BSoD when screen saver trying to start

[writhziden]

That should be fine. Post back how the test goes.

 

[rting]

That should be fine. Post back how the test goes.

I ran Unigine Heaven 3.0 (See: Heaven DX11 Benchmark | Unigine: real-time 3D engine (game, simulation, visualization and VR)) and the FurMark showed that the GPU temperature goes up to 104 degrees celcius

I got a FPS of 30.8 and scoring 777.

See attached picture.

And no BSoD and still running now.

 

[writhziden]

In the future, you may want to keep the temperature from reaching over 100 C. I usually try not to exceed 95 C on my card, and it is a similar model to yours. Also, check your graphics card for dust clogging the fan. That caused mine to get to higher temperatures, as well.

Blow dust out with compressed air from a can or an electronics duster; never use a vacuum on the inside of the system because the static buildup and discharge from the vacuum can fry components.

 

[rting]

No kidding. But it looks like system is fine and the BSoD is fixed and it is indeed a faulty card? The above is a test done with the card that was not flickering.

 

[writhziden]

I would see if the other card works as well. Sometimes problems result from running two cards instead of one, so both cards may be fine, but together they cause problems.

 

[rting]

I would see if the other card works as well. Sometimes problems result from running two cards instead of one, so both cards may be fine, but together they cause problems.

Now I have the "other" card in the PC. Will call this card A (the primary suspect) and card B (the secondary suspect).

Issues without / before stress test:
1. I noticed that no matter what I do, I cannot bring the temperature down to below 83 degrees Celsius.
2. That said, card A when idling (nothing is running) has a steady temperature of 86-87 degrees. Card B has a steady temperature of 83-84 degrees. My room temperature has not changed much.
3. The FurMark Burn-in test has a "furry object" setting. With it on, the GPU usages goes way up to 100% and temperature goes as high as 107 degrees. With it off, the GPU stays around 36-37% usage for card B and 47-51% for card A.

Will retest both cards to verify the GPU usage percentages and repost / edit this post.

 

[rting]

Hang on - ignore the FurMark GPU usage numbers above.

Let me start over with the same set of parameters.

 

[rting]

Ok, after some stressful tests and endless night with FurMark, Memtest (Memtest86+ - Advanced Memory Diagnostic Tool), BlackBox Test, IntelBurnTest, etc., I believe I have located the problem.

The two graphics card are fine. They had tons of dust collected inside the vents and now their normal idle temperature is 83-84 degrees. See attached screen capture.

CPU seems stable. See attached IntelBurnTest result screen capture.

The issue, I believe is in one of the 2GB RAM module. See attached photo file name IMG_0970.jpg (sorry being upside down, took it with my phone). When being tested with Memtest alone as the only module, it had 9 errors. When tested with its twin as a pair, it had 1 error.

So NOW, I took that pair out and running on only 4GB RAM (I had 8GB). If nothing else, hopefully, I can send the memory in question back to Kingston and get a matching set of modules with the other modules that showed no error.

Hopefully, now this machine will not BSD ... it has been fine for about 6 hours in which there were numerous tests above, Battlefield 3 sessions with and without CrossFire (can be fun ), and screen saver no longer flickers.

That said, these DDR2 memories are more expansive than the DDR3 but my Asus P5Q-E does not support DDR3.

 

[writhziden]

Definitely appears that you found the problem. Did you test the slots to make sure it was not a bad slot and was definitely a bad module?

 

[rting]

I believe I tested the two pairs and individual unit in the same slots ... there were actually less errors when the unit in question were tested with its twin pair.

So as such, I would assume I need to test the pair in question on the other two slots just to make sure - at least.

I hope it is not the motherboard. :shock:

 

[writhziden]

It rarely is the motherboard in these cases, but I have seen it a few times in the hundreds of threads I've helped in with these problems. Best to be thorough, but don't stress too much about it.

 

[rting]

Well, one more thing (and I am not Steve Jobs LOL) ...

The two memory modules in question are slightly older. They are:

Kingston KVR800D2N6/2G ... got two of these.

The newer modules are:

Kingston KHX6400D2K2/4G that comes in a kit of 2 modules.

Their specifications are slightly different. The newer ones are CAS 5-5-5-15 (whatever that mean :huh the older KVR ones are CAS 6-6-6-18.

One of the older KVR who had problems in Slot A1 is having errors in Slot A2 (of the motherboard - 8 errors so far as I type this). Will test the rest. But I am not sure if I had the right memories mixing the old and the new - which worked fine for the most part until recent BSoD, etc.

 

[writhziden]

Mixing RAM can often lead to problems, but if you are seeing errors, it is likely a bad module and not the mixing itself that caused crashes. That being said: I would recommend that you try to get the exact same RAM as the modules you keep when you replace the other RAM just to avoid any mixing and matching issues.

 

[rting]

You may be right about the motherboard is okay.

I am seeing some of the addresses which produced an error to be the exact same address on one particular module (one of the KVRs) ... the error happened in Slot A1, A2, and now B1.

Translation - it is almost 100% sure that one or more of the millions of transistors inside this DIMM module (a KVR) may be faulty - thus same address locations keep coming up as errors. Nothing wrong with the motherboard I think.

The problem I have is that the memory test software is not Windows base nor DOS based. I could not keep track of the address other than taking a photo with my phone. I am planning to contact Kingston for their supposedly lifetime warranty.

If they give me new modules, I will want the KHX ones to match as suggested.

 

[writhziden]

Sounds like good steps and you reached a logical conclusion. Let us know when Kingston responds or when you get the new RAM, or both, whichever you prefer.

 

[rting]

Ok! Finally!

Sent the defective memory modules back to Kingston and then Kingston sent me the replacement - received it by FedEx today.

DIMM0 and DIMM2 are of the HyperX variety by Kingston and both pass the memory tests with no errors. These are Kingston KHX6400D2K2/4G (2 modules in a set making 4GBytes) and remained with me.

DIMM1 and DIMM3 are those who had errors during memory tests. These are Kingston KVR800D2N6/2G (2 modules making 4GBytes). The new ones are just installed and so far no issues.

If nothing else, the only thing odd is that the KVR800D2N6/2G (those that were replaced and currently in DIMM1 and DIMM3 slots) are forced to run at lower 330MHz and not at the standard 400MHz? Additionally, I noticed that in the section where it says "Speed (O/C)", the KVR800D2N6/2G are constantly above 15% and sometime close to 20% while the other two in DIMM0 and DIMM2 (the KHX6400D2K2/4G) are at 2% or so.

Currently researching what all these means ...

Updates ...
DIMM1 and DIMM3 are actually -15% .... not +15% LOL. My bad.

 

[writhziden]

Your new RAM will be clocked at the same frequency as the other set of RAM you have to prevent conflicts. That is probably why they are reduced to 330MHz instead of 400 MHz. The important question is: How is the system currently responding to the new RAM? Any stability problems?

 

[rting]

Your new RAM will be clocked at the same frequency as the other set of RAM you have to prevent conflicts. That is probably why they are reduced to 330MHz instead of 400 MHz. The important question is: How is the system currently responding to the new RAM? Any stability problems?

The system seems stable ... I mean after some BF3 running at Full Screen, crossfired the two (now cleaned HD4850), etc. at default resolution.

Other programs loads faster.

The problem must have originated from dust build up together with my ignorance of not checking for BIOS updates. Which now I will do if I get a new computer, etc. I also checked these memory at retail - those they sent me cost like $70! Like way more expansive than the newer DDR3 memory!

Looking out, if I want to update the memory all together, it seems that I cannot find DDR2 memory modules that are 4GB each - making a total of 4x4GB = 16GBytes. 16GBytes is the theoretical DRAM capacity of my Asus P5Q-e. In other words, it is either I get a new motherboard and CPU all together (My Core2Quad Q6700 only fits 775 socket) ... which may proof even more costly.

 

[writhziden]

There is really no performance increase with 16 GB as opposed to 8 GB unless you are doing video editing or AutoCAD/Inventor/Solid Works tasks. I do a lot of video editing myself, and I have been perfectly happy with my 6 GB system and 8 GB system, so you would be hard pressed to find a use for all 16 GB.

 

[rting]

I trade financial instruments - 4 x 22" screens ...

Being cheap in fact with my internet connection but besides the connection, it is the memory and the hard drive.

Currently looking at SSD drives ...

If I engaged crossfire, I loose 2 screens ... but that would be my "gaming setup" LOL.