HIS Radeon HD 3850 crashing system in Game

[[SNG]McBain]

I've been working on this problem for about a week .... New card; and new system as follows:

• MSI K9A2 Platinum MB
• AMD Athlon 64 X2 5000+ Black Edition
• ZALMAN CNPS9500AM2 (AM2/754/939/940) AMD CPU Cooler
• HIS Radeon HD 3850 IceQ3 TurboX 512 MB Graphics card
• 2 x 1 GB Ultra DDR2 (667) RAM
• Enermax Liberty 620W Modular Power Supply
• SB Audigy 4 Sound card
• WD 500 GB SATA Hard drive
• 2 x generic DVD writer (one's SATA and one's IDE)
• APEVIA X-Cruiser Metal Case

The problem is that the system crashes while in game. So far I'm blaming this on the graphics card, but I'm open to other suggestions. When the crash occurs, the game will freeze .... then the screen goes black ..... a very loud and high pitch noise comes out of the speakers .... and the monitor displays no signal. The only thing I can do at that point is to reboot the system.

This is what I've tried so far: (and so far nothing has made a difference)

1. Re-installed the drivers
2. Updated all hardware drivers (including AMD's CPU Optimizer)
3. Ran CCC Auto-Tune - that made it worse
4. Un-installed ATI drivers (and used a Reg cleaner), and tried Omega drivers - that also made it worse
5. Ran Everest's System Test - no errors
6. Ran memtest 86 - no errors
7. Installed RivaTuner to adjust the fan control on the graphics card
8. While in game I would periodically minimize the game and check system and graphics card temps - CPU hardly reached 40 C and Vid card never goes over 70 C

I think I even tried an older ATI driver, but I can't remember. I will try that again though just in case.

I'm not sure why this is happening, and anything I've tried doesn't seem to fix it. I could play a game for two hours before the crash, or I could play a game for 10 mins before it crashes - it's not regular. I've read a few forums where people have posted similar problems, but there's been no clear solution.

I'm hoping someone here may have some idea of what's going on with my system. Any reply would be greatly appreciated.

Thanks for reading.

[SKYMTL]

Have you tried increasing airflow around your graphics card? Maybe place a 120mm so it is blowing directly at it in order to lower temps to see if that helps any. That being said, have you been monitoring your temps? If so, what do they read?

[[SNG]McBain]

I don't know the exact temp at the moment of the crash, but as I said, while in-game I would check the temps, and the graphics card would not go higher than 70 C. Usually hang around the 66 - 69 range. Do you think that is too hot?

[Keir]

If you have Riva tuner already there is a monitoring graph you can use. Run it in the background play your game for an hour, then quit. Go back and hold your mouse over the graph and you'll see the temp at that particular time.

This sounds a lot like the dreaded VRM problem from the 1950pro's but I'll try not to scare you too much with that.

But ya to rule out something else on the board overheating, point a fan at it, take the side case off, whatever you have to do to at least test it. If you can run a fan blowing on it and it still does it, well heat probably isnt your issue. It still may be hardware related though.

What about other board temps... like your Northbridge? You can use a program like Everest or SiSandra to monitor other temps.

are you o/cing? if so have you tried to just run on defualt(ish) bios setting and see if that works?

[[SNG]McBain]

Originally Posted by Keir

If you have Riva tuner already there is a monitoring graph you can use. Run it in the background play your game for an hour, then quit. Go back and hold your mouse over the graph and you'll see the temp at that particular time.

I've done this .... 70 C was tops

Originally Posted by Keir

But ya to rule out something else on the board overheating, point a fan at it, take the side case off, whatever you have to do to at least test it. If you can run a fan blowing on it and it still does it, well heat probably isnt your issue. It still may be hardware related though.

Good suggestion, I'll try that tonight

Originally Posted by Keir

What about other board temps... like your Northbridge? You can use a program like Everest or SiSandra to monitor other temps.

I have Everest, so I can check these. What are reasonable temps for the MB components?

Originally Posted by Keir

are you o/cing? if so have you tried to just run on defualt(ish) bios setting and see if that works?

Not over-clocking


Thank you Keir

[Ugly n Grey]

Try running the game and system with the side panels off the computer. A Cheap simple way of seeing if you have good case airflow. When I suspect heat, I usually stick a household fan facing right into the case and then stress it to see if I can invoke a lockup or failure.

Second possibility is PSU. If you still experience lock up, try a different PSU. The one you have puts up a theoretical 22amps on each 12v rail but only combines for a maximum of 36amps. While this is plenty of power under normal circumstances, heat lowers the efficiency of the PSU and it's possible that if the unit is weak at all you are exceeding capacity on one rail. I've been able to draw more than 18A from a 12v rail by stressing the snot out of a Brisbane with a high OC. So if your mobo rail (on the PSU) is marginal, you might be having problems there as well. (These are only possibilities, we are just trying to troubleshoot). I know you are not OCing, it's just worth checking out.

From your first post you say the CPU hardly ever reaches 40c... those readings are rarely correct in the case of Brisbanes and you can usually add +15c to that reading to get the real on die temps. Closest I could come with a thermal probe is the IHS and after doing the math I tend to agree with the much "better than I" experts on this issue. I don't think the 40c reading is dangerous because tcase max is usually 55-68 or something in there on the BE's.

Also don't forget to do the basic testing : Memtest for at least one full test suite. Dual prime overnight. If all is well then move on to the video card and stress it with loops of 3D testing to force a lockup.

Just thoughts and I'm sure you'll be able to resolve this issue. The 3850 is normally a cool non power pig type of card, so you should be able to narrow this down and get some serious game time in !
UnG

[[SNG]McBain]

Thanks for the reply UnG. I will pull the side of the case and use an external fan to either confirm or rule-out a temperature problem.

I have a couple questions specific to your reply (I'm still a bit new to building systems):

... what is "Dual prime overnight"? - not familiar with that term
and
... what would be the best way to stress the video card to cause a lockup, and what will that tell me? - is there a specific test that will tell me what is wrong when it locks up?

Thanks again

[Ugly n Grey]

well dual prime is just a way of running heavy math calculations on both cores to stress the CPU as much as possible. Popular programs for this are prime95, Orthos, Systool and many others.

As far as stressing video cards, I admit to being lazy and just running loops of Crysis benchmarking. It seems to work as well as anything, however, I am sure others here have more experience/suggestions on this than I can come up with. I will also use ATItool as it forces the card into 3D mode and checks for artifacting and is very good at it and this sometimes gives me hints about what's up with the card as it monitors temps as well.

[Keir]

yah Crysis benchmark can work for stressing, as can 3dmark you can set to loop.

As for what temps they should be at, its kind of a board by board thing, my Northbridge runs at 55 deg, which on some boards would be really hot, mine on the stock cooler was more like 65-70.
You can list them and we can see if the community thinks any are out of the normal range.

I really don't think your CPU temp is very high, you've got a nice cooler on it.

Some other things - Does this happen in any game, specifically what games have you gotten to happen in.

Have you updated things like Direct X? Chipset Drivers? The fact that your driver changing seemed to make it worse somewhat points to software. I'd suggest using the latest Cat drivers.

Have you run 3dmark06? Only reason I ask is if it fails on a certain test, that might point you in the right direction.

if your on XP and can use Ati Tray tools, that also has a 3d bench/stress type option. If it crashes while your doing these type of things, you can basically get it down to the video card vs something else.

[[SNG]McBain]

Wow, ok guys .... I've got some testing to do. Thank you for all the suggestions.

As for which games it crashed on, I've played Guild Wars, ET:QW, HL2:Lost Coast, and Portal - it's crashed on all of them at some point.

(I'm running XP, and Direct x 9.0c)