Overclocking Air: Intel i7 920, Asus P6T, G.Skill DDR3-1600 C8, Xigmatek HDT-1284E

[JakFrost]

Wasn't trying to be unhelpful, just didn't realize your processor was binned so low man.

Yes, that's the whole reason why I started this thread because I started hitting a wall trying to get past BCLK 180 with this processor D0 revision and batch number 3901A238 that was purchased in 2009-06 from Newegg.com. Other people are going 3.8 GHz on air easy and some 4.0 GHz with good air flow but I'm knocking my brains out trying to get passed 3.60 GHz.

A few folks on XtremeSystems made comments that some of the newer D0's are running "hot" even with low voltages. Others also commented on the fact that when Intel first releases a processor line with multiple speed products they don't really "bin" the processors to choose the best ones for the faster product but instead just build all of them to high-spec and label them as required for quantity. Later as they go on with the production line they learn to "bin" the processors so the not-so-good ones end up being labeled like the i7 920 but the best ones got out as i7 975 EE. My guess is that this is why a lot of folks with the old C0 revision were getting awesome clocks of 4.0 GHz+ on them.

Below is the info on my processor. I think that since Newegg is such a large distributor they have a large turnover of processors so the one that I got with the batch code 3901A238 is relatively recently produced in Costa Rica, year 2009, week 01. Since most folks in those threads report processors coming from the in the 38xx batches from last year and those are the ones with good clocks with 4.0 GHz+ stable systems.

The thread below details information about the batches and so far it looks like the 39xx batches are not so good and run hot.

XtremeSystems Forums > Hardware > Intel - Core i7 Overclocking by Batch #


Processor: Intel Core i7 920 2.66 GHz
Revision: D0 (SLBEJ)
FPO/Batch#: 3901A238

Date of purchase: 2009-06-29
Place of Purchase: Newegg.com

 

[JakFrost]

Hiya Jak!

Your temps seem kind of high. Have you considered reseating your heatsink?
As you can see, I'm @ 3.8 @ 1.211v, and 75c under load. Understand, I'm not bragging, or trying to be snide, I'm just telling you what I did wrong. Before, to get 3.8, I tried 1.325, and would hit 99 or sometimes even 00.

You see even when I hit this chip with 1.350 V and once with 1.450 V for CPU and QPI it still froze on BCLK above 180. At least you could run yours without freezing to try.

I will reseat the heatsink again today and try again but I have little faith that it will help.

I'm waiting on the intake fan though since I have more faith in that making a difference with this chip.

 

[gweminence]

You see even when I hit this chip with 1.350 V and once with 1.450 V for CPU and QPI it still froze on BCLK above 180. At least you could run yours without freezing to try.

I will reseat the heatsink again today and try again but I have little faith that it will help.

I'm waiting on the intake fan though since I have more faith in that making a difference with this chip.

If your heatsink is seated properly already, or, just as or more importantly, your TIM is properly applied, then yes, the fan will likely do more for you. If, however, like mine was, your TIM was not good...that's worse than a fan can make up for. So, it's good that you'll check it. You sound like you know what you're doing, so I doubt that it's bad, but then, I thought I was doing it right, and I wasn't. You should have seen the goop I had on mine

 

[JakFrost]

If your heatsink is seated properly already, or, just as or more importantly, your TIM is properly applied, then yes, the fan will likely do more for you. If, however, like mine was, your TIM was not good...that's worse than a fan can make up for. So, it's good that you'll check it. You sound like you know what you're doing, so I doubt that it's bad, but then, I thought I was doing it right, and I wasn't. You should have seen the goop I had on mine

Reseated and Reapplied Arctic Silver 5

Today I just removed the white ceramic based thermal grease with isopropyl alcohol and that came with the Xigmatech and polished the heat spreader plate and the heatsink copper pipes with Flitz metal polish to restore them to factory original condition and shine.

My friend also recommended to use scotch tape around the heat spreader and then apply the real Arctic Silver 5 compound that I had from a 3.5-years ago when I built my old AMD Opteron system. While it might sound that the compound is old I believe that since this is a non-degrading compound the age has no bearing on it. I debated between using the line-down-the-middle method of application that Arctic Silver 5 manufacturer recommends or the thin-layer method spread by credit card and went with a combination of both. I applied the very thin line like the manufacturer recommends over the four cores and then used the credit card to spread out the compound evenly and very thinly over the entire processor. I ended up scraping off about half of the compound that I applied very thinly onto the scotch tape all around the processor and what was left was a very thin layer in a perfect square pattern once I took the tape off. I put the heat sink back on and used the diagonal tightening recommendation to mount it to ensure that the paste spread evenly.

XtremeSystems Forums > Cooling > Air Cooling - application of thermal paste - what is the correct technique

Temperatures Are The Same

I did two runs of Prime95 Blend for 15-minutes each before I removed the heat sink with an ambient temperature of 25 C and got a peak temperature score of 83 C.

After the procedure I ran the same test with a little higher ambient temperature of 26 C because the sun came out a little more and I got a peak temperature of 84 C. Exactly the same expected result! No change or improvement with change of thermal paste and also with using a lot less of it.

Observations

One thing that I observed while polishing the heat sink's copper heat pipes is that they are a bit uneven with the aluminum base walls between them being a bit taller than the copper pipes. A bit of the thermal compound ended up being jammed in between the copper pipe and the aluminum wall. The copper heat pipes do seem to make contact with the processor since when I removed the heat sink earlier before cleaning the old grease off the flat surfaces on the heat pipes the grease left these little spikes after the heat sink was removed the same way that you see when you open up a jelly sandwich. This makes me believe that the heat pipes make direct contact to the heat spreader.

I think that this specific D0 revision batch 3901A238 processor just runs hot and that messing with the heat sink and thermal paste is not going to improve the situation.

I'm still eagerly awaiting the front intake bay and fan and I'm getting a little antsy about why it hasn't arrived yet from California since I ordered it in the middle of last week. I have my hopes set on this as at least providing some kind of better airflow to my case and possibly lowering the temperatures a few small degrees. We shall see.

Fan Speed and Blow Direction

Just wanted to mention that the fan is directed to pull the air from the middle of the case and then push it through the heat sink fins and directly at the rear exhaust fan to create an air stream effect. For an experiment I tried turning the CPU heat sink fan around to pull the air through the heat sink from the back of the case and then push the air over the memory module cooling fins but it turned out that my temps were at least 5-10 C higher this way so I put it back to the normal and recommended direction.

The fan is a Xigmatek hydro bearing 120 mm model with 800-1500 RPM speed and it runs at ~1,450 RPM on my system according to Asus Probe II. The other three 120mm fans in my system run around ~1,400 RPM also +-50 RPM.



FrostyTech - Xigmatek HDT-S1284EE Heatsink Review








Xigmatek HDT-S1284EE

 

[mastrdrver]

Have you tried running Real Temp 3.30? My temps got evened out better. I've got a 3903a batch and got my vCore down to 1.0 in bios and ran 8hrs of large OCCT without a fail. Temps saw a max of 54 with Real Temp 3.00. I know I'm on a P6T6, but I've read that my board is based off the P6T. I even have the same memory.

This was just last night to get an idea of how low I can go. If I loosen the timing to 9-9-9-24 1T, then I can get the voltage below .950 in bios. I know it says turbo enabled, but I don't as I have speedstep off which turns off turbo on my board. I've also read that running the multi at 20, instead of 21, allows for lower volts as same speeds.



My ambient is 21-22C here, but my cooler isn't much different (Vendetta 2). It's poorly seated too. I also pointed it toward the back of my 4870x2 in hopes that it would be better for the heatpipes going across the cores.

Also, why are your motherboard temps so high? Sounds like a possibility of poor airflow in the case might be to blame for you high temps on the cpu.

 

[JakFrost]

Have you tried running Real Temp 3.30?

Also, why are your motherboard temps so high? Sounds like a possibility of poor airflow in the case might be to blame for you high temps on the cpu.

RealTemp 3.30 Beta - Same Temp Results

I just downloaded this new version and I noticed that my temps are jumping up higher and quicker on this version than the previous one. After the same 15-minute Prime95 Blend test it hit 85 C peak while ambient is the same 25 C. I don't think it is the software because even Asus Probe II reports CPU temps of 88 C when RealTemp reports Core 0 at 84 C. That's pretty close and accurate.



Betting on Intake

I believe the problem with the high motherboard and processor temperatures is the lack of a dedicated intake fan from the front of the case for the top compartment where the most heat generating parts are. There is a nice large intake fan in the bottom compartment that blows over the hard drives to keep them cool and I noticed that the drives stay very cool to the touch because of the fan whereas when I used the same drives in an external enclosure on top of the computer without active air cooling the drives get almost too hot to touch.

 

[sodina]

thats some high temp
i recently put one up for my friend
i had same cooler as yours (darknight)

it was stable at 3.6ghz (200 x 18) with 1.120v
in prime 95, it never went above 77 even with pretty warm room temp

 

[JakFrost]

Testing the New Thermal Compound Application

Tonight I'm going to let my system test the new compound application to see how it does with it. Previously on the same CPU 1.20 V I hit 80 C peak with ambient 25 C. I am going to see how it goes tonight but I have a feeling that applying a thin layer like I did today did not provide enough compound for nice and even coverage so it might not do so well. The reason is that those gaps between the heat pipes and the aluminum block probably absorbed too much compound.

I might have to take it off tomorrow to see how well it did by looking for gaps in the contact points and if it is not good then I'll use the thin-line method to apply a little bit more without cleaning the existing thin film to get it to spread out a bit more.

I know that Arctic Silver 5 has a 200-hour curing period but the initial temperatures should be close to the white ceramic thermal grease that I used previously.

We'll see what happens.

Also the intake bay and fan are almost here. They are in a USPS depot since yesterday afternoon so they should be making their way to my house in the next day or two hopefully.

 

[JakFrost]

New Compound Application Worse - ~3 C Peak Higher

It seems that the thin-line along with spreading it out with a credit-card combined methods are not so good since my peak temps are ~3 C higher with an ambient temp difference of only 1 C. This could be just a normal fluctuation but I have a feeling that there just isn't enough compound to get into all the little nooks and cranies.

I wish that I listened to myself earlier yesterday and took pictures of the polishing and application of the thin coat method because the final results after the scotch tape was removed looked absolutely almost factory perfect.

I'm going to take the heat sink off today and re-do the process but this time leave a little very thin line in the middle like Arctic Silver recommends and then re-test the thermals tonight.

 

[JakFrost]

Intake Fan Installed and Thermal Paste Reapplied Thicker = 10 C Peak Temperature Drop

I got the intake fan and installed it and also reapplied the thermal paste but this time thicker. I ran the same stress test with the same ambient 25 C temperature and now my peak temp after 10-minute and 30-minute check is 74 C instead of 84 C so my temperature drop is 10 C. I am very happy with this now and the temperatures of this processor are back in line.

Very Thin-Layer Method Was Flawed

The previous very thin-layer application of the thermal paste was flawed since when I removed the heat sink now there were huge areas of no-contact zones. I decided to resolve this problem and apply the thermal paste thicker to the processor and at the same time to the heat sink base, making sure to fill in all the gaps between the copper pipes and aluminum block. Just for a bit extra I placed a very thin line down the middle of the cores to ensure that this line gets spread out and flattened upon contact with the heat sink base. It seems that the case of this heat sink is not exactly level so there is need for more thermal paste than normal.

I know that a few folks will cringe at how much paste it seems that I applied but I kept the layers very, very thin but evenly covered to ensure that the entire heat sink and heat spreader areas are covered because in the two previous attempts using too-little thermal paste was probably more of an issue than using a little too much. I rather error on the side of caution and use too much than to end up with no-contact zones like I did.

Intel Core i7 920 2.66 GHz D0 - Thin-Layer, No Contact Zones Showing

Xigmatek HDT-1284EE - Arctic Silver 5 - Thin-Layer, No Contact Zones Showing

Intel Core i7 920 2.66 GHz D0 - Arctic Silver 5 - Thin-Layer + Thin-Line

Xigmatek HDT-1284EE - Arctic Silver 5 - Extra Thick

 

[JakFrost]

Intake Fan Installed

Lian-Li PC-V1200 Plus Case - Front - No Intake Fan Installed

Lian-Li PC-V1200 Plus Case - Front - Scythe S-Flex 120mm 1,200 RPM Intake Fan Installed

Lian-Li PC-V1200 Plus Case - Back On Side Open - No Intake Fan

Lian-Li PC-V1200 Plus Case - Back On Side Open - Intake Fan Installed

Lian-Li PC-V1200 Plus Case - Front On Side Open - Scythe S-Flex 120mm 1,200 RPM Intake Fan Installed

Lian-Li PC-V1200 Plus Case - Lian-Li Intake Fan Installed - Holes Do Not Line Up

Lian-Li PC-V1200 Plus Case - Intel Core i7 920 2.66 GHz D0, Asus P6T 0603, G.Skill 6GB DDR3-1600 C8 2T, Xigmatek HDT-1284EE - Case Inside

Lian-Li PC-V1200 Plus Case - Inside Back

 

[JakFrost]

No Temperature Comparison Due To Blue Screen

I started running the Prime95 Blend stress test last night hoping to get a new lower temperature result with the new intake fan installed only to have my system crash at ~2:30 AM at night with a 0xFC error. Frak! What is going on with this machine.

*******************************************************************************
*                                                                             *
*                        Bugcheck Analysis                                    *
*                                                                             *
*******************************************************************************

[B][U]ATTEMPTED_EXECUTE_OF_NOEXECUTE_MEMORY (fc)[/U][/B]
An attempt was made to execute non-executable memory.  The guilty driver
is on the stack trace (and is typically the current instruction pointer).
When possible, the guilty driver's name (Unicode string) is printed on
the bugcheck screen and saved in KiBugCheckDriver.
Arguments:
Arg1: fffff8800d0feca0, Virtual address for the attempted execute.
Arg2: 800000013e56c963, PTE contents.
Arg3: fffff8800d0fea30, (reserved)
Arg4: 0000000000000000, (reserved)

Debugging Details:
------------------


CUSTOMER_CRASH_COUNT:  1

DEFAULT_BUCKET_ID:  VISTA_DRIVER_FAULT

BUGCHECK_STR:  0xFC

[B][U]PROCESS_NAME:  prime95.exe[/U][/B]

CURRENT_IRQL:  9

TRAP_FRAME:  fffff8800d0fea30 -- (.trap 0xfffff8800d0fea30)
NOTE: The trap frame does not contain all registers.
Some register values may be zeroed or incorrect.
rax=0000000000000000 rbx=0000000000000000 rcx=0000000000000009
rdx=0000000000000022 rsi=0000000000000000 rdi=0000000000000000
rip=fffff8800d0feca0 rsp=fffff8800d0febc8 rbp=fffffa80063ac840
 r8=0000000000000029  r9=0000000000000000 r10=fffffa80082f3a60
r11=fffff880009f2180 r12=0000000000000000 r13=0000000000000000
r14=0000000000000000 r15=0000000000000000
iopl=0         nv up ei ng nz na pe nc
fffff880`0d0feca0 56              push    rsi
Resetting default scope

LAST_CONTROL_TRANSFER:  from fffff80002ef7bb9 to fffff80002e8ff80

STACK_TEXT:  
fffff880`0d0fe8c8 fffff800`02ef7bb9 : 00000000`000000fc fffff880`0d0feca0 80000001`3e56c963 fffff880`0d0fea30 : nt!KeBugCheckEx
fffff880`0d0fe8d0 fffff800`02e8e06e : 00000000`00000009 fffff800`02e8bb79 00000000`00000000 00000000`00000000 : nt! ?? ::FNODOBFM::`string'+0x3d9c5
fffff880`0d0fea30 fffff880`0d0feca0 : fffff800`02e8bb79 fffffa80`063ac848 fffffa80`075fb000 fffffa80`00000000 : nt!KiPageFault+0x16e
fffff880`0d0febc8 fffff800`02e8bb79 : fffffa80`063ac848 fffffa80`075fb000 fffffa80`00000000 fffff800`02e91e82 : 0xfffff880`0d0feca0
fffff880`0d0febd0 fffff800`02e8b958 : 00000000`01e57000 fffff880`0d0feca0 00000000`0b977760 00000000`03265400 : nt!KiScanInterruptObjectList+0x69
fffff880`0d0fec20 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : nt!KiChainedDispatch+0x128


STACK_COMMAND:  kb

FOLLOWUP_IP: 
nt! ?? ::FNODOBFM::`string'+3d9c5
fffff800`02ef7bb9 cc              int     3

SYMBOL_STACK_INDEX:  1

SYMBOL_NAME:  nt! ?? ::FNODOBFM::`string'+3d9c5

FOLLOWUP_NAME:  MachineOwner

MODULE_NAME: nt

IMAGE_NAME:  ntkrnlmp.exe

DEBUG_FLR_IMAGE_TIMESTAMP:  49ee9439

FAILURE_BUCKET_ID:  X64_0xFC_nt!_??_::FNODOBFM::_string_+3d9c5

BUCKET_ID:  X64_0xFC_nt!_??_::FNODOBFM::_string_+3d9c5

Followup: MachineOwner
---------

Possible Causes - Too Much Thermals Paste, Voltage Dip Due To Current

I am going to enable temperature and voltage logging to see what is happening with them after Prime95 runs for 2-hours. I'm a little concerned now that the additional thermal paste might be too much and it might cause the processor to build up heat pockets with too much thermal grease between the heat spreader and the heat sink. Also, there is a chance that the additional fan that I plugged in directly to a molex connector through a 3-pin-to-molex adapter is drawing additional current and dropping the voltage too much and causing system instability after extended period of time.

+5 V Rail Voltage Dipping by 0.25 V On Stress Testing

One thing I noticed about voltage is that what the +12 V line is at 12.24 V on my multimeter and 12.19 V in Asus Probe II, when I start Prime95 is only drops to 12.23 V and 12.18 V in Asus Probe, showing very good current handling for the most important voltage rail in the system. However, the +5 V line is not doing so well. It shows +5.05 V in multimeter and +5.02 V in Asus Probe II, but when I start Prime95 it drops to 4.72 V in multi and 4.76 V in Asus Probe II. That is a sign of weak current handling on this rail. Luckily my power supply is the OCZ Adjustable kind so I'm going to turn the adjustment potentiometer on the back of the power supply to push the +5 V rail to +5.25 V to compensate for the voltage drop during Prime95 testing. The other raile +3.3 V stays solid in Asus Probe II at 3.26 V but I don't have lead that is easy to reach right now to test it with my multi. It doesn't budge when I do Prime95 though.

 

[JakFrost]

The +5 V Rail Dipping 0.75+ V

I ran a bit of a stress test today for just a few minutes and I saw the +5 V rail drop to 4.74 V as before but after a bit more time it started going lower 4.60 V and went as low as 4.35 V and held there.

I adjusted the potentiometer on the back of my power supply to up the rail by 0.25 V to 5.24 V but even with that increase it still dropped. My power supply is supposed to be able to do 40 Amp of current on the +5 V rail but it seems that it just cannot handle this load.

OCZ PowerStream 520W SLI ADJ ATX2.0 EPS12, +3.3V 28A, +5V 40A, +12V 33A



Power Supply Upgrade

I might have to upgrade my power supply to a more powerful model because of this problem. The problem is that the best power supplies have that 120mm fan inside and blow the air down onto the CPU. However my case has a reverse setup and a separate compartment for the power supply so it seems that I will be restricted to using only flow-through power supplies like the OCZ line. I'd rather go with a Corsair or Seasonic model that I like but this is just not possible I think.

 

[JakFrost]

Just made this post in the Power Supply forum.


[H]ard|Forum > [H]ard|Ware > Power Supplies - 0.30 V drop only on +5V rail during Prime95 - Is this normal?

I ran into a strange problem with Prime95 on a previously completely stable 3.60 GHz overclock speed after adding in a Scythe S-Flex 1,200RPM 120mm fan. Previously I did multiple 9+ to 11+ hour stress tests on the same settings and got a clean results but yesterday my computer blue screened. I disconnected the fan but the power issue is still the same so the fan is not the cause, just the reason why I noticed the power problem on the +5 V rail in the first place.

I inspected the fan and it draws power from the hot yellow +12 V rail line and returns it properly to the black neutral line right next to the yellow and I don't see any resistor packs on the lines so it doesn't seem to include a undervolt mod on the lines from the factory.

POW: OCZ PowerStream 520W SLI ADJ ATX2.0 EPS12, +3.3V 28A, +5V 40A, +12V 33A

I decided to look at possible power supply issues and noticed that previously my system's +5 V rail was running at +5.07 V and it would dip to ~4.72 V during Prime95 Blend within about a minute or two. I have an adjustable power supply so I used the potentiometer to up that rails' voltage to ~5.21 V and now it drops to +4.85 V after a minute and dips occasionally to +4.83 V after a few minutes of running it. I can't push my rail any higher or the green LED on the power supply turns read indicating that it is out of range and then Asus Probe II will also shown a warning. Still the +3.3 V does not change and +12 V drops only 0.05 V during Prime95 stress testing.



Now is this type of a drop normal for my power supply?

Below is the thread that I'm using as a working log of the overclocking issues with my computer.

[H]ard|Forum > [H]ard|Ware > Overclocking & Cooling
Overclocking Air: Intel i7 920, Asus P6T, G.Skill DDR3-1600 C8, Xigmatek HDT-1284E



Below is my current system config for reference.

[size=-2]
MOB: ASUS P6T, LGA 1366, Intel X58 ICH10R, 3 SLI, 6 DIMM, 6 SATA - FW: 0603, AiTweaker XMP, Turbo Off, HT On, SpeedStep On, C-Step On, (Settings) Auto
CPU: Intel Core i7 920, 2.66 GHz, 4C, 8 MB, 4.80 GT/s QPI, 45nm, 130 W, D0 (SLBEJ) 3901A238 - 3.60 GHz, BCLK 180, Multi 20, CPU 1.20 V (Manual), QPI 1.21250 (Auto)
FAN: Xigmatek HDT-1284EE 120mm Hydro Bearing Fan 800-1500 RPM 4 Heatpipes - 41 C Idle, 81 C Load
RAM: G.Skill 6gb (3 x 2 GB) DDR3 1600 (PC3 12800) 8-8-8-21 2T 1.60 V - XMP Profile 1, Ratio 2:8, Frequency 720 MHz (1,440 MHz), DRAM 1.60 V

VID: eVGA nVidia GTX 260 896 MB G92 192c 576/999 MHz PCI-e 2.0 16x 2xDVI 1xSVid 1xHDTV - 600/1100 MHz, DVI to LCD, HDMI 60' to 50" Plasma TV.
CAP: AverMedia AVerTV Combo PCI-e M780R ATSC/NTSC Capture MPEG-2 Hardware Encoder - MCE Remote, IR Receiver

CAS: Lian-Li PC-V1200 Plus Mid-ATX Aluminum 4x5.25 6x3.5 2x120mm - Best case to date in my opinion.
POW: OCZ PowerStream 520W SLI ADJ ATX2.0 EPS12, +3.3V 28A, +5V 40A, +12V 33A

SSD: Intel X25-M 80GB MLC SSD SATA II 3 GB/s - FW: 8820
HDD: Western Digital Caviar SE16 500 GB WD5000AAKS 16 MB SATA-II - (4) in (2) RAID-1 (Mirror) Arrays

ENC: MacAlly G-S350SU External HDD Enclosure 3.5" eSATA USB 2.0 Aluminum, Universal 12V Adapter - (2) external WD GP 1TB for Movies
ENC: i-Rocks 9410 External HDD Enclosure 3.5" eSATA USB 2.0 Aluminum, Univeral 12V Adapter - (2) external WD GP 1TB for Backups
HDD: Western Digital Caviar GP Green 1 TB WD10EACS 16 MB SATA-II - (2) external Movies, (2) external for Backups

CON: JMicron JMB363 PATA and SATA Controller, PCIe 1x, 1 PATA, 1 SATA, NCQ, Hotplug, RAID 0,1,0+1,JBOD
CON: JMicron JMB322 SATA II 3 GB/s Controller, PCIe 1x, 2 SATA, NCQ, Hotplug, RAID 0,1,JBOD
NIC: Realtek 8111C PCIe GigE
SOU: Realtek ALC1200 8 channel, HD Audio, Coax/Opt S/PDIF - Analog for Speakers, Motherboard Digital S/PDIF to nVidia GTX 260 S/PDIF for DVI->HDMI

DVD: NEC ND-3550A 16x DVD+-RW SL/DL 48X CDR PATA - FW: 1.07
BLU: LG GGW-H20L Blu-ray 6x BD-RE 16x DVD SL/DL 48x CDR SATA - FW: YL05

MON: HP LP2475w 24-inch LCD H-IPS Monitor Wide Gamut 102% NTSC 6-12ms Response - 1920x1200x32 60Hz, DVI, GIG122/GIG052, Bri 15, Con 80, 9300K
KBD: Microsoft Natural Keyboard Pro USB/PS2 Media Keys
MOU: Microsoft IntelliMouse Explorer 3.0 Optical USB/PS2 Scroll-Wheel 5-Button
JOY: CH Products Fighterstick USB 3 axes 24 buttons
THR: CH Products Pro Throttle USB 3 axes 24 buttons
PED: CH Products Pro Pedals USB 3 axes

ROU: Linksys BEFSR81 EtherFast Cable/DSL Router 8-port Switch 100MBps FW: 2.51.3

OSS: Microsoft Windows 7 Release Candidate Build 7100
[/size]


Rail +5 V on After Prime95 Blend Stress Test - +5.02 V

Rail +5 V on Prime95 Blend Stress Test - +4.78 V

Power usage during Boot-up - 251 W

Power usage during Idle OS - 234 W

Power usage during Prime95 Blend Stress Test - 384 W

Intel Core i7 920 2.66 GHz D0, Asus P6T 0603, G.Skill 6GB DDR3-1600 C8 2T, Xigmatek HDT-1284EE - CPU 3.6 GHz (180 x 20) 1.20 V, DRAM 720 Mhz 2;8 8-8-8-21 2T 1.60 V, QPI 1.2125 V - Prime95 Blend Stopped 76C Peak - PowerStrip

Intel Core i7 920 2.66 GHz D0, Asus P6T 0603, G.Skill 6GB DDR3-1600 C8 2T, Xigmatek HDT-1284EE - CPU 3.6 GHz (180 x 20) 1.20 V, DRAM 720 Mhz 2;8 8-8-8-21 2T 1.60 V, QPI 1.2125 V - Prime95 Blend 5-Min 76C Peak - PowerStrip

 

[keto]

I would say you have very poor mating surfaces, to start. Check for flatness using a straight edge, both the bottom of the cooler and the top of the heat spreader. Likely both, but for sure one or the other could stand to be lapped, using a known flat surface (ie., panel of glass) and fine grades of sandpaper starting at (depending who you read) 400-600-800 and going up to at least 1000 if not higher. Search around, you will find descriptive articles on the process.

I don't have the reference at hand, but I read a review of your cooler which was much less than flattering.

FWIW, my new (48 hours) chip is a C0 and on a Gigabyte mobo and so not necessarily a direct comparison, at 3.6 I'm full load temps not higher than 75 using an unlapped (also cpu unlapped) TRUE 120 Black and 2 1600rpm fans in push-pull. I am not yet stable but have been to 3.8 and not seen 80C. vcore 1.30 now 1.35, again I recognize not necessarily direct comparison to your system.

 

[MittWaffen]

I have the same setup as you.

Do this, bring your QPI to 1.37 run P95 and bring down the voltage after you hit the 12 hour mark..notch by notch bring it down ever hour and eventually you'll crash. Bring it two or three notches higher that what you crashed at and you'll be fine. its over intels spec but really, .25 over isnt very much when you lower v-core massively.

Running 4.2ghz stable.

 

[adobian]

I have the same setup as you.

Do this, bring your QPI to 1.37 run P95 and bring down the voltage after you hit the 12 hour mark..notch by notch bring it down ever hour and eventually you'll crash. Bring it two or three notches higher that what you crashed at and you'll be fine. its over intels spec but really, .25 over isnt very much when you lower v-core massively.

Running 4.2ghz stable.

What is your set up again? I want to mimick yours to get 4.2ghz stable when I get to build my i7 gig.

 

[d50man]

you multi is holding you back use the 19X and crank the bclk

 

[JakFrost]

Do this, bring your QPI to 1.37 run P95 and bring down the voltage after you hit the 12 hour mark..notch by notch bring it down ever hour and eventually you'll crash. Bring it two or three notches higher that what you crashed at and you'll be fine. its over intels spec but really, .25 over isnt very much when you lower v-core massively.

What is your set up again? I want to mimick yours to get 4.2ghz stable when I get to build my i7 gig.

you multi is holding you back use the 19X and crank the bclk

BCLK Is Unstable

I have already mentioned this before in this thread but I can't push the BCLK higher than 180 even if I hit my processor and QPI with 1.35 V or as high as 1.45 V it just freezes. I already tried low multi's like 19, 18, and 12 and it doesn't work because BCLK won't raise higher.

I'm waiting for the new power supply to come in so that I can swap it out for the current one that is giving me unstable +5 V rail voltages hoping that the new one might improve my luck.



New Power Supply - Corsair 750HX

[H]ard|Forum > [H]ard|Ware > Power Supplies - 0.30 V drop only on +5V rail during Prime95 - Is this normal?

I ordered a Corsair 620HX power supply but decided that with SLI there wouldn't be enough head room so I went with the Corsair 750. Long story short from the other thread is that I ended up on purchasing a 750 W power supply in preparation for possible 2x SLI graphics card configuration in the future.

CORSAIR CMPSU-750HX 750W ATX12V 2.3 / EPS12V 2.91 SLI Ready CrossFire Ready 80 PLUS SILVER Certified Modular Active PFC Power ... - Retail
Cost: $164.99

 

[JakFrost]

I would say you have very poor mating surfaces, to start. Check for flatness using a straight edge, both the bottom of the cooler and the top of the heat spreader. Likely both, but for sure one or the other could stand to be lapped, using a known flat surface (ie., panel of glass) and fine grades of sandpaper starting at (depending who you read) 400-600-800 and going up to at least 1000 if not higher. Search around, you will find descriptive articles on the process.

I don't have the reference at hand, but I read a review of your cooler which was much less than flattering.

FWIW, my new (48 hours) chip is a C0 and on a Gigabyte mobo and so not necessarily a direct comparison, at 3.6 I'm full load temps not higher than 75 using an unlapped (also cpu unlapped) TRUE 120 Black and 2 1600rpm fans in push-pull. I am not yet stable but have been to 3.8 and not seen 80C. vcore 1.30 now 1.35, again I recognize not necessarily direct comparison to your system.

I saw the recommendation for my cooler as a good value for the low price and I thought that 4-heat pipe direct contact would be a good idea. Turns out that it isn't so bad now with the intake fan and it is giving me close to the same temperatures that other people are getting with their other coolers.

I'll read around about lapping the heat sink to see if I can get some kind of flattness to it but frankly after this last application of AS5 and intake fan it is going pretty well. If I get bored one of those days then I'll take it off and polish it out flat. I've seen some people report 5+ C temp drops after lapping the heat sink and that is good news.

 

[Flexion]

Jak,

Looking at your case, if you're looking to shave a couple of degrees C off, you can reverse the airflow by changing both your new front intake fan and rear exhaust fan, to becoming a rear intake and front exhaust. You would probably also re-mount your cooler to flow from rear to front too.

Lian Li, did this on a few of their smaller cases, and I found it odd at first, but I actually had my D0 running in one of those for a few weeks.

Alot of the low temp air cooled results are running on test bench type setups that are open air. Don't pay too much attention to those. Also, in my quest for OC'ing my D0 I found alot of posts by another [H]'er named Eva2000. Though most of his posts are pertinent to the DFI UT x58 mobo, his info on memory and CPU types are useful.

My D0 is a funny one too. I could make it up to 4.0- 4.2GHz, but had to run with my case open on a Thor's Hammer in order to keep things in the 85C range HT on all cores maxed. Some D0s like to having their Vcore and QPI volts at the same volts. So going higher on QPI isn't always the key. I found my processor didn't like QPI higher than 1.3, and clocked higher with lower volts. It got just as hot using lower volts..

I decided to go with watercooling, and have not been running my i7 for a few weeks just so I can set my system up the way I want.

There is also the possibility that the heatpipe orientation isn't optimal. I found that when my heatpipes were horizontal (end to end) rather than vertical, I had a 3C drop as well. I didn't draw any conclusions, because a change like that could simply be attributed to a core to cooling surface mating issue.

Also, it seems like you've been reapplying TIM and re-mounting quite frequently. AS5 needs many hours of thermal cycling to hit optimal temps.

 

[RADEoN]

Holy shit Jak, did you seriously use that "extra thick" as5 spread? If so, that' why your temps are so high.

 

[mike311271]

Hay Guys im new to this overclocking lark..lol , I have just built a core i7 setup and 4ghz was easy, my setup is:
1 nzxt lexa case
2 asus p6t v2
3 core i7 920 Do 2.66ghz at the moment overclocked to 3.50ghz
4 ocz reaper memory 3 x 1gb 1866
5 samsung 750gig sata 2 hard drive
6 nero cpu cooler
7 tagan 900w mod psu
8 sapphire 4870x2 (This as just been RMA Due To Memory Fail)
9 Dell 2408 Ultrabright 24" lcd display
10 logitech lx10 keyboard and mouse combo

my ? i have noticed my motherboard temp go to 60c and above, i emailed asus and they told me that this motherboard will go as fare as 70c. some1 told me that if i put the sapphire 4870x2 in pci-e slot 2 this will be away from the northbridge enc cooling motherboard. not been able to try this yet as due to graphics card going back to sapphire, at the moment the card i have got for gaming is a sapphire hd3450 it will play IL2 brilliently anything else forget it....lol

my cpu at 4gz idle at 48 and when playing prototype max settings everything full res at 1920 x 1200 after 1 hr maxed at 65c - nothing changed in bios set clock to 200. At 3.50ghz idle at 30 maxed at 51 so i have kept this setting just incase my temps are bad

 

[stiltner]

I agree with what was stated by one other person about mulitiplier limiting.

I'd also lean towards the board being a flake possibly. But what do I know, the last Asus board I admit to buying was a P3V4X (you might remember that bastard child if you been around long enough).

I'm not beating up on Asus here, but if you can't get higher than that you have either reached the limit of your cpu, board, or memory, or you're just plain unlucky.

Me personally, I'd hit up someone else with a i7 system, take my CPU and memory and slap it in another board and try it. Thats just me though, that could possibly help narrow down what might/might not be your weakness point and see if you can adjust from there.

Regardless if you're stable @ 3.8, you're still talking almost a 50% OC, so its hardly something to scoff at you realize right?

 

[JakFrost]

you can reverse the airflow by changing both your new front intake fan and rear exhaust fan

Eva2000. Though most of his posts are pertinent to the DFI UT x58 mobo, his info on memory and CPU types are useful.

There is also the possibility that the heatpipe orientation isn't optimal.

Also, it seems like you've been reapplying TIM and re-mounting quite frequently. AS5 needs many hours of thermal cycling to hit optimal temps.

I might want to try that but frankly I don't know how much of an improvement that would make especially since the back of my case blows into a wall and using that as a intake would probably blow hotter air inside instead of cooler air from the center of the room. My room is in the sunny corner of the house so the walls and ceiling heat up more in this part of the house.

Holy shit Jak, did you seriously use that "extra thick" as5 spread? If so, that' why your temps are so high.

No, that was the case before I allied AS5, even with the ceramic paste before that, and even with a super-thin layer of AS5 that caused no-contact zones. The thick later is probably a big improvement now that all of the heat pipes that are uneven make contact with the heat spreader.

I agree with what was stated by one other person about mulitiplier limiting.

Regardless if you're stable @ 3.8, you're still talking almost a 50% OC, so its hardly something to scoff at you realize right?

I'm hitting 3.6 GHz and it is mostly stable, except the last time I tried to run a stress test it barfed after 2-hours. Probably due to power supply +5 V rail instability as I found out afterward. The new Corsair 750HX power supply should be here on Monday and it should stabilize the power in my system to start some more testing.

After it comes in I'm planning on starting from scratch on my overclocking efforts. My first step is to test the maximum BCLK so I will set the multi to the lowest 12x and then bump up the voltage on CPU and QPI to 1.35 or even 1.40-5 V to see if I can get this system to not freeze at BCLK 200. So far it has been very stable at freezing at 200 immediately and freezing after a few minutes at 195. If I can't get these numbers I'll see what it takes to make it work at 185 or even 190 if I can get this lucky. That is my only option since like I mentioned above, on this motherboard and BIOS version setting the Multi to 21x turns on Turbo even if it is disabled in the BIOS.

We shall see...

PS: Still, I've been running this system at 3.60 GHz at 1.20 V and I'm happy with how it performs. I did some STALKER Clear Sky gaming and it does very well along with some MeGUI x264 8 multi-threaded encoding and it is very fast. The SSD makes it super responsive and Windows 7 is a pretty good OS from my short usage of it. Happy all around, except for having to deal with such issues with overclocking.

 

[keto]

my cpu at 4gz idle at 48 and when playing prototype max settings everything full res at 1920 x 1200 after 1 hr maxed at 65c - nothing changed in bios set clock to 200. At 3.50ghz idle at 30 maxed at 51 so i have kept this setting just incase my temps are bad

65C at load is well within acceptable, your cooling is doing its job. You would be OK even in low-mid 80's. I would stress test with Prime95 blend 8 iterations tho, that will jack you up higher in temp than any game. See where you top out for temps and adjust accordingly.

I now have my C0 stable at 3.760 (21x179) with HT on. Still have not seen 80C (getting close tho, 78C) under load - again, on air TRUE 120 with 2 1600rpm fans in push pull and no other cooling, tho the case is open at this point. Noise is not objectionable at all. 21 is obviously 20x + turbo - Jak, I forget what your objection to turbo was...? I also think that, for your purposes on air, chasing bclk is a red herring....leave that multi up. Be a different story if you (we) were on a more intense cooling solution. Most of my extensive reading AND my experience so far point in that direction. Best of luck as you chase the speed

 

[JakFrost]

Well, I got the new Corsair 750HX power supply today and I am very happy with it. I tested it and it is okay, installed it, and ran a Prime95 stress test on it for a few hours to make sure that it works okay. Everything is fine and there were no problems with any voltage drops during the stress test on any of the rails.

I love the packaging with the velvet bag for the power supply and the cables. The thin black ribbon cables make cable management a breeze and I don't need to go crazy tying the cables together as before because the case is neat right away.

I also replaced my SATA cables with new blue ones with a metal latch to get a secure lock on the cables but found that the Intel SSD doesn't use the click lock and the crappy JMicron internal ports are of the SATA I variety and don't even allow cables with click-locks to connect.

I'll post pictures and details of this power supply tomorrow. I'm planning a full CPU and GPU load test along with hard drives and optical drives to see if I can get maximum power draw out of it. Should be fun.

PS: Bad news, one of my Western Digital Caviar 500 GB drives died today. This happened after I connected it to the new power supply but I have been getting failure events on some of hard drives for a few weeks now. I think that the old power supply just was killing my drives and it finally finished this one off. The SMART Status today changed on it to BAD so I used WD's web site to do a quick Advanced RMA on the drive and I should have a replacement in a few days. Good thing that WD gives such long warranty even on OEM drives, because the production date on the drive is 2006-08-15.

 

[JakFrost]

21 is obviously 20x + turbo - Jak, I forget what your objection to turbo was...? I also think that, for your purposes on air, chasing bclk is a red herring....leave that multi up.

I'd love to leave the Multi at Auto for Turbo 21x multi-core and 22x single-core overclocking but I found my system to be unstable once it hits and stays at Turbo 22x.

I also didn't experiment with trying to find out a stress test suite that would focus on only one core to force the system to run at Turbo 22x multi for extended time since when I try Prime95 1-thread I still only get 21x multi out of it. Only when the system is idle and I open up a single app it hits 22x for a few seconds before it goes back to 21x.

I know that I could disable the other 3 cores in the BIOS to test 22x single core but that wouldn't be a realistic test since the other cores would be artificially unused and the test results would not apply once those cores are re-enabled.

 

[MittWaffen]

http://www.hardocp.com/article/2009/03/15/asus_p6t/6

Go there use settings close to that HARD did and you'll be stable. I have a d0 and still am close to C0 voltages on this board.

 

[MittWaffen]

Correction, you can tighten cpu voltage after you follow that guide which will stabilize your system easily with a high overclock. and use V-tune from there under load.

 

[JakFrost]

[H]ard|Forum > [H]ard|Ware > Power Supplies - 0.30 V drop only on +5V rail during Prime95 - Is this normal?

New Power Supply

Here are the pictures of the load testing on the new Corsair HX750W power supply.

This power supply already seems more efficient since on CPU Load my old OCZ PowerStream 520W pulled 384 W but this new one only does 326 W, for a difference of 58 W. On Idle the old one did 234 W and new one 200 W for a difference of 34 W. Overall it seems more efficient.

I did not run a CPU & GPU Load test on the old power supply but on this one I did a Prime95 8-threads and also did 3DMark Vantage GPU Test 2 - New Calico and hit a peak of 424 W.

One thing about this power supply is that the +5 V rail and the others are absolutely stable, unlike my other power supply that had a 0.30 V drop in the +5 V rail under load and probably further fluctuations that I didn't notice. This was the most important reason for this power supply.

I included the output from my APC Back-UPS XS 1500 VA showing the various power usage of the computer, monitor, router, cable modem, and desk lamp during the different stress test scenarios. Basically the numbers are Idle 302 W, CPU Load 423 W, and CPU + GPU Load 510 W.


750 W Is Too Much Honestly

I think that the 750 W power supply is too much honestly. The peak CPU + GPU test showed only 423 W and CPU 326 W for a 98 W difference. Now if we include a second video card into this picture it would likely pull less than 98 W because in SLI mode the second card is never fully used so it doesn't draw so much more power. At the maximum my system with nVidia GTX 260 in SLI would draw 521 W which would be handled just fine by the Corsair HX620W model. In reality when doing just the GPU load tests that I didn't post the power draw is around the ~330 W for most of 3DMark Vantage and with SLI it would be less than 428 W so a 620 W power supply would be more than enough.

At this point I paid the extra $25 ($45 with rebate) for the 750W power supply and frankly it is not worth it! However, I already opened and installed this power supply and a bit of it got a little scratched on the side so I'm stuck with it. The other power supply is already going to be delivered tomorrow so I can Refuse Delivery on it so it goes back and I get a refund.

If I just ran the full CPU + GPU Load test on the old power supply and saw the ~482 W result I would have probably went with the 620 W model and it would have been enough.

I'm sure that having the extra 130 W head room with this 750 W model over the 620 W model is not a bad thing to have but if you consider that it cost me $45 more for a ephemeral benefit then it was not worth it. Especially when normal maximum usage will usually be either CPU load with 326 W maximum draw or GPU with slight CPU load which ends up almost the same at ~330 W. With SLI that would be the same for CPU load and GPU load with a maximum of ~428 W. So a 620 W power supply should handle this type of a load just fine with 200 W head room. Now I'm stuck with a 750 W power supply that is only loaded at 200 W (26%) when the computer is idle, 326 W (46%) when CPU loaded or doing gaming, with a possibility of SLI for ~428 W (57%) load. That is a huge waste of potential for this power supply since it will never go above 60% usage.

I'm sure that some of you will point out that the efficiency of this power supply is the best at roughly 50% load so that it would be the most efficient when it is loaded but frankly the little ~3% change in efficiency is so small that any energy savings attributed to it are probably fractional compared to the extra $25 ($45 with rebate) costs of it. It would probably take months of non-stop usage at 50% load to get the money back and nobody does that.

I should have listened to some of the smarter reviewers like Christoph Katzer from AnandTech who say that the latest craze with huge power supplies is just an unnecessary marketing fad and a race for watts to nowhere that nobody really needs. Word of warning to you folks, don't overbuy a power supply that you don't need.

The only saving grace for the Corsiar HX750W power supply is that it includes 4-PEG (6/8-pin) connectors for the PCI-e graphics cards and new cards that require the 8-pin PEG connector where as the Corsair HX620W only includes 2-PEG (6-pin) connectors. That is about the only reason that I am not quite angry about this power supply since I know that if I go with SLI in the future or when I replace my GTX 260 with something else this power supply will be ready.



Results From Corsair HX750W

Kill-a-Watt Power Meter
Intel Core i7 920 2.66 GHz D0
Asus P6T 0603
G.Skill 6GB DDR3-1600 C8 2T
Xigmatek HDT-1284EE
Corsair 750HX

Idle - 200 W

CPU Load - 326 W

CPU + GPU Load - 424 W

Power Usage - Idle - Kill-a-Watt 196 W, APC UPS 302 W

Power Usage - CPU Load - Kill-a-Watt 326 W, APC UPS 423 W

Power Usage - CPU + GPU Load - Kill-a-Watt 424 W, APC UPS 510 W

Lian-Li PC-V1200 Plus Case - Corsair 750HX - Drive Bay

Lian-Li PC-V1200 Plus Case - Intel Core i7 920 2.66 GHz D0, Asus P6T 0603, G.Skill 6GB DDR3-1600 C8 2T, Xigmatek HDT-1284EE, Corsair 750HX - Case Inside

 

[keto]

I'm up to 21 X 185 = 3.885 on my C0, idle 42C, loaded 81C, 1.4125vcore.

 

[JakFrost]

I'm up to 21 X 185 = 3.885 on my C0, idle 42C, loaded 81C, 1.4125vcore.

Did you have to change any other voltages except for CPU, like QPI?

 

[keto]

Yes. I have a different mobo, Gigabyte EX58-UD5 and don't know if all the bios terminology is the same as Asus'.

I'm running vdimm 1.660, QPI-vtt (aka Uncore) 1.460, CPU-PLL at 1.880, Load Line enabled. There are several other changes available, I don't recall all the names offhand (IOH Core, ICH Core, QPI PLL are some) but I did do 1 thing - take everything off auto and put in a number, even if it's the same number as set by auto. I don't know if it's just the Gigabyte boards or all '58 chipsets, but some of the voltages will spike if left on auto.

And I need to amend my prior post, 185x21 didn't fully stabilize, but 182x21 did. Good enough for me, though I still need to go ramp up my RAM speed to 10x, as opposed to the 8x I am running now.

 

[JakFrost]

New BIOS 0707 - System Freezing when Idle

Just a little update, the new Asus P6T BIOS 0707 makes this system freeze when idle. This is likely due to a change that Asus made to their BIOS and the freezing is likely the symptom of under voltage. I will start the process of diagnosing the subsystem that is causing the freezing only if I can figure out a repetitive way of making the system freeze.

The system was perfectly stable with the 0603 BIOS but now I cannot downgrade after I updated it so I need to fix the problem.

BIOS Setting Screens

Below are the BIOS settings screens for this system.

Asus P6T BIOS Downloads - 0707, 0603, etc.

 

[mastrdrver]

Do you have the cd that came with the board? There is a windows app that lets you install an earlier dated bios than the one you have currently.