THE TEST BENCH
This is our test bench for the RAID and Enterprise testing. Currently in a state of flux, this bench has undergone some transitions as of late. Our extensive water cooling, with additional radiators under the table, allows for us to take the i7-3820 as far as she will go without any concern for cooling. While this may look a tad messy to the reader, it is actually very organized by our standards!
CPU: Intel Core i7-3820K Sandy Bridge-E 3.6GHz (3.9GHz Turbo) LGA 2011 130W Four-Core Desktop Processor Overclocked to 5.0 Ghz
RAM: 16GB Crucial Ballistix 1866, 9-9-9-27
GPU: Dual EVGA GTX480
POWER: ST1500 Fully Modular 1500 Watt Power Supply 1500W (Peak 1600W) 12v1320W/110A (Peak 120A) combined+3.3 5v 280W
CHASSIS: Danger Den Torture Rack
CPU COOLER: HeatKiller 3.0
WATER SYSTEM: Two KMP-400 w/reservoirs in a Serial loop, two MCR320-QP rads, and 1 BIPS 240 rad, CPU only. Loop 2- MCP-655 and Honda Radiator on dual 480GTX GPUs with EK Blocks.
OVERCLOCKING THE i7-3820
While overclocking of the six core SNB-E chips is very simple by adjusting the multiplier up to 57X, the 3820 has a ‘limited unlocked’ multiplier that maxes out at 43X. This will effectively limit the overclocking potential to 4.3GHz.
Luckily there is the ability to adjust the blck up from 100MHz to 125MHz, 166MHz and 250MHz. This allows the user to circumvent the multiplier limitations, via bclk modifications, on the way to some impressive overclock speeds!
One of the greatest things to come along in the overclocking world lately is the preset default overclock profiles. Overclocking “Purists” do not hold preset profiles in high regard, but this overclocker certainly has no issue with them. One of the greatest benefits to these profiles is that they are true time savers, and should be treated as such. They are also allowing casual users to get solid overclocks without taking a ‘deep dive’ into overclocking 101.
These profiles, provided you have a good motherboard, will give you some good ballpark numbers to start with. This eliminates some of the discovery time that was previously needed to find the correct voltages. All automatic overclocking profiles should be used only as a guideline, as they are prone to using higher voltages than may be required. With some time and adjustment, they can be excellent tools.
We experienced absolutely no problem overclocking the processor right up to 5.0 with minimal voltages. It truly was a very simple and easy process to get stable at 5.0, so much so that it was quite routine. Finding the optimum settings for voltages did require a bit of time and fine tuning, but as a solid 5.0 overclock goes, this was about as easy as it gets.
To be quite honest, a 5.0 overclock with 1.48 Vcore is rather good. There was some modification of this and other voltages down from the auto-tuning settings. Running at a multiplier of 40 and the bus speed of 125 MHz was actually an auto setting, so it was just as easy as it gets!
We will be using HyperPi, Cinebench and PCMark Vantage for comparison purposes.
There are many more measurements than just base performance levels of one CPU compared to the other, so we will include power testing as well.
We will not be measuring the total system power draw. With varying power usage from devices such as GPUs, CPUs, RAM, and add-on cards, the results can be skewed quite a bit. The purpose of this testing is to determine the power consumption of the most important component, the CPU Socket.
All CPU power monitoring will be done with the AIDA 64 application. For users wishing to check their own power usage, you simply navigate to the computer selection, then highlight the Sensor selection. Scroll down the right hand side of the screen, and there is CPU Package, which is your CPU socket power draw.