x1950 pro crashes

[s4fun]

Hello All,

Anyone experiencing random lockup with the x1950 pro AGP?

Given that the absence of evidence is not the same as evidence of absence, or perhaps I'm delusional, all I can say is that I think I got thing stabilized more or less, where it does NOT crash in 3dMark03/05/06, Call of Duty 2, Half Life 2, Empire At War, GTR 2, and many other 3D apps. Or at least it is not obvious that they are crashing or corrupting the video.

But what is obvious is that it is crashing quite often with Company of Heroes. If it doesn't crash it gets corrupted shadows, or textures on the screen that says bad splat. I don't know if this helps or confuses the matter, but similar behavior have been observed on my x850 xt too. Alternatively, the nVidia 6800gt or 7600gs seem to be orders of magnitude more stable.

Is it the card or is it that THQ/CoH has not done adequate testing with ATI card? Or is it that ATI drivers are having issues? Can anyone confirm or deny any of this?

Thanks.

 

[Old Bear]

Could be the drivers. Someone say that 7.1 CCC is bad so I use old.

 

[s4fun]

I've tried a number of drivers, 6.11, 6.12. 7.1 and the one that came on CD I think it was 6.8 or so. It didn't seem to matter. I've also tried different motherboard chipset drivers and BIOS revisions for the motherboard. I've tried raising the AGP voltage, disabling fast writes too. I leaning to suspecting the application because it doesn't seem to happen when I configure the game to use "low shader model", and if I do "high shader model", if I run the performance test before starting a level/skirmish, I don't crash to desktop 95% time, and 100% it doesn't hard lock. Without running the performance test, I get a hard lock almost 95% of the time as soon as the level starts, I get a blank screen, like it is almost guaranteed. Additionally although it doesn't hard lock on the x850xt, like the x1950 pro does, it nontheless also gets the occassional crash to desktop. And the shadow corruption will still occur after while usually past the 30 minute mark after a lot has been going on. Which leads all leads me to suspect the game itself. I was wondering if anyone else has seen this on any of their ATI video cards?

 

[lemonadesoda]

Other comments on other threads suggest overheating of card... depending on your case temperature... the stock cooler is starting too late. (ie at 35% minimum and not going 100% until too late).

Try using a fan tool to get it 100% permanently. See if it still crashes. If this solves your problem... then you are OK. Its a heat issue that you can manage with a better cooler. (Arctic Accerlero 2)

It the problem persists, RMA the device.

 

[WarEagleAU]

I agree Lemon...its exactly what I was going to say. If heat is the problem (be it on memory or core) then this most likely will happen. While manufacturers try to make defect free cards (they have in house testing) problems can happen with something not being soldered right, wrong polarity, etc. If it dont work with the heat fix...RMA

 

[s4fun]

The x1950 pro agp I got currently is the HIS Iceq3 1950 pro. I had RMA'ed a Powercolor with the Accerlero 2 three weeks ago. I seriously doubt temperature is the problem because with ATItool logging I've only seen a max of 55c and that is using ATItool to peg the thing to max. From the logging while running the games, the temps hover in the mid 42-48c. The system temp stays in the high 20's and the CPU temp stays below 37c. I tried touching the card by hand and most of it feels barely warm to the touch. Of course none the of the fans are running at maximum because I'm trying to run the system quiet while cooling it adequately, so I've been monitoring temps closely.

Here's question from the left field, the HIS x1950pro has the thing factory overclocked, anyone knows how to down clock the thing to standard speeds like 575/1380. I can get the core clock down with ATI-tool no problem, but as soon as I touch the memory clock I get a blank screen, though the system isn't really locked up. This is so that I may explore the possibility that the factory overclock might be a bit too aggressive.

 

[Namslas90]

Your Corsair HX 520 only supplies 18 amps on the 12v PCI-E rail. That card requires a minimum of 24 amps on the 12v PCI-E rail. Best bet is you are starving the card of power, causing it to shut down during high demand game play.

 

[s4fun]

The HX 520 provide three 12V rails all three are capable of providing 18A each. The PSU spec says that it can provide a total of 40A combined. Is there away to combine rails to provide the power required? I can solder and redo wiring harness if need be. Thanks.

 

[Namslas90]

The HX 520 provide three 12V rails all three are capable of providing 18A each. The PSU spec says that it can provide a total of 40A combined. Is there away to combine rails to provide the power required? I can solder and redo wiring harness if need be. Thanks.

That information is kind of misunderstanding. When the card pulls more than the available amperage; amperage will be automaticly transfered from one of the other rails. The total transfer is usualy only 20%-30& of the second rail total. So, 18 + 30% = 23.4 amps. The only way I know of to combine three rails is to have a card that has 2 power connectors.

 

[Zubasa]

You have to also consider that one of the +12V rails can be dedicated to power the CPU.
So basically you might have only two +12V rails left.

 

[tkpenalty]

Hey man, the AcceleroX2 version was perfectly fine, you should have asked us first. I don't trust corsair for PSUs, because they are inadept with them, I'm not saying the PSU sucks, however not all the amperage will go to your GPU as Zubasa stated. Both cards are not faulty, and Tul isn't going to be happy, constant lockups mean only one thing in certain graphically intensive parts of a 3D application; not enough power is being supplied.

Yes guys, to rub it in, the X1950PRO is not the angel of power consumption. So isn't the 7900GS, but that doesnt come in AGP. I say 450W minimum that ATI stated is for the top-tier 450W PSUs

Btw, combining rails is extremely hazardous and almost impossible, rails arent a "thing" but a collection of electronics. Just get a new PSU, like what I got, cheap, powerful and can run at more than 600W without a sweat

 

[Zubasa]

I just returned my Sapphire X1950 Pro today and they basically test the card and exchanged a new one for me.
Now the new card has much better fan speeds in the bios.
Not bad service after all.

 

[s4fun]

The Corsair HX520 is the 4th power supply I got after RMA-ed the last 3. I tried 600W 30A single rails. 550W 22Ax2 dual rails, etc. Somehow I get the feeling it isn't the PSU.

Futhermore, this what I've found after a bit more searching:
http://forums.relicnews.com/showthread.php?t=129695&highlight=x1950+pro

It seems to suggest that it is the application. It's either that, or I'm gonna be doing the RMA for the x1950 pro agp.

I think the Call of Duty at 1680x1050 with most of the settings on high puts a fairly substantial load on the board, PSU, and the system and it seems to be very stable, not even a stutter bad textures or messed up shadows (at least not anything obvious) even though frame rates are pushed down into 40 fps. It seems Company of Heroes has bugs and THQ has got their work cut out for them.

 

[lufrey]

Hello All,

Anyone experiencing random lockup with the x1950 pro AGP?

Given that the absence of evidence is not the same as evidence of absence, or perhaps I'm delusional, all I can say is that I think I got thing stabilized more or less, where it does NOT crash in 3dMark03/05/06, Call of Duty 2, Half Life 2, Empire At War, GTR 2, and many other 3D apps. Or at least it is not obvious that they are crashing or corrupting the video.

But what is obvious is that it is crashing quite often with Company of Heroes. If it doesn't crash it gets corrupted shadows, or textures on the screen that says bad splat. I don't know if this helps or confuses the matter, but similar behavior have been observed on my x850 xt too. Alternatively, the nVidia 6800gt or 7600gs seem to be orders of magnitude more stable.

Is it the card or is it that THQ/CoH has not done adequate testing with ATI card? Or is it that ATI drivers are having issues? Can anyone confirm or deny any of this?

Thanks.

I've had similar problems with my old 6800, in the end, the trouble ended up being overheating. The stock cooler was crap, and the temp often exceeded 115C. Replaced the stock cooler with a Zalman VF900-CU, and all the problems went away.

 

[s4fun]

I wish it was as simple as overheating. I have some experience and some cheezy hack job techniques with fans and aluminum ducting that can almost always guarantee to lower video card temperatures. Heck I can duct some -10F from the outside right now. As it stands, even without running the fans at max, the x1950 pro is running at 35C idle, 47C normal application load, and 58C with ATItool stress test. The whole system just air cooled is at 22-27c and CPU is running from 24c-37c. I can put thermal probe on the thing, but in general just touching it by hand, it feels barely warm.

 

[s4fun]

As for the Powercolor x1950 pro AGP with the Accerlero 2, they did NOT provide adequate cooling for the VRM. The VRMs get real hot real quick and the thing becomes unstable. I've heard recently that the Arctic Cooling and Powercolor/Tul updated the Accelero 2 with VRM heatsinks.

 

[deftonesmx17]

I had a similar issue. At first it was only COH that would crash, other games worked fine. After some time, every game started doing it or should I say anything that was in 3D mode would crash the vidoe card only(VPU recover), but windows never crashed, the system would not crash in a 2D mode. So, I bought a new video card thinking it was obvious my card was fudged. Same issues with new card as my old card. I tried everything from drivers, making sure I had nothing overclocked through my BIOS, and to a clean install of windws. Nothing worked. What did I find out in the end................my motherboard chipset was overheating when I switched to 3D. I assume it had to do with the extra load that was put on the graphics port (PCIe port in my case, AGP port in your case).

 

[InfDamarvel]

Power Color has released a vrm cooling unit for the X1950pro pci-e and it can be brought on ebay or recieved for free if you already have a X1950pro with one on it I believe.

Just check their site lol.

 

[Completely Bonkers]

Are you overclocking your mb?

The rialto PCIe to AGP chip HATES, and I mean HATES, being overclocked.

Go back to stock clocks and check again before you go bonkers.

 

[s4fun]

I am not overclocking a single thing. The HIS Iceq3 x1950 pro (which I got after I RMA-ed the Powercolor one and got this one in exchange)comes factory overclocked, so its standard is at 620/1480 and that is the lowest I can get it to go, and I definitely am NOT going to try overclocking that. Anyone know how I can get that thing to run a ATI standard clock of 575/1380 which is where the Powercolor board came standard with.

I make sure the AGP port is locked in at 66mhz, and I have the CPU run at stock clocks. What other clocks do I need to verify?

 

[Completely Bonkers]

The important bit is the AGP port. Just make sure that's at 66Mhz. But if the mainboard and AGP is NOT overclocked... then the rialto chip isnt the problem...

You seem to be very unlucky so far. I hope you get some luck to solve it.

 

[mullered07]

if youve gone through 3 psu's and this is your 2nd x1950pro maybe theres a fault with your motherboard, have you tried any other graphics cards? tbh to go through 3 psu's and 2 x1950s and still have the same problem then that says to me its neither of them 2

 

[s4fun]

I upgraded from an nVidia 6800gt. I plug back the 6800gt for a couple of weeks after the first x1950 pro was being RMA-ed thru the mail. The system is rock solid as it has been for 2+ years with the 6800gt. Other games seemed to be stable as far as I can tell with the x1950 pro. Perhaps they don't stress it enough, but I got think Call of Duty 2 at 1680x1050 with most settings on high should be plenty demanding.

What would help achieve a peace of mind would be some tool down clock (undo the factory overclock) of the HIS x1950 pro to the standard ATI "reference" settings, and something to test the memory on the video card to see if there is any thing wrong with them at standard settings. Perhaps the factory overclock might just be a bit over the edge even though the temperatures are low and the cooling seems to be well managed.