New q9550 fails intel burn test at stock speed

[the DRIZZLE]

I just dropped a new q9550 into my Abit IP35 Pro. It fails intel burn test within 15 mnutes everytime. It also fails linx and the OCCT linpac tests, which are the same thing as intel burn test. The temps seem a little high but not crazy. They are roughly 67, 62, 62, 62 under full load, and all in the low 40s at idle.

Is this something I should worry about? I am running the regular OCCT test now for 10 hours so I will see what happens with that test.

 

[Idontcare]

Your temps sound ok, so I'm not suspecting a poorly seated HSF or TIM...but my next suspicion is that the cpu is getting undervolted somehow...maybe vdroop is excessive with your mobo.

What is your VID (stock vcore) and what does CPUz say the Vcore is when you have the burn test running?

 

[dguy6789]

I second Idontcare's advice. CPU Z says the core voltage on my stock voltage Q9550 is 1.192. I imagine yours should be close to that if it is running at the proper voltage.

Could you give your entire system specifications?

 

[n7]

Sounds like a RAM issue.

How much RAM?
What speed/voltage is your RAM at?
How much are you setting the programs to stress?

I've heard of people having issues on 32-bit OSes with LinX/IBT.
Make sure it's not stressing more than you have available.

 

[the DRIZZLE]

Full system specs:

Abit IP35 Pro
q9550
2x2gb ram (I think corsair)
Corsair 520 hx
EVGA 9600gt
Windows 7 32bit build 7600

I will check all the voltages when I get home tonight. I am dual booting 32 bit windows 7 and 64 bit vista. I can rerun the tests in vista tonight.

 

[taltamir]

run memtest overnight.. if memtest PASSES then troubleshoot the CPU, if memtest fails, you have a mobo or ram issue

 

[dguy6789]

Is it possible for memtest to fail if the cpu is operating at an unstable setting?

 

[Idontcare]

Originally posted by: dguy6789
I second Idontcare's advice. CPU Z says the core voltage on my stock voltage Q9550 is 1.192. I imagine yours should be close to that if it is running at the proper voltage.

Could you give your entire system specifications?

FWIW CPUz doesn't tell you your cpu's stock voltage (as in factory set VID)...CPUz tells you the current software-based reading as to the actual voltage your CPU is being fed. (i.e. its Vcc)

Vcc will be less than or greater than the VID depending on the motherboard's Vdroop characteristics combined with any active line conditioning in place to counter Vdroop.

To check your VID you can use Coretemp.

To determine Vdroop (which if excessive this can be a sign that your mobo is a walking wounded mobo with dead caps or vrms...this happened to me once) you take your coretemp VID value and subtract the CPUz Vcore value while the system is loaded with Prime95 small FFT (or something equally rigorous).

I love this AT article's graphics in showing the differences between the terms.

 

[Binky]

Remember that the IP35 Pro stopped getting bios updates right as the Q9550 was hitting the streets. I have the same combo with Bios 18 (the last bios released). It works fine but there were many quirks before bios 18 with the Q9550. I'm not sure that bios 18 is anything more than a beta bios...

 

[the DRIZZLE]

Originally posted by: Binky
Remember that the IP35 Pro stopped getting bios updates right as the Q9550 was hitting the streets. I have the same combo with Bios 18 (the last bios released). It works fine but there were many quirks before bios 18 with the Q9550. I'm not sure that bios 18 is anything more than a beta bios...

Yeah I am aware of that. I am running the 18 beta bios. Can you run intel burn test on your setup an let me know if you pass?

 

[BTRY B 529th FA BN]

Update to the latest bios
enable vdroop control
disable C1E and the other EIST
Make sure your vcore is close to your VID
run everything stock
Memtest86 which ever version, run test 5
raise your nb voltage slightly

 

[the DRIZZLE]

Ok I think I got it working, it just passed a 20 test run, but I will do more testing to confirm. Heres what I did:

1. Changed ram voltage from auto to 1.8V. It was reading 1.95V in abit uguru when set to auto
2. Changed ram frequency from 800mhz to 1:1 (666mhz)
3. Changed the cpu voltages from auto to manual
4. Changed FSB from 340 to 333

 

[Binky]

If that works out, you should get new ram. Running any 800mhz DDR2 at 800mhz with 1.95v should be very easy.

 

[geokilla]

I personally don't run Linpack. No need for that much stress testing. Linpack makes my E7200 hit 80C, thus failing the test. I have a OCZ Vendetta 2 too. I just run 6 hours of OCCT and couple hours of Memtest test #5 and overnight Memtest and call it a day. Folding@Home stresses my computer too.

 

[dguy6789]

I run LinX(Linpack) because I want my processor to be able to run any code that anyone could come up with for an unlimited amount of time if the need ever came up. If your processor fails Linpack, it is unstable. General use may not bring the load up enough for it to crash, but it will still crash if the processor is working hard enough. I couldn't live with that.

 

[the DRIZZLE]

I did pass IBT with the setting that I stated but I was still having stability issues OCCT was causing my system to crash after about 20 minutes. I looked into it and 1.9V was actually the correct voltage for my ram so I changed it to that. I also changed the GTL Ref to 65%. I passed OCCT for 8 hours last night. I started memtest86 before I left work this morning. Hopefully I get can this ironed out so I can start overclocking.