SOLVED Need help with i7 930, ASUS P6X58D Premium random freezing

Hello,

I really need some help.

I purchased parts for a new computer and had the guys at NCIX (www.ncix.com) put everything together for me. The system has been having issues ever single it was first booted up. The problem is that the computer will randomly freeze (can’t move mouse, no BSOD, no crash dumps). It even freezes while I’m browsing or idling in the BIOS. The freezing occurs anywhere between 1 min to 7 days of the computer running. Which makes things very hard to troubleshoot.

MoBo: ASUS P6X58D Premium (LGA1366)
CPU: Intel Core i7 930 @ 3.20GHz
Cooler: Noctua NH-D14
RAM: Patriot Extreme Performance Viper 3x2GB PVT36G1333ELK
GPU: 2X NVIDIA GeForce GTX 470 SLI
SSD: 40GB OCZ VERTEX2
PSU: Corsair HX1000W Power Supply
OS: Windows 7 64-bit

What I Know:
- When the computer freezes while playing music, the music goes on. Once it even crashed during a Skype call and I was able to keep talking with my contact (but no video was being sent or received since the screen was totally frozen).
- Setting all the BIOS settings to default makes the system most UNSTABLE
- Setting the RAM timings and DRAM Voltage according the memory specs improves stability slightly
- Overlocking (specifically increasing the QPI) in the BIOS improves stability noticably
- Prime95 will sometimes go for 24 hours without a freeze, but the system will then freeze randomly while on idle when I'm sleeping
- Launching any RAM-heavy application causes the computer to stutter and lag (most evident when playing music at the same time), making me think it's a RAM issue of some sorts

What I've Tried Replacing:
- CPU (to the same type)
- RAM (to the same type, and to a completely different manufacturer: Patriot, OCZ, Corsair)
- Motherboard (to the same type and to a different type by the same manufacturer)
- PSU (to the same type, and to one with more wattage)
- Video Card (to the same type, to single card, to 2 cards in SLI, to different PCI-E slots)
- SSD (to the same type and to a different SSD)

BIOS settings I've tried:
- Everything on Auto (reset to defaults)
- Everything on defaults except for specific RAM timings
- Overclocked to 3.2GHz based on some online overclocking tutorials

Other things I've tried:
- Using only 1 RAM stick
- Moving the 1 RAM stick to other DIMM slots
- A clean installation of Windows
- A clean installation of Ubuntu
- Testing the system in another house in case there's some problems with my power outlet

Nothing has helped.

Since this means I can rule out any hardware malfunction or operating system problems (it crashes in the BIOS after-all) I am completely out of ideas.

I will accept any and all suggestions - no matter how crazy or absurd. Aside from building a completely new rig from scratch with all different parts, I don't know what to do.

SOLUTION

One of my GeForce GTX 470's was faulty and was causing all the issues. It had nothing to do with the RAM or overclocking settings afterall.

Well with that setup, you should be able to run everything on auto in the BIOS and have no issues whatsoever. What power supply are you running all of that on? It's a decently power hungry system so if your psu isn't up to snuff that may be your issue...

Have you tried going into the Bios and loading the fail safe default settings? That would be my first try, then might begin to eyeball windows.

Moved to Intel Motherboard section. Please update your way of keeping track because the redirect will expire in two days.

As to your problem, I've never even heard of a PC freezing up in BIOS...it should definitely not do that. First, +1 to the PSU question, what have they put in there to power the system?

Second, I'd get back to basics and clear CMOS with a good, deep cleansing. Move the CMOS jumper to the clear position and yank the CMOS battery. Let it sit at least an hour.

After replacing the battery and restoring the CMOS jumper to the run position, boot immediately into BIOS. Go to the setting in your BIOS titled something like "load optimal defaults" and apply that.

Then set your memory manually. Go to the voltage section and set your memory voltage to 1.65V. Change your memory divider so that you're running DDR3-1333 or lower. Go to the memory timings menu and set the first four timings (CL-tRCD-tRP-tRAS) to the specified timings for your memory (9-9-9-20).

Leave everything else having to do with hardware on auto (except boot order/boot settings, if you need to change those). See how it treats you after that.

Lastly, welcome to OCF! In time, perhaps you'll change your mind from "I have absolutely no desire to overclock or get better performance." to "How can I squeeze some more performance out of this machine while keeping it stable...just because I can?"

globex said:

OCZ OCZ3G1333LV4G 12.0GB Triple-Channel DDR3 @ 534MHz (8-8-8-20)

Click to expand...

That screams I need more VTT. Fully populating all of the DIMM slots, is rarely acceptable on stock voltage. You were on the right track with the initial increase, which is why you noticed the increased duration of stability. As a precautionary measure I would increase it up to 1.275V-1.2875V (A bit of stability testing will be needed: IBT, Prime 95, or the like ...) You must make sure your loaded temperatures are under control, around 80C- 85C maximum for a synthetic stability program (this may be tough if you're using the stock cooler). This increase should be all you need for the stability you're looking for.

Welcome to OCF

kayson: I'm using a Corsair 750W PSU. When I reset the BIOS settings to defaults, the computer starts freezing between every 1min - 10hr.

GoD_tattoo: Are "fail safe settings" different from "reset default settings"? If so, where do I find that?

nzaneb: The DIMM slots aren't fully populated. I'm using OCZ's 3x4GB modules.

hokiealumnus: Thank you, I will give those suggestions a shot.

Make sure they're on the compatible vendor list. Did you set your ram speed, timings, and voltage manually? You'll most likely need to.

nzaneb said:

Make sure they're on the compatible vendor list. Did you set your ram speed, timings, and voltage manually? You'll most likely need to.

Click to expand...

How do I know what values to set? OCZ only says it should be 1.65V for DRAM. And that's what I have.

go into the memory timings area and set the values to 8-8-8-20-2t and the volts to 1.65

the memory timings should be on their own page

See my post above regarding timings & speed. When you change your RAM divider it should tell you what speed you're setting.

psionic, the product I found with that model number showed timings of 9-9-9-20. Where did you see 8's?

Is this your Ram http://www.ocztechnology.com/produc...66-12gb-gold-low-voltage-triple-channel-.html

You'll need to enter the BIOS memory configuration and change these value.

Manual configuration
Ram ratio will be 2:10 for 1333mhz effective (at 133 base clock)
CAS - 9
TRCD - 9
TRP - 9
TRAS - 20

edit: I guess I'm a little slow, lol.

If the OCZ memory has the XMP data, it should set your QPI/DRAM voltage. If you do need more voltage though, you can easily go up to 1.35 even 1.4V.

sorry read OP note and had timings on the end of the string of ram..

globex said:

Intel Core i7 930 @ 2.80GHz 36 °C (Stock cooler)

Click to expand...

kayson said:

If you do need more voltage though, you can easily go up to 1.35 even 1.4V.

Click to expand...

I wouldn't feel comfortable doing that on the stock cooler. If the OP was open to getting a better HSF... definitely

I think on a stock cooler and stock CPU speeds it shouldn't be an issue. My RAM (see sig) sets qpi/dram to 1.35V at stock speeds. Granted, I have a venomous x, but a 10% voltage increase on something not so major won't bring up temps too much.

kayson said:

I think on a stock cooler and stock CPU speeds it shouldn't be an issue. My RAM (see sig) sets qpi/dram to 1.35V at stock speeds. Granted, I have a venomous x, but a 10% voltage increase on something not so major won't bring up temps too much.

Click to expand...

You're also on 32nm and the OP is on 45. I would consider VTT a pretty major voltage (it's one of the 3 most crucial voltages for bloomfield) changes to it will directly effect CPU temps as it sits on die. 1.35-1.4V is a 12-15% increase, and is definitely too high for stock operation. Most 930's-950's don't need 1.4V for 4.2ghz operation, let alone stock. Maybe I'm missing the point of increasing it that high Overvolting that much on the stock cooler, is never good in my book. But to each his own I guess...

nzaneb said:

You're also on 32nm and the OP is on 45. I would consider VTT a pretty major voltage (it's one of the 3 most crucial voltages for bloomfield) changes to it will directly effect CPU temps as it sits on die. 1.35-1.4V is a 12-15% increase, and is definitely too high for stock operation. Most 930's-950's don't need 1.4V for 4.2ghz operation, let alone stock. Maybe I'm missing the point of increasing it that high Overvolting that much on the stock cooler, is never good in my book. But to each his own I guess...

Click to expand...

I'm just saying there's room to move the voltages up. I wouldn't go beyond 1.3 just to get it stable at stock, but I'm at 4.2GHz with 1.35Vcore , 1.35 qpi/dram, and my idle and load temps haven't gone up more than 2-3C from stock. So it's a pointless but possibly useful comparison

Edit: Also, power dissipation (so to some extent temperature) is roughly proportional to voltage squared. So with small voltage increases you get an even smaller power increase. It's directly proportional to frequency though, and that's what really gets ya

Thank you all for your very helpful replies. Here is what I did:

1. In BIOS I set "Reset Setup Defaults". Rebooted. Computer freezes every few minutes as reported.

2. In BIOS I set
- AI Overclock Tuner: X.M.P.
- BCLK Frequency: 133
- PCIE Frequency: 100
- DRAM Frequency: DDR3-1333MHz
- QPI/DRAM Core Voltage: 1.275V
When I saved settings, the computer wouldn't restart. Black screen. I turned the power supply off and then back on, and then the computer booted again. I am testing these new settings now. They seem to be working alright for now. The CPU temperature is currently just below 40C. Speccy/CPU-Z however is now reporting my RAM as "11-9-9-24". I'm guessing this is bad since my RAM is supposed to be 9-9-9-20.

However... I wasn't able to find where to find CL-tRCD-tRP-tRAS memory voltages. The BIOS has a "DRAM Timing Control" menu, but in there are A LOT of options. They start with:

- DRAM CAS Latency
- DRAM RAS# to CAS Delay
- DRAM RAS# to PRE Time
and like 20 other similar settings.

Where do I enter the 9-9-9-20 timings?

CL = Cas Latency
tRCD = RAS# to CAS# Delay
tRP = RAS# Precharge
tRAS = Precharge Delay

Make sure you adjust the DRAM Voltage to 1.65V

Make sure everything is compatible: i.e. cpu is supported by your mobo/bios and RAM is on the compatibility list for your mobo as well. On the P6X58D Premium, the top 4 RAM timings should be 9-9-9-20 in order. But the XMP should set that automatically...

Edit: also make sure your command rate is set properly