Overclock Core 2 Quad on Nvidia 7xxi Motherboard

[Ladle]

Hello community, I've been having trouble getting a stable overclock on my PC and was hoping to get some direction from someone more experienced with system overclocking. This is my first attempt at trying to overclock a system, and after reading up on it quite a bit, I gave it a try. Some of the more successful details of my attempts are below.

My System:
Intel Q9450 2.66 GHz with Cooler Master TPC 812 Heat Sink
EVGA 780i SLI (132-CK-NF78)
8GB DDR2 PC6400 (4 x 2GB – Kingston ValueRAM KVR800D2N5K2)
EVGA GeForce GTX 460 (main video card)
XFX GeForce GTS 250 (for PhysX processing)

First attempt:
Based on “[p]'s 780i Overclocking Guide” (http://www.evga.com/forumsarchive/tm.asp?m=222688). I made many attempts at changing CPU core voltage in order to get a stable system. However, I was unsuccessful, below are the settings that seemed to last the longest under stress testing. I used Prime95 for stress testing and CPUID HW Monitor for temperature measurements.

Goal
QDR: 1666 MHz
DRAM: 800 MHz

Voltages
CPU Core: 1.3750 V
CPU FSB: 1.4 V
Memory: 1.85 V
SPP: 1.4V
MCP: 1.525V
SPP<->MCP: 1.25V

Max Temperature
System: 63 C
CPU (SPP?): 107 C
CPU1 (MCP?): 78 C
Core 1: 51 C
Core 2: 50 C
Core 3: 58 C
Core 4: 52 C

Results
Prime 95 ran for about 2:53:00


Second Attempt:
Also based on “[p]'s 780i Overclocking Guide”. I attempted to lower the FSB QDR to something that appeared to be much more attainable. 400MHz FSB is afterall what the motherboard is rated for by the manufacturer. I made many more attempts at changing CPU core voltage. Below are the settings that seemed to last the longest under stress testing. I used Prime95 for stress testing and CPUID HW Monitor for temperature measurements.

Goal
QDR: 1600 MHz
DRAM: 800 MHz

Voltages
CPU Core: 1.34375 V
CPU FSB: 1.4 V
Memory: 1.85 V
SPP: 1.4V
MCP: 1.525V
SPP<->MCP: 1.25V

Max Temperature
System: 58 C
CPU (SPP?): 107 C
CPU1 (MCP?): 78 C
Core 1: 51 C
Core 2: 50 C
Core 3: 58 C
Core 4: 52 C

Results
Prime 95 ran for about 2:53:00


Third Attempt:
I than found this post which suggested starting with the motherboards “auto” settings for some of the values (http://www.evga.com/forumsarchive/tm.asp?m=499530&mpage=1&key=%F1%B9%BD%8A).
I decided to go back to the drawing board and used Memtest86+ to test two of my 4 memory sticks to make sure they were working properly. Each was tested by individually in the same DIMM slot using the manufacturers recommended settings in the BIOS (5-5-5-18 @ 1.80V). 7 full passes of each showed 0 errors. I then proceeded to stress test the system with stock BIOS settings, and only these two memory sticks. A one hour OCCT stress test, and ~2 hours of Prime95 showed no failures.

I then began trying to overclock my CPU again by slowly increasing the CPU core voltage from 1.20 V and got as high as 1.4250 V. Unfortunately, I was unable to get a stable system for more than 20 or 30 minutes.

Goal
QDR: 1600 MHz
DRAM: 800 MHz

Voltages
CPU Core: 1.4250V
CPU FSB: Auto(1.30 V)
Memory: 1.8 V
SPP: Auto(1.3 V)
MCP: Auto(1.5 V)
SPP<->MCP: Auto(1.2 V)

Results
Prime 95 ran for about 0:20:00
OCCT for about 0:15:00

I didn't want to attempt raising the CPU core voltage any higher than 1.4250 V, and attempting to alter any other voltage value would be more or less a random guess on my part. Instead I thought I would post here in an attempt to get a more guided approach. Can anyone with provide me with some suggestions or a direction for what I should be looking at next?

 

[wonder44]

Hello community, I've been having trouble getting a stable overclock on my PC and was hoping to get some direction from someone more experienced with system overclocking. This is my first attempt at trying to overclock a system, and after reading up on it quite a bit, I gave it a try. Some of the more successful details of my attempts are below.

My System:
Intel Q9450 2.66 GHz with Cooler Master TPC 812 Heat Sink
EVGA 780i SLI (132-CK-NF78)
8GB DDR2 PC6400 (4 x 2GB – Kingston ValueRAM KVR800D2N5K2)
EVGA GeForce GTX 460 (main video card)
XFX GeForce GTS 250 (for PhysX processing)

First attempt:
Based on “[p]'s 780i Overclocking Guide” (http://www.evga.com/forumsarchive/tm.asp?m=222688). I made many attempts at changing CPU core voltage in order to get a stable system. However, I was unsuccessful, below are the settings that seemed to last the longest under stress testing. I used Prime95 for stress testing and CPUID HW Monitor for temperature measurements.

Goal
QDR: 1666 MHz
DRAM: 800 MHz



Voltages
CPU Core: 1.3750 V
CPU FSB: 1.4 V
Memory: 1.85 V
SPP: 1.4V
MCP: 1.525V
SPP<->MCP: 1.25V

Max Temperature
System: 63 C
CPU (SPP?): 107 C
CPU1 (MCP?): 78 C
Core 1: 51 C
Core 2: 50 C
Core 3: 58 C
Core 4: 52 C

Results
Prime 95 ran for about 2:53:00


Second Attempt:
Also based on “[p]'s 780i Overclocking Guide”. I attempted to lower the FSB QDR to something that appeared to be much more attainable. 400MHz FSB is afterall what the motherboard is rated for by the manufacturer. I made many more attempts at changing CPU core voltage. Below are the settings that seemed to last the longest under stress testing. I used Prime95 for stress testing and CPUID HW Monitor for temperature measurements.

Goal
QDR: 1600 MHz
DRAM: 800 MHz

Voltages
CPU Core: 1.34375 V
CPU FSB: 1.4 V
Memory: 1.85 V
SPP: 1.4V
MCP: 1.525V
SPP<->MCP: 1.25V

Max Temperature
System: 58 C
CPU (SPP?): 107 C
CPU1 (MCP?): 78 C
Core 1: 51 C
Core 2: 50 C
Core 3: 58 C
Core 4: 52 C

Results
Prime 95 ran for about 2:53:00


Third Attempt:
I than found this post which suggested starting with the motherboards “auto” settings for some of the values (http://www.evga.com/forumsarchive/tm.asp?m=499530&mpage=1&key=%F1%B9%BD%8A).
I decided to go back to the drawing board and used Memtest86+ to test two of my 4 memory sticks to make sure they were working properly. Each was tested by individually in the same DIMM slot using the manufacturers recommended settings in the BIOS (5-5-5-18 @ 1.80V). 7 full passes of each showed 0 errors. I then proceeded to stress test the system with stock BIOS settings, and only these two memory sticks. A one hour OCCT stress test, and ~2 hours of Prime95 showed no failures.

I then began trying to overclock my CPU again by slowly increasing the CPU core voltage from 1.20 V and got as high as 1.4250 V. Unfortunately, I was unable to get a stable system for more than 20 or 30 minutes.

Goal
QDR: 1600 MHz
DRAM: 800 MHz

Voltages
CPU Core: 1.4250V
CPU FSB: Auto(1.30 V)
Memory: 1.8 V
SPP: Auto(1.3 V)
MCP: Auto(1.5 V)
SPP<->MCP: Auto(1.2 V)

Results
Prime 95 ran for about 0:20:00
OCCT for about 0:15:00

I didn't want to attempt raising the CPU core voltage any higher than 1.4250 V, and attempting to alter any other voltage value would be more or less a random guess on my part. Instead I thought I would post here in an attempt to get a more guided approach. Can anyone with provide me with some suggestions or a direction for what I should be looking at next?

Yea this is odd. I have a Q6600 running at 3.6 stable with a voltage of 1.46v on my 780i. Your voltages seem a little high thoug. Also what power supply do you have? I have only changed the core speed and voltage and i am running stable. I got 48 hours on prime 95 without any issues. Maybe you just have an unlucky chip perhaps.

 

[Ladle]

Thanks for the reply. I am using a Zalman ZM750-HP 750W power supply.

When you say my voltages seem a little high, which ones are you refering to, just the CPU core voltage or something else as well? I have not really experimented too much with the other voltage settings.

I have since read that having four memory modules and multiple video cards can have an effect on the stability of an overclocked system, but I was unable to find a specific reason why. Is this something that can be compensated fore in the voltage to the motherboards northbridge which I beleive connects the CPU, primary graphics card and memory?

 

[wonder44]

I have 2 sticks of ram and one GPU. Id say it could possibly be that your GPU's are drawing too much power from the PSU which is leaving your CPU with not enough headroom to overclock higher. Although i could be wrong, however if you get the time you may try removing one card to see if that helps. Also the other voltages seem ok. Also i did have stability problems with the FSB at 1600mhz so i raised it to 1601mhz which seemed to help A LOT, which is rather strange. I recently just moved the voltage up to 1.47v however. your temps are fine. Doing better than me, i have watercooling and im still getting 55c load lol. Although your chip is smaller die. and considering that, your voltages will not need to be as high as mine. Id Overclock to 3.4ghz with a vcore of 1.40. if it is stable. Set CPU at 3.6ghz and try 1.43v

 

[Ladle]

Thanks for the suggestions. I've tried bumping the FSB up to 1601 MHz but there was not change in stability.

I then tried removing my secondary video card and started bumping the CPU core voltage up again. Using the same voltage settings as my "second attempt" above, I got as high as 1.4250 V for the CPU before I stopped with no increase in stability. It would appear, I'm not gaining anything by removing the second video card.

When I have some more time later this week, perhaps I'll try one video card with only two sticks of RAM. Something else in my system must be limiting the stability...

 

[wonder44]

Although it is possible your inparticular chip is not quite as good as the others. You must remeber, just because it is the same chip, does not mean it was made EXACTLY the same as another. For instance, some people could their Q6600's to 4Ghz stable, while others could barely get 3.4Ghz. So it just depends on how the chip was made, Even if it is the EXACT same model, although your chip should make it past 1600Mhz rather easily. So, i now suggest that is trying to get upto 1.44v slowly. You do not want to go above 1.45v for that cpu so its probably best to stay under 1.44 being 1.44v maximum for daily use. Another thing to take note, is that most applications in the real world will not stress a CPU to the limits such as intel burn test or prime 95. So basically if you can survive 10-15 runs on intel burn test with a standard memory option selected and/or 24 hours on prime95 you should be fine. My old celeron laptop froze running the 16th run of intel burn test. And i have never had any problems with it. And this is a laptop so its not overclocked and cannot be. So thus, leads me to believe that you should be fine if your making 24 hours on prime or 10-15 runs on IBT. So slowly raise the voltage up and upjust trying not to go above 1.44v. 1.44v at the maximum, even though intel suggests you do not go above 1.45, its best to always stay a little under the recommended just to be sure. So give it a shot, just make sure your temps are ok as you get higher with the volts. Also your CPU will be a lot faster at 3.4ghz than my Q6600 will be at 3.6Ghz. And my Q6600 at 3.6Ghz is really quite fast so im sure you will doing fine. Hope all goes well

 

[Ladle]

I have read that every CPU is different, I guess I was hoping that mine was capable of what I considered a moderate overclock.

Are you sure I should attempt to bump the voltage up even higher? Intel specifies that the maximum voltage for this chip is 1.3625V (http://ark.intel.com/products/33923/Intel-Core2-Quad-Processor-Q9450-12M-Cache-2_66-GHz-1333-MHz-FSB). I thought I was already getting quite high with 1.4250 V.

I'm going to try overclocking with one video card and the two memory modules I tested with memtest now to remove one more potential variable.

Thanks for the continued suggestions.

 

[nforce4max]

Is the LLC circuit still enabled or did you shader one of the resistors for the vdroop issue? I got the xfx version and learned a few things about these boards. Once you start jacking up the volts for the spp and the temps really start to rise performance can degrade while stability can be dicey.

XFX780i, q8200 at 3.26ghz 1.168v after vdroop, antec he550, and a ancient zalman cnps7000a. As for the ram I am using the gskill 1066kit, 4.4.4.12 20 at ddr2 874 (1.875v).


Check you board to make sure that it is flat and doesn't have a bend. Your SPP and MCP temps are to high. You won't get good clocks with those temps.

 

[Ladle]

I have not done any mods on my motherboard, but I have searched the vdroop mod you mentioned. It sounds like this is a work around for the board providing slightly less power to the CPU than what is set in the BIOS, is this correct? Could this be a factor in the overclock stability? Would a safer work around not be to just use a slightly higher voltage setting in the BIOS than what you need?

As for my motherboard being level, it appears to be. If you are suggesting that the SPP and MPP temperatues I listed in my post above are too high, then perhaps the heat sinks may not be properly seated. I'll have to take a look to see how easy it is to remove and re-attach them.

Physically modifiying the hardware is starting to get out of my confort zone, but I can still handle heat sinks. Resister shading is a bit more risky in my opinion.

Thank you for your suggestions.

 

[nforce4max]

The resister shading is easy and if it don't work it is easily undone. The reason why I mention the board's cooler is that some have gotten boards that had poor contact with the cooler. There was one that still had the plastic tape still on the core and so it didn't have contact with the cooler.

The reason why most did the vdroop mod was due to stability issues at higher clocks when the voltage dipped to low for the cpu to be stable. For me I did it to keep the voltage from dipping to low while undervolting.