I7 950 overclock (some hand holding needed)

[Skwall]

Hello everyone, I'm a bit new to the over clocking world though I've been an avid builder of my (and my friends) gaming PC's for quite some time now. I finally decided to venture down the road of overclocking and its been a bit bumpy. I don't know how many times I've cleared the CMOS, but I'm sure not as many as some.

I've been reading this forum along with watching YouTube videos and my own trial and error but here's what I've got so far. I've been able to get from 3ghz to 3.6ghz stable with no voltage changes to neither the vCore or VTT, all voltages are on auto at this point. Though, a 28C idle tells me there might be more. So I went back into the BIOS and made more changes to the FSB, upped it to 200mhz with a multiplier of 20, giving me 4ghz. It didn't get into windows before a BSOD. So, back to the BIOS, upped the vCore to 1.3v then again to 1.35v and the VTT +50mv (1.15v): My bios doesn't show me the VTT voltage just options to increase by a set amount. F10'd to save and it booted into windows and sat at 32C idle CPU.

I took some screen shots from the EVGA E-LEET Tuning Utility that is pretty much CPU-Z, but you can alter settings which I don't use it for. With the current settings I get a BSOD when I start Prime95 and all of the cores start to load up. Looking at the temps I feel it might be a voltage issue but I'm a novice that's why I'm here.


Computer:
MOBO - EVGA X58 SLi3
CPU - i7 950 @ 4ghz
Cooler - Cooler Master 212+ EVO
Ram - Patriot Viper II Sector 7 Edition 6GB (3 x 2GB)
GPU - EVGA GTX580 1.5GB (stock clocks)
PSU - Corsair 750w 80 plus certified
HDD - WD 250gb 3gb/s
Chassis - Cooler Master HAF 932

The first thing I noticed was the fact that the NB was running shy of 200mhz, Voltage issue?


This RAM is set to run at [email protected] from the factory, though these auto timings are different. I was torn about manually setting them as I was focusing on more the CPU at this point, but any suggestions would be welcomed.


You can see the the changed in voltage I spoke of earlier.


I noticed here that the VTT voltage seems to have varied from the earlier mentioned 1.15v.



So that's it, that's what I have to work with, I built this PC in Feb of 2011 so its just over a year old.

 

[EarthDog]

Welcome!

1. Idle temps do not matter. You will want to know load temps (Prime95, Torture test, Small FFT).
2. Increase Vtt a bit... to 1.25v and see if that helps.

In my signature is an overclocking guide you should read. You are well on your way and I think by reading that, should help you out a bit.

 

[Skwall]

Thanks, that was a quick response. In the BIOS it showed my VTT voltage at 1.10, so I added the 50mv that made it 1.15v but the snip of the monitor tab in the EVGA utility shows it at 1.23v. Is that registering wrong or is CPU VTT different than VTT?

Also at 3.6ghz the CPU temp during a 2hr Prime95 run never got above the high 40's to mid 50's. With the stock cooler at 3.2ghz (turbo mode) it was at 75C I think. I'm going to read the OC guide now to see if it can answer some other questions before asking anymore.

Thanks!

 

[Skwall]

I could be about to blow something up, not sure.

Prime95 running - Second thought, not sure how close is too close to TJ Max while running this kind of test.

 

[EarthDog]

85C is about as close as you want to come... not so much because it will kill your chip, but because they will sometimes show instability around that range+.

 

[Skwall]

So I guess 3.8 is my stopping point. I made it stable at 4ghz but I didn't like the temps I was seeing. I guess I could always put a pulling fan on the cooler I have, currently only has a pushing 120mm.

 

[psionic98]

that heatsink isn't really made for handling the output of the 1366 chip when overclocked really.. its better than a stock heatsink, but to really hold these chips at high overclocks a high end air cooler is needed (if you wanted to stay with air that is)

 

[EarthDog]

212+ especially with push/pull should be able to rock 4Ghz... I mean hell, it handled 5 on the 2600k with HT.

 

[Skwall]

that heatsink isn't really made for handling the output of the 1366 chip when overclocked really.. its better than a stock heatsink, but to really hold these chips at high overclocks a high end air cooler is needed (if you wanted to stay with air that is)

Yeah, but for 35 bucks I think its an amazing product. I decided to just tag it on the order for the gtx 580 I bought a few weeks ago. I was still rocking the stock cooler.

I had considered going with one of those corsair setup's that area already put together and in a closed loop. But I'm not sure if I would need more than a 750w PSU. It's a good PSU made with the high quality Japanese capacitors, though I don't know how much that means next to wattage requirements for running the pump. If you guys have any thoughts on the matter it's always appreciated.

This would be my choice.
http://www.newegg.com/Product/Product.aspx?Item=N82E16835181016

 

[Skwall]

212+ especially with push/pull should be able to rock 4Ghz... I mean hell, it handled 5 on the 2600k with HT.

Well as of now, the vCore is at 1.3v and the VTT is set to +100mv. To make it stable at 4ghz I moved the vCore up to 1.35 but the core's got quite hot during Prime95. Does that make it acceptable for every day use and gaming?

I guess if Prime95 is just to test an absolute worst case scenario and my computer doesn't explode at 4ghz it just run's between 85-90c during an 8 thread torture test. My current settings run at 80-82c during the same test but never gets above the high 60's when actually gaming.

 

[EarthDog]

Here is the thing.. P95 stress testing will be 10-15C hotter than most anything else you do on your PC. The high 60's is nothing when gaming. I say you can rock 4Ghz if you so choose.

 

[Skwall]

Here is the thing.. P95 stress testing will be 10-15C hotter than most anything else you do on your PC. The high 60's is nothing when gaming. I say you can rock 4Ghz if you so choose.

That's a reassuring thought. I will crank it back up to 4ghz this weekend and run realtemp for a bit while gaming to see what those numbers look like.

 

[Skwall]

Ok, I had another 120mm Thermaltake fan that puts out up to 97.5cfm, thankfully it has a control switch on it. So I put it on the heat-sink set to pull, its pretty much lined up with the 200mm exhaust fan so that kinda worked out.

It's stable enough to run a game but after a few minutes of P95 it BSOD's. The cores are only at 80-82c when this happens.

Here's the results while running a game.


I'll play for a while and see what happens.

 

[Skwall]

I would think more voltage some where would give more stability but I'm not sure what the limitations are and how much further I can push it. Is heat the deciding factor?

 

[redduc900]

Fill out this BIOS template for the EVGA X58 w/ all of your 4.0GHz settings (just disregard the settings in the template and replace w/ your own)...

Frequency Control  
CPU Clock Ratio ( 21X )  
CPU Host Frequency (Mhz) ( 183 )  
CPU Uncore Frequency (Mhz) ( 20x )  
PCIE Frequency (Mhz) ( 100 )  

Memory Feature          
Memory Speed ( Standard )  
Memory Control Setting ( Enabled )  
Memory Frequency ( 2:8 )  
Channel Interleave Setting ( 6 Way )  
Rank Interleave Setting ( 4 Way )  
Memory Low Gap ( Auto )  
tCL Setting ( 9 )  
tRCD Setting ( 9 )  
tRP Setting ( 9 )  
tRAS Setting ( 24 )  
tRFC Setting ( 88 )  
Command Rate ( 1T )  

Voltage Control  
EVGA VDroop Control ( With VDroop )  
CPU VCore ( 1.3 )  
CPU VTT Voltage ( +200 ) 
CPU PLL VCore ( 1.650 )  
DIMM Voltage ( 1.65 )  
QPI PLL VCore ( 1.200 )  
IOH VCore ( Auto )  
IOH/ICH I/O Voltage ( Auto )  
ICH VCore ( 1.2v )  
PWM Frequency ( 800 )  

CPU Feature  
Intel SpeedStep ( Disabled )  
Turbo Mode Function ( Disabled )  
CxE Function ( Disabled )  
Execute Disable Bit ( Disabled )  
Virtualization Technology ( Disabled )  
Intel HT Technology ( Enabled )  
Active Processor Cores ( All )  
QPI Control Settings ( Enabled )  
QPI Link Fast Mode ( Enabled )  
QPI Frequency Selection ( 4.800 GT/s )

 

[Skwall]

Code:
---------
Frequency Control
CPU Clock Ratio ( 20X )
CPU Host Frequency (Mhz) ( 200 )
CPU Uncore Frequency (Mhz) ( 16x (3205mhz ))

PCIE Frequency (Mhz) ( 100 )

Memory Feature
Memory Speed ( Standard )
Memory Control Setting ( Enabled )
Memory Frequency ( 2:8 )
Channel Interleave Setting ( 6 Way )
Rank Interleave Setting ( 4 Way )
Memory Low Gap ( Auto )
tCL Setting ( 9 )
tRCD Setting ( 9 )
tRP Setting ( 9 )
tRAS Setting ( 24 )
tRFC Setting ( 88 )
Command Rate ( 1T )

Voltage Control
EVGA VDroop Control ( Without VDroop )
CPU VCore ( 1.3125 )
CPU VTT Voltage ( +100 )
CPU PLL VCore ( 1.8 (auto))

DIMM Voltage ( 1.65 )
QPI PLL VCore ( 1.100 (auto))

IOH VCore ( Auto )
IOH/ICH I/O Voltage ( 1.5v (Auto))
ICH VCore ( 1.1v (auto))
PWM Frequency (don't see this.)

CPU Feature
Intel SpeedStep ( Disabled )
Turbo Mode Function ( Disabled )
CxE Function ( Disabled )
Execute Disable Bit ( Enabled)
Virtualization Technology ( Disabled )
Intel HT Technology ( Enabled )
Active Processor Cores ( All )
QPI Control Settings ( Enabled )
QPI Link Fast Mode ( Enabled )
QPI Frequency Selection ( 4.800 GT/s )

 

[Skwall]

As I continue reading and thinking about it, I feel like I might be on to something. After 3 minutes of P95 the cores hit about 80c or so and that's when the BSOD happens. Before BSOD my desktop gadget that loads when I turn my computer on crashes. I can tell before it happens and the last time I was able to stop P95 and prevent the crash. I'm thinking I need to increase the MIC voltage. Mabe +100mv more, +200mv total.

 

[Skwall]

So I moved the VTT voltage to +200mv and the CPU down to 1.3 even. Still not 100% stable there. Even though temp's are OK when the BSOD happens.

 

[Skwall]

Ok so I've bumped the CPU Core voltage to 1.325 and left the VTT at +200mv. Seems to be stable running P95, after 5-6 minutes it does get up to 85c. Also disabled the "Execute Bit".

 

[EarthDog]

Why did you disable that? It shouldnt have anything to do with your stability.

That said, Those temps while stress testing are fine. My preference is after small FFT runs for a few hours, go to blend and see if that works for a few more hours.