Core i7 920 Overclocking Help

[StrategyFreak]

Following the guide shown on this website, I tried overclocking the core i7 920 d0 I got from microcenter a month ago.

I am currently using an Asus p6x58d premium updated to the most recent BIOS.

Here is the relevant hardware:

6 GB Corsair PC12800
2xGTX 260
PC Power and Cooling 750w PSU
Swiftech Apogee XT waterblock
Swiftech MCP655 Pump

I was able to achieve a maximum stable bclk (Prime 95 8 hours, LinX stable 20 passes) of 210 at 1.350 QPI, according to the first section of the guide. (I did manage to get it to boot at 215, but I never was planning to take the system to above 4.2 ghz so I wasn't going to bother with getting it stable). I decided to skip to memory overclocking step and instead just use a really low multiplier because memory overclocking usually doesn't pay off in terms of the time put in.

I am having difficulty getting 4.0 ghz overclock stable at 1.325 vcore. At 1.325 vcore, I managed to get the system stable for about 5 hours in Prime 95 before encountering a BSOD. The system was stable for 32 LinX passes before giving an error. This is normally acceptable for me, but I was planning to do folding on this rig, so I need even more stability.

Is it abnormal for this system to require so much voltage just to get stable at 4.0 ghz? I can boot to windows at 1.250 vcore but it simply isn't stable.

Another observation: I never seem to get any failed tests in Prime 95, only BSODs. For the most part they are two letter/number combinations like 9c (don't know what to do, i usually just increase the vcore), but occasionally i get some like 101 (indicating vcore increase needed) and 124 (qpi increase?).

Ai Overclock Tuner.....................[Manual]
CPU Ratio Setting......................[21.0]
Intel(r) SpeedStep(tm) Tech............[Disabled]
Xtreme Phase Full Power Mode...........[Enabled]
BCLK Frequency.........................[191]
PCIE Frequency.........................[100]
DRAM Frequency.........................[DDR3-1531MHz]
UCLK Frequency.........................[3063MHz]
QPI Link Data Rate.....................[Auto]

CPU Voltage Control....................[Manual]
CPU Voltage............................[1.32500]
CPU PLL Voltage........................[1.80]
QPI/DRAM Core Voltage..................[1.35000]
IOH Voltage............................[1.36]
IOH PCIE Voltage.......................[Auto]
ICH Voltage............................[1.20]
ICH PCIE Voltage.......................[1.50]
DRAM Bus Voltage.......................[1.64]
DRAM DATA REF Voltage on CHA...........[Auto]
DRAM CTRL REF Voltage on CHA...........[Auto]
DRAM DATA REF Voltage on CHB...........[Auto]
DRAM CTRL REF Voltage on CHB...........[Auto]
DRAM DATA REF Voltage on CHC...........[Auto]
DRAM CTRL REF Voltage on CHC...........[Auto]

Load-Line Calibration..................[Enabled]
CPU Differential Amplitude.............[800mV]
CPU Clock Skew.........................[Delay 300ps]
CPU Spread Spectrum....................[Disabled]
IOH Clock Skew.........................[Auto]
PCIE Spread Spectrum...................[Disabled]

C1E Support............................[Disabled]
Hardware Prefetcher....................[Enabled]
Adjacent Cache Line Prefetch...........[Enabled]
Intel(r) Virtualization Tech...........[Disabled]
CPU TM Function........................[Enabled]
Execute Disable Bit....................[Enabled]
Intel(r) HT Technology.................[Enabled]
Active Processor Cores.................[All]
A20M...................................[Disabled]
Intel(r) SpeedStep(tm) Tech............[Disabled]
Intel(r) C-STATE Tech..................[Disabled]

 

[jarlmaster47]

take the multiplier down to 20. I oc'ed my 920 to 4ghz easy. I have bclk at 200 and multiplier at 20. I also have my vcore at 1.36V, allowing for plenty of power to be delivered. basically copy my overclock and then report back. willing to bet that it will be perfectly stable for u

 

[MIAHALLEN]

Hey SF to OCForums

IOH Voltage............................[1.36]
CPU Clock Skew.........................[Delay 300ps]

Can you explain why you chose these settings?
The clock skew specifically can cause problems (although usually not @ 300ps)....you shouldn't need any skew for these speeds.
The IOH seems way too high How many PCIe cards are you using in the rig?

Oh, and Vcore is not too high, my 920 DO needs 1.4V for 4GHz stability....just make sure your cooling is up to the task

BSOD during P95 small/large FFT or blend?

 

[Widowmker]

If you're using the 808 beta bios try using the 703 bios. I've heard the 808 beta isn't quite as stable as the 703. You also might want to check your memory timings if they're set to auto,
sometimes that board gets a little too aggressive in auto and you might have to loosen them up a little manually.

Miah - Congrats on your HWBot top ten.

 

[StrategyFreak]

Hey SF to OCForums



Can you explain why you chose these settings?
The clock skew specifically can cause problems (although usually not @ 300ps)....you shouldn't need any skew for these speeds.
The IOH seems way too high How many PCIe cards are you using in the rig?

Oh, and Vcore is not too high, my 920 DO needs 1.4V for 4GHz stability....just make sure your cooling is up to the task

BSOD during P95 small/large FFT or blend?

I am currently following the advice given in your guide for IOH voltage for 2 PCIe cards. 2x GTX 260

As for the clock skew, i saw on some forum that it might help stability, but i haven't noticed. I'll turn it off.

It BSOD'ed during blend.

I upped the voltage to 1.337 vcore and dropped the qpi to 1.25v (i had gotten stability at 1.20v at 190 bclk when finding the bclk maximum). It passed 50 LinX passes and I am currently running SMP folding along with 2 GPU clients to test for stability.

 

[MIAHALLEN]

Nice sounds good

Don't waste too much time "stress testing" with those synthetic apps.
If you're building the rig for folding/crunching....better to stress test it by folding/crunching

 

[StrategyFreak]

To clear things up, I'm using the 0703 BIOS. I'll take a look at the timings of the memory as well.

I got the system stable in folding for 8 hours with 1.337 vcore and 1.300 qpi. I got a 124 BSOD when testing with folding at 1.337 vcore, but i haven't gotten one since after bumping up the qpi from 1.250.

I'm using an Apogee XT waterblock with a 240 mm radiator with 2x Panaflo ultra high speed fans at 7v, so heat shouldn't be a problem. I am currently getting temps of 70C when running Linpack and 65-67C with folding. If temps ever get too high I could just switch the fans to full speed with the side effect being greatly increased noise. I might increase the voltages even more and try to hit 4.1 ghz or even 4.2 ghz since I think the CPU can handle higher temperatures, but the power consumption of running folding 24/7 at that would probably be quite high.

Right now I'm thinking about dropping back down to 3.8 ghz and try 1.25 vcore to help lower my electricity bill.

Thanks for the help guys. I guess I just didn't get as good of an overclocker. I might replace the i7 920 with a 930 later on, but from what I've read, those chips aren't that great overclockers either. I'm somewhat disappointed with the result with this chip.

 

[CompuTamer]

I've also found that just cause i can Prime for 24 hours doesn't mean that my CPU'll take MP3 encoding for my Zune. That constant 100% 10% 100% 10% load over and over and over again seems to be a really good stability test in itself... got a crash a few times there when i could go 24 hours under prime95.

 

[StrategyFreak]

I've also found that just cause i can Prime for 24 hours doesn't mean that my CPU'll take MP3 encoding for my Zune. That constant 100% 10% 100% 10% load over and over and over again seems to be a really good stability test in itself... got a crash a few times there when i could go 24 hours under prime95.

yeah, i've learned not even to trust linpack, which seems to push my cpu the hardest thermally.

i guess the only test is the program i'm running itself (Folding@Home). I just use the synthetic tests to establish a good baseline so I don't end up failing a bunch of WUs.

I'm disappointed that all the extra work and money put into the watercooling system will likely yield me an overclock that most people could probably easily achieve on air because my particular chip doesn't overclock as well.

 

[chris89]

Well the 920 and 930's are both great overclockers but they are both power hungry monsters. Even the new 32nm Gulftown 6 core i7 is going to have that earth shattering 130w TDP. It drives me crazy!!! why not engineer a super powerful 1366 with a 95W TDP???!?!? I would love these chips to death. The best stability test is Intel Burn Test. Set it to Maximum stress and use 32 threads which is equal to say encoding 10 1080p bluray movies all at once. It'll push the cpu so hard and is the best way to see if it is truely stable. I can run Everest or prime 95 for 8 hours on an overclock and fail in 5 minutes on intel burn test. These cpu's need a lot of Vcore for stability. 1.35V for 3.8Ghz, and 1.4V-1.425V for 4ghz. I haven't been able to get my system stable for intel burn test and i want to encode 10 blurays at once and let them run for 8 hours. I need that kind of stability. So i feel like my Rampage II Gene knows best when it comes to Voltage since everytime i set it to auto the board will intelligently know what's the best voltage for the overclock for stability. I'm running an i7 930 which is lapped to a perfectly flat mirror finish, and a Cocage True Spirit lapped to a mirror finish as well. I'm running 5 Scythe slip stream 110CFM fans in the case and 1 Scythe Ultra Kraze for the cpu cooler. Also using IC Diamond 24 carat thermal paste.

 

[Neuromancer]

I'm disappointed that all the extra work and money put into the watercooling system will likely yield me an overclock that most people could probably easily achieve on air because my particular chip doesn't overclock as well.

For me watercooling has 2 benefits to air...

1) It can be quieter than air cooling with the same or better heat dissipation
2) A radiator can be put anywhere not directly on top of the CPU so it opens up the air flow in the case and keeps VRM/Chipsets cooler by allowing more airflow to the sinks near the CPU.

Last winter I ran my rad outside. for some nice OCs

 

[chris89]

For me watercooling has 2 benefits to air...

1) It can be quieter than air cooling with the same or better heat dissipation
2) A radiator can be put anywhere not directly on top of the CPU so it opens up the air flow in the case and keeps VRM/Chipsets cooler by allowing more airflow to the sinks near the CPU.

Last winter I ran my rad outside. for some nice OCs

That sounds freakin sick! Radiator on the outside of the case. I would really like water cool a LIAN LI PC-V354B @ http://www.newegg.com/Product/Produ...cm_re=lian_li_microatx-_-11-112-300-_-Product and mount a radiator on top of the case with swivel mounts so i can mount 2 120mm fans on it and suck fresh air over the radiator and exhaust it into the fresh air. I think that would keep it running much cooler... And I could swivel it back to being flat on the case and the single 120mm case fan exhaust interior case air over it when idling and low load applications. Doesn't that sound like a good idea? I also have plans to modify the hard drive cage as well to still be able to mount the twin 120mm fans in the front of the case and hopefully push the cage forward enough to be able to mount a 10.5" graphics card in as well as be able to mount 7 3.5" hard drives.