E8500 & P5Q Deluxe - Stability Issues

[n3xu5]

I've hit my 4 Ghz mark, and I'm in windows. However, I'm having stability issues. Every now and then a game or program may crash on me. I also ran Orthos last night for 57 minutes, and got 1 error and it stopped. Most of my settings on the P5Q Deluxe are on auto, but I'll post everything for anyone that may be able to offer some helpful advice. Originally I had no intentions of overclocking my memory, but if i can get this system stable I would like to have it running per GSkill DDR2-1000 spec.

System Setup:

E8500
8GB Gskill DDR2-1000
Asus P5Q Deluxe
ATI 4870x2
WD Velociraptor
WD 6400 AAKS x2
SB X-Fi Titanium Fatality Pro
Custom Built Water Cooling

I'm running a 9.5x multi, 425fsb, and 2.0 on dram voltage. Everything else is set to AUTO. Cpu spread spectrum, pcie spread spectrum, speed step, dram static read control, and c1e support are all disabled / off.




ASUS P5Q DELUXE Template:

AI Overclock tuner: MANUAL
CPU Ratio Setting: 9.5
FSB Frequency: 425
PCI-E Frequency: 100
FSB Strap to North Bridge: AUTO
DRAM Frequency:AUTO
DRAM CLK Skew on Channel A1: AUTO
DRAM CLK Skew on Channel A2: AUTO
DRAM CLK Skew on Channel B1: AUTO
DRAM CLK Skew on Channel B2: AUTO
DRAM Timing Control: AUTO

1st Information: 5-5-5-15-3-52-6-3
2nd Information: 8-3-5-4-6-4-6
3rd Information: 14-5-1-6-6

CPU Voltage: AUTO (Reads 1.272 in the latest CPU-Z)
CPU GTL Voltage Reference (0/2): AUTO
CPU GTL Voltage Reference (1/3): AUTO
CPU PLL Voltage: AUTO
FSB Termination Voltage: AUTO
DRAM Voltage: 2.0
NB Voltage: AUTO
NB GTL Reference: AUTO
SBridge Voltage: AUTO
PCIE SATA Voltage: AUTO

Load Line Calibration: Enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew : AUTO
NB Clock Skew : AUTO
CPU Margin Enhancement: OPTIMIZED

Orthos Error:

 

[setscrew]

Did you run Memtest on the RAM yet? Manually input memory values specified by GSkill.

What is the CPU voltage under load? What power supply are you using? Try running at stock speed first while you find the problem.

Once it passes Memtest, try bumping the Northbridge voltage as you are populating all four DIMMs.

 

[n3xu5]

Did you run Memtest on the RAM yet? Manually input memory values specified by GSkill.

What is the CPU voltage under load? What power supply are you using? Try running at stock speed first while you find the problem.

Once it passes Memtest, try bumping the Northbridge voltage as you are populating all four DIMMs.

Memtest+ pass 100% before OS install
Corsair 620HX
Stock Speeds ran Prime 95 for Vista 64 for 24 hours no issues.

as far as bumping NB voltage I wouldn't know where to start .. it was on AUTO ... but I tried some settings a friend suggested and this is what I have so far ...

I am currently in windows typing on it ...

vCore: 1.29 (CPU-Z shows this as 1.280 and switches to 1.288 every so often)
vPLL: 1.52
NB: 1.26 (I tried 1.28 and it was yellow ... )
Set Dram Frequency to 851

I also tried FSB Voltage @ 1.12 and 1.14 (this was the last thing I tried after booting several times into windows for the other settings first. When i set FSB to 1.12 or 1.14 the machine restarts with no video, so i had to reset the RCT / CMOS clock.) Right now it's on AUTO .. although standard minimum voltage for 45nm cpus is 1.10 ... weird. I don't know what to make of this or just leave it on AUTO.

 

[n3xu5]

I'm wondering if I should start over ... and shoot for a higher FSB like 450, and a 9x multi. Will it really make that much of a difference?

Also ... are 5:4 dividers for memory bad? I understand how they work, but have no experience with them. I was thinking of going 1:1 for 450 for 900Mhz DDR2 but if i can push it farther for everyday use .... does anyone have an opinion on this?

 

[Ocellaris]

Try running IntelBurnTest to test you OC, its going to give you results (a failure) a lot faster than Orthos testing. I would try 9.5x400 for starters, run the RAM 1:1 in spec and try to isolate the CPU testing for right now.

 

[Impulse]

I think you're pushing it plenty far for everyday use already... The dividers aren't inherently bad, you're just likely to not derive much of a benefit from running the memory faster if it comes at the cost of tRD (Anandtech has some good info on this, on the ROG Rampage review), so it's a balance game. Real-world benefits are probably minute either way, on the memory side... As far as getting the proc OC stable (which should be the priority if you already verified the memory to be stable on Memtest), I'm sure someone else can better help ya with that.

 

[kiajoon]

Your memory has not enough voltage. PC6400 4-4-4-12 memory or PC8000 5-5-5-15 memory will need a default minimum 2.1V to run as assigned by the manufacturer. Some may even specify 2.2V. If I am not wrong your G.Skill PC8000 memory should specify 2.1V.

Also set your memory to 6-6-6-18 to try. If this is stable then you can start playing from there.

And use IntelBurnTest to find out your srtability in a10 mins compared to Prime or Orthos.

 

[n3xu5]

IntelBurnTest? never heard of it ... thanks! I'll give it a look, Also I've heard OCCT 2 hours is like Prime 95 for 12 hours ... any knowledge on that?

 

[Impulse]

Uhh, sounds pretty bogus, seeing as OCCT is just a pretty GUI-wrapped Prime95 made specifically for stress testing...

 

[geok1ng]

if the memory could reach CAS4 DDR2 900 on 1:1 divider it would be great.CAS DDR2 1000 is possible on a few memories, the hard goal is to make this using tRD5...
The great problem in the C2D plataform is the NB cooling- put some air over it, if it doesnt help then you can add more voltage
remember anything beyond 400 FSB is past the 1600 rated mark for the NB, and without a CPU fan it will not withstand nearly as much heat as you think