"New" PC keeps freezing / crashing!

[drboon]

Hi

HARDLY AN INTRODUCTION
Well my (fairly) new computer keeps crashing . I'll just post the most important components of the build first before going into detail.

COMPONENTS
1x ASUS Striker II Extreme nForce 790i Ultra SLI
1x PC Power&Cooling Quad Silencer (750W)
1x Intel Core 2 Quad Extreme QX9770 (@3.2GHz, cooler: Thermalright IFX-14)
2x EVGA 9800GX2 SSC (1 card's disconnected from the PS atm)
1x Creative X-Fi Xtreme Gamer
1x 2GB OCZ PC3-12800 CL7 XTC Platinum
3x 3.5" WD Raptor X (WD1500AHFD)

=> The system is running with Windows XP (32-bit) and Windows Vista (64-bit).

THE PROBLEM
That system was built about a year ago. Everything looked good at first, but after setting up everything nicely I couldn't help but notice that the computer keeps freezing / crashing. I was - and still am - deadly certain that it's not caused by software, so I looked for any unusual temperature values.

TEMPERATURES (IDLE)
http://www.zshare.net/image/63094204be2e2865/

Correct me if I'm wrong but I don't think that there's any unusual temperature value in this listing. I'm not quite sure about the NB and SB though: 70-80°C (under load - saw this in the temperature monitor of the BIOS) seems pretty high (?).

The RAM has also been checked (no errors).

So, after doing this: for quite some time I'm finally asking you guys to help me fix this

 

[Fatal]

When does it crash with media and gaming or it happens in time?

http://www.tomshardware.com/forum/252088-30-solution-790i-freezes

Found that it says about freezing up for the 790i.

Edit:I do not know much about intel but I am sure some one will come along that can assist you.

 

[IINexusII]

2x EVGA 9800GX2 SSC (1 card's disconnected from the PS atm)

remove the card completely if your not using it

 

[kyle2020]

Im surprised the system would even boot with a card installed that isnt getting any power

Remove that card and see what happens.

Could also be down to the motherboard needing flashing to the latest bios - try that.

 

[drboon]

@Fatal: its crashing random

@IINexusII: why would I have to do that? I was afraid that the SB suffers from the heat of the second card but now that its not connected to the PS its not heating up at all.

 

[drboon]

already flashed the bios to 1104

 

[MaximusE]

upgrade your bios. also update ur drivers to the latest. also use a auslogics disk defrag program. is free. and scan ur pc. also check ur temps and ur cards ar properly connected from the psu

 

[MaximusE]

1x 2GB OCZ PC3-12800 CL7 XTC Platinum? if u can upgrade to 4 gb

 

[drboon]

1. the bios is up-to-date
2. latest drivers for all components are already installed
3. how does defragmenting help me fix deadlocks ?:>
4. temps are up there^^
5. the card does not need to be connected (consider it as not plugged in - i detached it from the psu yesterday for those of you who insist on connecting it to fix the problem )
6. I don't think that additional ram will fix the problem..I'm solely using winxp32bit so I'm more or less hitting the cap anyways

 

[MaximusE]

use registry defrag and disk defrag and u will see.

why the cards not need to be connected? what u mean

of course the additional ram will increase the power. if u check now u using around 1.5 gb on process. 2 gigs with this rig is not usual. use 4

 

[drboon]

with all do respect but defragmenting or cleaning up the registry does not solve that kind of a problem (i reinstalled xp about a month ago anyways).

I detached the 2nd card yesterday -> I'm having these problems for about a year now (with the 2nd card connected - obviously ).

additional ram won't help either since I'm not looking for more power but stability
1.5gb of unreserved memory is more than enough to run winxp - and as I said: it's crashing randomly so just leaving it in idle mode for some time suffices to make it freeze.

 

[95Viper]

Just a thought. Try moving that ram from the A1 (blue) slot to the A2(white) slot or vice-versa...and see if it still has random crashes.

In Vista goto the admin tools and look at event viewer to see if it gives any info on the crash.

If it blue screens write down the info and codes it gives you.

Random crashes are a pain sometimes, Goodluck

 

[phanbuey]

your MCP may be overheating... Mine was causing crashes until I put an 80MM fan over it. Just a thought. I have the 750i with circupipe cooling on my nb.

 

[drboon]

already did both - the event log doesn't show any errors.

yep..sounds kinda strange but I wish the OS would smack me with a bluescreen to get some sort of clue

 

[phanbuey]

so it crashes regardless of quad SLI, or does that make the crashes worse?

Have you tried overclocking at all?

Have you tried turning off c1e or any other bios power management?

More juice to the processor?

Can you post a CPU-z shot?

/20 questions - it sounds like a processor issue - you checked the mem and that was fine, so you should not be getting crashes due to bad sticks... although hard locks can be caused by not enough voltage to the CPU - possibly the cpu is being fed wrong volts, maybe due to power management being borked on the MB.. or it is degraded (did it have a previous owner) and needs more powah.

Last question... have you tried raising the volts and bumping the FSB term on the cpu?

 

[drboon]

1. yep, its not affected by quad sli
2. nope, manufacturer timings
3. all additional features are turned off
4.0 CPU voltages: VCore - 1.2875V, PLL - 1.5V, VTT: 1.3V
4.1 Mem voltage: 1.88V (its actually running at 1.9V)
5. CPU-Z (under load): Screenshot
=> no previous owner
6. which one of these would I have to change to do that? Voltage settings (these are NOT my settings. it's just a random pic of the available voltage settings)

 

[phanbuey]

1. yep, its not affected by quad sli
2. nope, manufacturer timings
3. all additional features are turned off
4.0 CPU voltages: VCore - 1.2875V, PLL - 1.5V, VTT: 1.3V
4.1 Mem voltage: 1.88V (its actually running at 1.9V)
5. CPU-Z (under load): Screenshot
=> no previous owner
6. which one of these would I have to change to do that? Voltage settings (these are NOT my settings. it's just a random pic of the available voltage settings)

thanks for answering all that i have to do some research on the optimal GTL ref settings of the proc - its might where your problem lies as your other voltages look absolutely fine.

BUT Your V-core shows 1.344V in CPU-Z but you say it is set at 1.2875V, thats a massive difference. Maybe loadline calibration (LLC) is upping the V's too much at load? Or worse, the motherboard is having issues controlling processor voltage.

Can you verify your load voltage using everest or speedfan and that the setting in the bios is 1.2875V? Because if so, and your CPU is jumping to 1.344v during load that's a pretty big deal.

Also FSB Term = VTT voltage... usually its fine to leave GTL refs on auto, unless youre trying for some insane clocks... Also - what is your NB voltage? this should be probably at least around 1.4+v as the 7-series chipsets are known for having comparatively weak NB's for quads.

 

[drboon]

Thanks for looking up the GTL ref settings I tried to do that myself but couldn't find anything :\

Agreed - the vcore's actually pretty high compared to the voltage in the bios. Should I disable LLC at this point?

NB voltage is 1.5V.

 

[phanbuey]

Thanks for looking up the GTL ref settings I tried to do that myself but couldn't find anything :\

Agreed - the vcore's actually pretty high compared to the voltage in the bios. Should I disable LLC at this point?

yes... and watch the proc's vcore at idle... see if its dipping.

1. Motherboard power
2. North Bridge shenanigans

Ill brb with the gtl's

EDIT: 1.5v that's what mine is at too - (1.51v) has to be or it craps out ... i have to have active cooling on it to keep stable at that though.

EDIT:
From Anubi @XSforums:

You adjust each Reference voltage separately. They aren't dependent on each other, they each do a specific function.

...There are reference voltages for the DIMMs, for the memory controller, for the FSB controller in the NB, and for the FSB controller in the CPU. They do the same thing in each case. The controller uses the reference voltage to determine if the signal it is receiving is high, giving a logic 1, or low, giving a logic 0.

Looking at memory for example, DDR2 has a single reference voltage for each DIMM, and Intel chipsets have a single DDR reference voltage for the chipset. Each reference only comes into play when that device is receiving. So the DDR reference for the NB is only used when the DIMMs are sending data to the memory controller in the NB. And the DIMM reference is only used when the memory controller is sending data to the DIMMs.

The same is true for the FSB GTL reference voltages. Each one needs to be tweaked independant of the rest. If you adjust them at the same time, you won't have any way to know which one is causing a failure. The same goes for if you adjust the reference voltages at the same time as changing other voltages or clock frequencies. If you get a failure, you can't tell what is causing it.

And this source advocates the best FSB and CPU GTL's as 67% of VTT - which at 1.3 would be ~.871-.872

 

[TheMailMan78]

I can't believe no one has said this yet...

1. Set everything back to default. No OC at all anywhere. No volt adjustments. No nothing.
2. Run OCCT for an hour and get back to us.

If that clears then it isnt a hardware problem. Then slowy tweak and test each adjustment there after until she bombs again. That should be it man.

This forum is full of HIGHLY intelligent individuals. However some lack common sense sometimes.

 

[phanbuey]

Basically it seems the 790i is notorious for hard locking if the vcore of the processor goes below a certain level... it definitely seems like a motherboard power issue -

One way to make sure that windows power management doesn't kick in is to go into vista power options>advanced settings> processor>min processor state 100%

@Mailman: Im sure he has done that already - he pretty much knows what he is doing. Also the 790i is known for this problem at stock. (read link below)

http://www.tomshardware.com/forum/252088-30-solution-790i-freezes

 

[m4gicfour]

1x 2GB OCZ PC3-12800 CL7 XTC Platinum? if u can upgrade to 4 gb

Wow. That's as bad as Compaq telling me to buy more ram for my laptop when I asked them if there was any difference between compaq's - 2005 - display driver package and ATi's :shadedshu (turned out that a crappy overlay for the volume control buttons needed compaq's version... I sacrificed that and took the double frames per second from updated ATi drivers.

On topic, I don't know about your specific hardware, but on my DFI board the NB was showing ~50-60C and I was having lockups when the OS power management was enabled. I removed the heatsink, took the crappy thermal pad off, reapplied a quality thermal grease, and added a fan. Temps dropped to ~30-45C and I stopped having lockups.

Your chipset may be able to take that kind of heat, but mine sure couldn't. I could have increased voltage too, but lowering temps seemed a better option.


also, if you haven't already, you could try running memtest (although memory errors in my experience cause bluescreens rather than hard-locks).

 

[TheMailMan78]

Basically it seems the 790i is notorious for hard locking if the vcore of the processor goes below a certain level... it definitely seems like a motherboard power issue -

One way to make sure that windows power management doesn't kick in is to go into vista power options>advanced settings> processor>min processor state 100%

@Mailman: Im sure he has done that already - he pretty much knows what he is doing. Also the 790i is known for this problem at stock.

http://www.tomshardware.com/forum/252088-30-solution-790i-freezes

Never under estimate some people. The guy could build a space shuttle from a can of tooth picks doesn't mean he can change a tire

 

[phanbuey]

Never under estimate some people. The guy could build a space shuttle from a can of tooth picks doesn't mean he can change a tire

 

[TheMailMan78]

If he had done what I suggested he would already know what the problem is. Or at least narrowed it down.