Archive Home arrow Guides arrow Intel Sandy Bridge Overclocking Guide
Intel Sandy Bridge Overclocking Guide E-mail
Articles - Featured Guides
Written by Servando Silva   
Tuesday, 19 April 2011
Table of Contents: Page Index
Intel Sandy Bridge Overclocking Guide
Why Should I Overclock?
Overclocking Applications and Utilities
Processor Stability Testing
Understanding OC Variables
UEFI and Testing Methodology
Overclocking Frequency vs. Voltage
Overclock vs. Power Consumption
Overclock vs. Temperatures
Final Thoughts

Processor Stability Testing

Before starting, I must say stability is a concept that can be perceived in different ways depending on your needs. While some people think SuperPi runs are enough to check stability, others want their PC to successfully run 30 hours of OCCT's Linpack test. I'm gathering all these concepts into a PC stable enough to do daily applications no matter how hard they are because that's what we're aiming for. I would be unsatisfied if my PC suffered a BSOD while doing multi-tasking, playing my favorite game online or coding and rendering the next-day's project. What you must understand, is that every component needs to be tested in different situations to check if it's 99.9% stable. In this case, since we're overclocking our CPUs (mainly), my favorite tool is Prime95. Run it through a Blend Test to check both CPU and Memory stability, or run in place large FTT for maximum heat and power consumption. Usually, a couple of hours means your PC isn't prone to suffer BSODs, but I prefer to leave it running at least 10 hours before working. Usually, a night running Prime95 should be enough.

Since we're talking about 4 core CPUs here, I usually fire up Prime95 and start doing Blend tests. Prime95 allows you to check which CPU core is failing, and so you can add voltage or compensate the CPU knowing which core is the weakest of all. However, if you're just starting to overclock your CPU, you might want to test it with something lighter, and Cinebench R11.5 is a good choice to ensure CPU is (at least) stable enough to boot into OS and open your OC tools. Even Cinebench uses all your available cores; it won't define any kind of stability by itself.

SandyBridge_OC_Analysis_Prime95.PNG

Next step would be testing your RAM. RAM is a little bit more difficult to test because while you can spend 10 hours running stability tests, it could fail while opening a very simple application (e.g.: Photoshop, MSN messenger, etc.). What I do is: start with MemTest and put all unused RAM to test until it reaches 100% (at least). Many people think this is enough, but RAM can be a real pain if you don't stress it enough. Try opening Prime95 again and start Blend Test (uses lots of RAM). After a couple of hours running Prime95, you might want to check with your daily applications and GPU benchmarks like Unigine's Heaven 2.1. Trust me; RAM fails exactly when you don't expect it to fail, so it's better to make sure it will be stable enough for hard daily work.

One thing people hate is watching high temperatures on their PCs. If you're using Intel's stock cooler I would recommend that you not raise CPU vCore. You might want to give a read to our latest cooling articles to change you cooler to something a little bit more appropriate. My favorite temperature monitoring tool for Intel Core processors is Real Temp. This little application has a sensor test to check if any of your cores has a stuck sensor. I've also found this is very accurate software which is a MUST in this case. Below you can see an example of our Core i7 2600K at idle mode while Noctua's NH-D14 is sitting at the top of it. Ambient temperatures were below 25ºC, and our heatsink was good enough to keep it 1-3 degrees above ambient. The latest edition of this particular software offers CPU load and Power consumption readings. With these features we can isolate CPU power from all other components.

SandyBridge_OC_Analysis_Real_Temp.png

You might be wondering what the appropriate range of maximum temperatures is. Have a look at the image above. If you add "temperature" plus "Distance to TJMax" values, you'll find a TJunction value. For example: in this case 28+70=98. That's the maximum temperature your CPU will support before turning the computer off for self-protection. Of course, it's very unlikely to reach these temps, and motherboards normally have protection limits somewhere between 80-90 degrees. That's why it's actually very difficult to burn a CPU nowadays unless you're giving it too much voltage.

If you're overclocking a Sandy Bridge processor, you might want to keep your temps below 70-80 degrees at full load (Prime95 FTT test). Daily applications shouldn't stress it enough to pass 70-75 Celsius. Keep in mind CPU temperatures are meant to be kept below those limits at overclocking conditions. Thanks to the 32nm manufacture, you might be able to keep your CPU below 60 or even 50 degrees at factory settings (no OC, no extra voltage). Now that you've downloaded and understood your weapons, let's get a little bit more technical and analyze all the variables you should pay attention to while overclocking. In the next page I'll explain each variable and how to control it from your BIOS/OC utility.



 

Comments 

 
# RE: Intel Sandy Bridge Overclocking GuideJackNaylorPE 2011-04-21 12:37
Would have liked to see a comparison of setting the Turbo / Voltage to a fixed number versus using the Max turbo frequency / voltage offset method.
Report Comment
 
 
# RE: RE: Intel Sandy Bridge Overclocking GuideServando Silva 2011-04-21 13:53
Hi Jack. Please check our forum as I've just explained why I didn't test that way. It would need another article just for that, or at least, a new set of tests.
Take care.
Report Comment
 
 
# RE: Intel Sandy Bridge Overclocking GuideRobert17 2011-04-21 13:24
Nicely done. But you did make it seem like the "good old days" of mounting a PVC tank with a fish-tank pump feeding from an ice chest full of frigid water may be a thing of the past.

Not to get too far away from your core article regarding Sandy Bridge, but do you have any insight as to whether or not AMD will maintain OC potential in their new lineup?
Report Comment
 
 
# RE: RE: Intel Sandy Bridge Overclocking GuideServando Silva 2011-04-21 13:57
Yeah. I've been overclocking for the last 10 years and I know many others who have been doing it for 15 years or so.
You know that phrase: "Like the old days". I'm not saying I don't like this new way of overclocking. Actually, I like not to pass several hours to find such a nasty or complex configuration for my PC, and doing it the the old way. New OC tools and features help a lot when what you just need is to bump your PC speed and get back to work, but sometimes it was funny to test and read a lot for that.
Also, sub-zero overclocking is quite fun, but with Sandy Bridge they somehow killed it.

Regarding the new AMD processors, I still have no information about them, so I'm anxious to test one.
Report Comment
 
 
# RE: RE: RE: Intel Sandy Bridge Overclocking GuideDavid Ramsey 2011-04-21 14:30
The new way of overclocking-- raising the maximum multiplier use by Turbo Boost-- has one huge advantage over the old ways of increasing BCLK or the base multiplier: to wit, the processor can still downclock to low speeds when you don't need the performance. My 4.1gHz 980x always runs at 4.1gHz. A 4+gHz Sandy Bridge can idle at the same speed it does non-overclocked, saving a lot of power (and generating less heat).
Report Comment
 
 
# RE: RE: RE: RE: Intel Sandy Bridge Overclocking GuideRobert17 2011-04-21 15:51
I'm not complaining. Maybe just pining a bit. I set up a simple OC on my MB, seldom vary it due to the stability, change to game, pretty much ignoring the power consumption. My bad. What limited understanding I have of UEFI seems to indicate that rebooting to change configs may become simpler.

And certainly the advantages of having a MB/CPU combo that is "self-monitoring" power, thrust, pitch and yaw outweigh the "good old days". And yes, I mostly like automatic transmissions over three-on-the-tree these days as well.
Report Comment
 
 
# RE: RE: RE: RE: Intel Sandy Bridge Overclocking GuideServando Silva 2011-04-22 11:45
Completely true! As I've said. I'm not really complaining. I like the new way as it benefits final users. It's just they took away that "chilli spice" when overclocking. Also, they limited it to Unlocked processors and certain platforms.
Report Comment
 
 
# RE: RE: RE: Intel Sandy Bridge Overclocking GuideOlle P 2011-04-22 10:30
"... sub-zero overclocking is quite fun, but with Sandy Bridge they somehow killed it."

You can say that again! I read somewhere that Sandy Bridge reach its peak performance at about 20C. If you cool it more than that it won't reach quite as high speeds.
Report Comment
 
 
# RE: RE: RE: RE: Intel Sandy Bridge Overclocking GuideServando Silva 2011-04-22 11:48
Do you want me to say it louder? Yeah!
I've heard some processors do worst when going below 10-20C degrees. Again, it's good because now many users will be able to play and overclock without going extreme, but it won't be as interesting for extreme users.
Report Comment
 
 
# RE: Intel Sandy Bridge Overclocking GuideJackNaylorPE 2011-04-24 15:35
One other thing I'd love to see addressed ... POITA ain't I :) ..... the 2600k I assume is running hotter cause of the HT..... Prior to reading the article, I cam to the same conclusion that 4.4 Ghz was the sweet spot for the 2600k for 24/7/365. Since this was a B'day build for Son No. 3, w/ Spring break and all I haven't been able to get near the thing in a week.....I used the Asus BIOS Profile feature to store OC Profiles from 4.0 to 4.8 GHz w/ these temps on the SIlver Arrow cooler:

Max Core Temps under (Idle - P95 load)

GHz..... 3.8 ... 4.00 ...... 4.2 ........ 4.4 ........ 4.6 ....... 4.80
Core 1 (51) (31 - 52) (29 - 54) (29 - 56) (31 - 62) (29 - 69)
Core 2 (53) (30 - 54) (30 - 56) (30 - 60) (31 - 66) (28 - 75)
Core 3 (53) (23 - 55) (22 - 57) (22 - 60) (31 - 68) (28 - 79)
Core 4 (51) (29 - 52) (28 - 55) (29 - 57) )31 - 65) (28 - 72)

Hope that formats well

What I am thinking now is making a "gaming profile" w/ HT turned off since I can prolly drop 7 - 10C at 4.8 Ghz ..... will give it a # when kid gets back in school but wondering if anyone's tried yet on SB.
Report Comment
 

Comments have been disabled by the administrator.

Search Benchmark Reviews
QNAP Network Storage Servers

Follow Benchmark Reviews on FacebookReceive Tweets from Benchmark Reviews on Twitter