Q9550 Fails OCCT @ stock clocks

[TRIPTEX_CAN]

I built a system for a friend last winter and it's crapping out already. I've ran OCCT and Orthos on the system and it fails within seconds @ stock frequencies and 100% auto voltages. Temps look OK. up to 60c (coretemp) with Orthos full load on CPU0 and CPU1.

Running Orthos with affinity set to CPU0 and CPU1 the system is perfectly stable. CPU2 and CPU3 fail every time.

Might try either underclocking the CPU or upping the vcore more. I already set the vcore to 1.3v and it didnt really appear to do anything.

Has anyone see this before?

Do you think it's time for an Intel RMA?




Specs:

Q9550 (Xigmatek HDT-RS1283 Red Scorpion)
P5Q Deluxe
4GB Mushkin Accent PC6400
4870x2
TX750w

 

[theonedub]

Was it ever clocked?

 

[newtekie1]

What temps is it reaching when running OCCT? 60°C seems a little high for just two cores being under load.

 

[TRIPTEX_CAN]

Was it ever clocked?

By me yes.

Originally I had set the CPU to 1.3v, 3.2Ghz, 400x8. 1:1.

 

[sneekypeet]

What is the error at the top of OCCT. If its a core error it will say, but Im leaning on the fact it may be ram related.

 

[stanhemi]

CPU Margin Enhancement :try switching from compatible mode to performance mode

 

[TRIPTEX_CAN]

What temps is it reaching when running OCCT? 60°C seems a little high for just two cores being under load.

What is the error at the top of OCCT. If its a core error it will say, but Im leaning on the fact it may be ram related.

Will have screens in a sec

CPU Margin Enhancement :try switching from compatible mode to performance mode

BIOS I'm assuming. I'll have a look.

 

[stanhemi]

Will have screens in a sec



BIOS I'm assuming. I'll have a look.

yep in the bios

 

[newtekie1]

What is the error at the top of OCCT. If its a core error it will say, but Im leaning on the fact it may be ram related.

I was thinking that also, but why would it crash on cores 3 and 4, but not 1 and 2? If it was RAM I would think it would crash on all 4.

 

[sneekypeet]

Cus ram is funny like that IDK, but I do find ram crashes OCCT alot sooner in my runs than the cores do. Thats why I asked if OCCT shows a "error on core 3" type of failure.

 

[TRIPTEX_CAN]

I was thinking that also, but why would it crash on cores 3 and 4, but not 1 and 2? If it was RAM I would think it would crash on all 4.

It says "Error on core 3"..

Before I moved the vcore up from stock it crashed on core 1.

 

[sneekypeet]

If everything is set stock (cleared CMOS) and just reset the ram the CPU should function fine. Now the tough issue to solve is, is it the board/socket. or is it the CPU.

Got another CPU handy to slam in for some testing?

Also what is the stock Vcore for that CPU on your board?

 

[TRIPTEX_CAN]

Coretemp and OCCT are reporting different temps.

-sneekypeet. the stock vcore was 1.16v load.

no other CPU to test with and no thermal paste in the office to reapply that either. Office support workstation = fail.

WOW

 

[sneekypeet]

Put the CPU vcore to like 1.25 and see if it corrects it plz. Those should be able to run at 1.15 but for giggles see what happens?

 

[DonInKansas]

1.3 is too high for 3.2. That chip was never a gem to start with. If it's failing stock clocks at 1.25/1.3 I'd RMA it.

If you've got a spare chip you might also try that one to rule out any mobo issues.

 

[TRIPTEX_CAN]

I think the board is incorrectly reading the temps. Same thing with the test i just ran. the log shows the CPU temp hitting 95c but coretemp and the OCCT real time monitor never broke 63c.

im going to put the vcore back to 1.25v.. reset cmos, re-seat RAM, and see what happens.

 

[sneekypeet]

use Realtemp too plz! Also u can get into the OCCT options and set the temp to 100*C to rule out temperature shutdowns.

 

[95Viper]

Re- seat the CPU, replace TIM, and check the fan(s) on the cpu cooler. You beat me to the re-seating the memory modules.

 

[sneekypeet]

Last minute thought....is the ASUS on either the latest bios, or at least one known to support Quads better?

 

[vega22]

i would clear cmos and let it run on auto/optimised defaults and see what that throws up.

no offence but maybe somebody has changed a bios setting (p45is a very tempermental chipset) and thats the issue?

 

[TRIPTEX_CAN]

Last minute thought....is the ASUS on either the latest bios, or at least one known to support Quads better?

When the system was setup last year I flashed the board to the latest and best BIOS for the 45nm quads. should still be functional.

i would clear cmos and let it run on auto/optimised defaults and see what that throws up.

no offence but maybe somebody has changed a bios setting (p45is a very tempermental chipset) and thats the issue?

On complete stock settings system will crash.. tried this morning.


OK... reset CMOS, changed vcore to 1.25, re-seated the RAM, changed CPU Margin Enhancement to "performance", and changed the OCCT max temp to 120c from 80c. OCCT ran for 15min with no error.

Temps aren't he best but that's easier to fix than a faulty core.

 

[sneekypeet]

Now just back down the core voltage until it fails.

You could try to RMA it, but I dont know what they will say other than up the voltage

I think you may have gotten an inefficient chip.

WOW do you have load line calibration on or something? OCCT is reading 1.31Vcore with 1.25v set in bios?

 

[TRIPTEX_CAN]

Now just back down the core voltage until it fails.

You could try to RMA it, but I dont know what they will say other than up the voltage

I think you may have gotten an inefficient chip.

I put it to 1.25v but It was only a test. I had it @ 1.3v before.. I'll back it down as far as it goes to see what happens.

LLC is disabled. I'll enable it to fix that.

 

[TRIPTEX_CAN]

Last post. System is stable and running stock clocks at 1.16v. Thanks everyone.