NF4 crashing... Please help!

[tompson]

Ok, so I'm experienced OCer and I'm doing my masters in EE so while I'm not by any means an expert, I usually have a fairly good idea what I'm doing with these sorts of things... I'm this case I don't have any idea what’s going on.

 

I know the following description is long but please bear with me.

 

Bottom line: ON STOCK I'm getting blue screens plus other strange side effects. A few weeks ago I was running at 245 x 11 without problems. Then the blue screens started happening so I dropped it all back but I still get them at stock. I can't seem to find a pattern for the blue screens, they just happen randomly (load or idle). Also, when I restart the computer sometimes it will just hang. The fans spin up but they stay high and the computer doesn't beep. It requires a restart and then the computer usually post fine.

 

I'm Prime 95 stable for 4+ hours (I haven't had enough time to let it run longer but I'm sure it would). Also I ran Memtest for 3 hours today without problems. My CPU temps are 48 Celsius under prime 95 (both cores) + 3D Mark 06 looping in the background (51 Celsius at 245x11)... Which is a little higher than I would like, but I find the 7800s and the 4800+ run pretty hot on the same cooling loop... and silence is important to me (since my computer is in my bedroom) so the res. fans aren't running very hard.

 

BIOS settings are:

CPU VID 1.3V

Dram Voltage 2.9V

CPU: 200 x 12

HTT ratio: 5x

 

(increasing the CPU voltage doesn't help...)

 

My BIOS memory settings are:

TCL 3

TRP 4

TRCD 9

Trc 11

Trfc 15

Trrd 4

Twr 3

Twtr 2

Trwt 4

Twlc AUTO

Tref 3120 cycles

Bank Interleave Enabled

Async Latency 7.0 ns

Idle Cycle Limit 16 cycles

 

The PSU voltages are 11.83V on 12V rail, 3.21V on 3.3V rail and 4.99V on 5V rail in Smartguardian under load.

 

Also, I'm running Windows XP x64. I just did a reformat and reinstalled everything (I thought maybe the NTFS file system was currupt).

 

My thoughts are to the power supply because of the random nature of these blue screens. How would I know if it was the power supply? If not that then maybe the memory heat sinks on the video cards aren't very good. They're the ozc ones with some thermal tape. I would use thermal epoxy but I want to have the option to remove them in the future. But, would over heating video card memory cause random blue screens?? I would have though vid artifacts first.

 

If anyone has any experience with this sort of thing (on this board) then please let me know.

[blick]

Try to unplug your audiocard just to exclude it as a potential cause.

For how long did you OC your CPU and what was your Vcore?

[Sorrento]

The PSU voltages are 11.83V on 12V rail, 3.21V on 3.3V rail and 4.99V on 5V rail in Smartguardian under load.

Those readings are from the ITE Smart Guardian aren't they? Do you have a DMM? The voltage reading between windows software, BIOS hardware doctor and a multimeter are always diferent... and this last one is the one to trust.

 

12v: yellow/black cables from 4pin molex

5v: red/black cables from 4pin molex

3.3v: first orange from ATX connector/black cable from 4pin molex.

 

The PSu 12v output on both rails should be around 12.32v at idle and around 12.28v at load... lower than this and the PSU might be overloaded by your hardware, and thus might be the source of your problems; two 7800GTX cards and the 4800+ Toledo are a lot of load. I have an Enermax Noisetaker 600w and it is failing to do its job on my system.

 

Also why are you using this latency: CL3 9 4 ???? and why is Tras missing?

 

Your ram latencies, either written at the ram sticker or from Corsair's specs consist on four numbers: first second third fourth. First corresponds to TCL at BIOS, the second corresponds to Trcd, third is Trp and fourth is Tras. I am not sure about your XMS4000's latencies, but I don't think those are right. Try adjusting them and the rest of the RAM settings according to Corsair's suggestions: check out their site:

 

http://www.houseofhelp.com/forums/forumdisplay.php?f=128

[madgravity34]

I am going to say right now that its your PSU. There is no way in hell a single 600w noisetaker can handle 2 gtx and a DC AMD chip let alone the h20 loop and other stuff. I have to run dual-psu's now just to cover my DC opteron and vmodded 7800gtx. The powerstream 520 runs the proc, while the enermax powers the rest of the system. I would stronly suggest you take out one of the gtx's and I guarantee your problem will go away. Buying a second PSU to power those two cards will probably solve all your problems.

[tompson]

Sorrento, I'm using higher than rated latencies at 200MHz. As you know, the ram is rated to 250MHz at faster latencies than this. I'll play around with the timmings but I'm not sure that that is the problem. In addition, the stability on memtest and prime 95 support that the memory is fine.

 

blick, the CPU was OC for about 4 months on CPU VID 1.5. The temps NEVER got above 51 Celcius. They would usually run around 48 load and 42 idle. I really hope that I haven't burned out the chip. This would be my first. I'm usually pretty careful.

 

madgravity34, thanks for the responce. It's interesting that you say that. I'll try taking stuff out and seeing what happens. I'll keep you guys posted.

 

Thanks so much for the help.

[tompson]

update: Underclocking the CPU seems to help. I pulled the multiplier back down to 11x (200) and put the HTT ratio on 4x. The restart issue goes away... I'm going to run some prime 95 + 3DMark loops and see what happens.

 

Not sure what this would mean. Could still be the PSU.

[HarryTaco]

I was wondering why you had such loose timings on your ram at stock freqs with 2.9 vdimm. Isnt the voltage a little high for those Corsairs?

Why dont you visit the Stock Settings Forum and look up someone elses settings for your board and ram and give it a shot before you go replacing hardware and spending money. My two-cents.

[tompson]

Loose timmings can only help stability. The fact that dropping the multiplier from 12x 200 to 11x200 without chaning HTT speed helped, would indicate that the ram is not the problem.

 

I'm guessing it's either the CPU or PSU. Of course I would hope it's the latter. For now I'll just have to run the CPU under stock because I have a EE thesis due in 2 weeks and I can't afford to not have a working computer for simulations.

 

Thanks for your help guys.

[Sorrento]

"Loose timmings can only help stability" This happens up till a certain point, I had XMS ram that wouldn't run at any other CAS than its default, and changing this value would make it unstable. But if looser = ok, then its ok for testing other hardware.

[supershanks]

It may be your psu like the guys say, removing your oc has also reduced demand on 12v from higher vcore & htt.

 

luck

[Sharp]

Hello,

 

Remove 1 gfx card, reduce your power load even more and see what happens.

 

Make sure that all 4 power connectors are plugged into the board.

Clear the CMOS for a few mins (10mins).