[Thanatos421]

10:1 you have an overheaing GPU. I get the same errors in Oblivion (used to) until I turned it down to 640x480 and now just barely keep the load temps on the GPU under the shutoff point. It's amazing how quickly my card would reach nearly 90c while playing Oblivion.

Get RivaTuner and monitor you temps while you play. It's nice if you can "borrow" a 2nd monitor for just that purpose.

 

[mitch074]

not really likely: overheating GPU usually results in corrupted graphics before an error ('snow', erratic triangles on-screen...). From cool playing to complete system crash indicates an interface failure more than overheating.

Now it wouldn't hurt to clean the fan and sink of the graphics cards...

Note that I've been told of unstable AGP voltage on this mobo: you should be able to raise the AGP voltage to 1.6 or 1.7V and see what gives.

 

[Thanatos421]

I beg to differ. I have sat and watched my GPU do the exact same thing he described. It heats up to right at 90c and the system shuts down. I think I caught the shutoff temp right at 87c. You can watch the temp monitor and I can tell you exactly when it will shut down.

It occasionally blue screens, but usually the video just dies and then the system reboots. It's heat in my case, which sounds a lot like what he described.

I only say that because of the striking similarities between his system and mine.

I also received the same BSOD message about the ATI loop. I also was able to run Prime 95 stable for more than 24 hours with my Opty OC'd. If you are telling me it's not heat, then I woud LOVE to know wtf the problem is so that I can play Oblivion at a nice resolution with some decent eye candy. It looks like a damn 8-bit Nintendo game at 640x480.

I'll add one more note that makes me HOPE you are right that it isn't heat. Running ATI Tool, I set it to find the max mem/core clocks. The card heated up to over 90c w/o a hitch, no shutdown, no BSOD. When in Oblivion, it dies at 87c consistently. Don't know if that helps, but it didn't seem to make sense to me at the time. I'm begging you to tell me that it's just the AGP bus not getting enough juice. I'll wet myself if I can actually play Oblivion like it should be played. I had relegated myself to playing in 8-bit mode until I could get a new system.

 

[mitch074]

...and I got the exact same symptoms on a system that ran rock stable when I reduced the AGP rate to 4X - thus making the AGP bus signal less difficult to parse (2 signals on climbing phase, 2 on falling phase, instead of 4 and 4).
I hardly think that there is a 'cut-off' temp at 90°C on this video card; these pieces of machinery would sometimes run at up to 110°C (which is too high; 90°C isn't though).

Try upping the AGP voltage or reduce the AGP rate to 4X. Dust your card!

 

[Thanatos421]

My card is spotless, as is the rest of my system. I'll try the AGP 4x trick, or the bus voltage adjustment tonight. If that fixes it, then you need to give me your address so I can send you some cookies or something. It just always seemed like heat when I could time the shutdown to a certain temperature.

If this fixes my problems, then you sir, are the man.

BTW - Where were you when I originally posted this problem

 

[sdrac]

Mitch, Thanatos,
Thanks for the inputs.
Its not a heat issue - at least as far as I can tell. I've had ATI Tray Tools on my system for a while and I was monitoring temps with it. Until recently I never saw it go over 50C while playing (Just recently I got up to about 55C which I might be incorrectly linking to these problems). My X1950 Pro has an Artic cooling Accelero X2 on it (with the heat spreader for the voltage regulator) and its kept it pretty cool under the highest loads.
I'll try up'ing the voltage on the AGP to 1.6V and if that doesn't work I'll try 4X too. Let you know how it goes.

Thx,
Sdrac

 

[mitch074]

I was off dusting my own system - a system isn't clean as long as you can't see yourself in your heatsink's fins

 

[Thanatos421]

OK, neither option worked for me

I tried it in combinations of AGP 8X/1.5, AGP 8X/1.6, AGP 4X/1.5 and AGP 4X/1.6

Each time, in Oblivion again, when the card hit about 87c, my monitor goes into standby, and the system reboots.

What driver suite should I have installed. I currently have Catalyst Control Center/ATI Tool/Riva Tuner all installed and in system tray.

I have a sneaking suspicion that what's causing my problem is the CCC VPU Recovery option. However, this says that it will simply reset the GPU if the driver stops communicating with the device. Shouldn't make a difference should it?

Also, what about AGP fast writes? I remember years ago, that setting could cause problems. Everything is enabled in CCC SmartGart: AGP Fast Writes/AGP Write/AGP Read and PCI Write/PCI Read all enabled. I haven't messed with the CCC options much, as I thought that once you installed ATI Tool, it took over operation of the card.

Also, this x1950xt came with the TEC cooler on it. I don't know if this card's hardware had some kind of call hard wired into it that would be some kind of "trigger" for the TEC to come on. Now that I have an aftermarket cooler on it, since the TEC isn't there, is it possible that the card is calling for extra voltage at a certain temp, causing some problem?

I don't know wtf is up with my system anymore. This only happens playing Oblivion at high(er than 640x480) resolutions.

I'm stumped... Hopefully, sdrac's problem can be fixed with the suggestions. I wish you the best of luck. Sorry if I kind of hijacked your thread bud, but our problems looked so similar.

 

[systemlord]

Oblivion is very hard on most G-cards even mine and its PCI-E. Crashes to desktop were the norm back when I played Oblivion. The Oblivion graphics engine relies on a least 400mb onboard VRam @ 1280x1024 with most setting set to medium. I bet if you upgraded to a new mainboard & G-card it wouldn't happen anymore. AGP running Oblivion I never thought possible.

 

[Thanatos421]

Hehe, yea, it barely runs at 640x480, with the only additional setting being bloom. I guess I'll have to play it "old school" style til the Phenom/Penryn battle is decided. I'm not wasting any more money on this old system.

 

[sdrac]

I don't know - prior to these issues starting for me I was running smooth at reasonably high settings: 1280x1024 (best my LCD can do), 6AA, 16AF set via ATI Tray Tool, and most settings in side Oblivion set to medium high or even high. Looked great and rarely stuttered.
I still have those settings now - but I'm having these problems now. I could back them off and see if things improve, but it used to work fine - something had to change.
Anyhow, I up'd the AGP voltage to 1.6V. It ran better - similar to what I saw when I up'd the Vcore. I played once for about an hour with no CTD's at all. Stopped for a bit, then played again for about 40 more min and had a CTD.
The CTD's are werid though - they don't normally occur during alot of action - in fact most seem to happen when Im going through my inventory. Werid.
I thought maybe it was the HD's kicking in originally, but I have them on a different 12V rail now so it shouldn't be that (at least what I think is a different rail).
Anyway, I'll try 4x next with both voltages and see what happens.

By the way Thanatos, no problem on the highjack. Our systems are pretty similar - hopefuly one of us will get the problem fixed. IMHO, your issue is with whatever aftermarket cooler your using - if its allowing you to get up to 90C, I'd say you shoud look for another.

I'll drop a note after the next round of tests.

Sdrac

 

[Thanatos421]

It's the Zalman Fatality GPU cooler, with the fan speed dial, which I leave wide open all the time.

I might have to take a look at the one you have, might be my voltage regulators taking a dump. If I can save this card for 50 bucks, I just may cave.

 

[mitch074]

the GPU Recovery option led me to more crashes than I can mention before I lowered the AGP speed. Fast Writes don't give much of a boost, disabling them may help.

Now if what I said about the AGP voltage/speed didn't help, then it may indeed be a bad power feed - try another PSU.

 

[sdrac]

So, by changing the interface to 4x, things seemed to improve a bit.
First I tried 4x/1.5V on the AGP interface. I played quite awhile until it finally CTD's with an error report. Seemed similar to changing the 8x mode to 1.6V.
Then I played with 4x/1.6V - I played a long time and it didn't crash to desktop at all. The only two issues: I had some artifacts during game play - I have *never* had that. And then when I exited out of the game, I pulled an error and generated an error report.
For chuckles I ran some benchmarks to see if using the 4x mode would affect the system greatly. No real change in 3Dmark05 or 3dMark03 (I never loaded 3Dmark06). So I guess thats good.
One strange thing I noticed though - in the info screen of either, it said that my card had 1G of video ram and texture memory. The card is only 512MB - but I do have 1G of regular RAM. Does anyone know if this number reflects the card RAM or the system RAM?
Overall I still think its the card - I think something has gone bad on it. I'm going see if I can RMA it if its not to late (I bought it back in January).

Mitch - thx for the advice on the 4x. I think I can live with this if I can't get the card RMA'd. I don't want to buy another for this old system -so its good enough for now!

Sdrac

 

[liquidx]

I doubt you will be able to RMA it, but definitely turn that in on warranty. Many cards these days have a lifetime warranty, and under what your describing, it may very well be the card.

As for Than, when you installed the new cooler, you did put thermal paste on it right? It seriously sounds like your gpu core is not touching the heatsink right. Even if you did use thermal paste, you might want to pull it back off, clean it, repaste, and then reseat. You never know a 20 min job, could get you back in business. Those cards are too nice of cards to leave overheating like that.

 

[Thanatos421]

Yea, I applied AS5 to it. Although, I did spread it VERY thin. I may try that and be a little more liberal this time, just to see. I hate to ditch this card just yet. When I bought it, I was hoping it would hold me over until Penryn vs Barcy could be decided BY FACTS (don't turn this thread into AMD vs Intel people!). Now I just hope it doesn't die on me. I may purchase a new cooler as well. The Zalman *should* be plenty to cool the GPU, but it has nothing to cool the VRM. I don't know if that's what's causing my problems, but it probably is, knowing my luck and my quirky computers

 

[sdrac]

Thanatos,
I'd definetly get something to cool the VRM. On the x1950pro series, without the heat spreader the card runs real hot - similiar to what your reporting. I'm not sure if the layout of the xt - does it have two molex connectors at the top edge of the rear of the card with the VRM right next to them? If so, you may be able to use the same heat spreader. You can buy them from arctic for like $1.00 or so.

 

[mitch074]

Try making the warranty work, yes. Those cards usually carry year-long warranties at the very least, use it.
If it has started showing off artifacts, something is quite wrong with it: I'd say a leaky capacitor.