|by AkG | March 1, 2009|
To ensure that the results from one review to another are not only reproducible but actually pertinent to this review, the Testing Methodology will be the same throughout all reviews used. If something does change we will be sure to make a special note of it and explain why this change was done and more importantly why it had to be changed or altered.
Any all CPU Cooling Solutions which do not come with their own fan, a Noctua NF-P12-1300 will be used if it accepts 120mm fans, if it only accepts 92mm a Noctua NF-B9-1600 will be used.
Except where noted all comparison testing was done on an open bench with an ambient temperature of 20c. Recorded temps were as reported via CoreTemp's "Temp Log". Average load temps were taken after 15 minutes of running Prime95 v25.4 “small fft” and are taken directly from CoreTemp’s temperature text file. Excel was used to average the results of all cores. Idle temps were taken 15 minutes after Load testing ceased. Motherboard temperatures were recorded using SpeedFan. All CPU throttling technology was disabled in the BIOS; as was all CPU fan speed control. More importantly, the Everflow fan’s built in fan speed control was set to full speed.
Arctic Cooling MX-2 thermal paste was used for all coolers during these tests unless otherwise noted. Application of thermal paste was in accordance with TIM manufacturer’s instructions; and while not necessary, the TIM was allowed to cure for 48 hours under moderate to high loads (with periods of low loads) prior to testing.
All tests were run a minimum of 4 times and only best results are represented.
Please Note: When viewing the results of the Q6600 and the e4600 please understand that in this instance the quad @ 1.45 volts actually is a cooler running chip than the relatively bad overclocker chip e4600 used. This is usually not the way it works but due to the variable nature of overclocking we happened to get a "good" quad and a bad "dual"; in that the quad is a good cool running chip when extra voltage is applied where as the dual heats up very quickly as extra voltage is applied. It would not surprise us if 1.4 volts is significantly shortening the life of the dual e4600 and that it will die a lot earlier than the quad q6600.
Please Note: To keep the motherboard chipsets from overheating a single 40mm Scythe Ultra Kaze was used, but was orientated in such a way as to not interfere with nor help the CPU cooler (i.e. it was basically on top of the South Bridge and pointed down). The 120mm Scythe E on the side of the open test bench was unplugged during temperature testing.
Notes about Overclocking:
For Q6600’s I consider 1.45 volts to be the most that I would seriously consider for a moderate-to-long term overclock.
For E4600’s I consider 1.4 volts to be the most that I would seriously consider for a moderate-to-long term overclock.
Yes you can go much higher but the longevity of the CPU is then called into question. Just as importantly the CPU should average out at LESS than 65c as this is also what I consider the safest, maximum long term overclocking temp. For the purposes of these tests I was willing to overlook temperatures as long as they averaged below 70c and did not peak over 75c. If 75c was displayed for more than 10seconds in CoreTemp all testing was stopped and that test run was considered a fail.
With these two general guidelines I overclocked both systems until either one (or both) of these "rules" was needed to be broken to continue.
Overclocking was accomplished by increasing FSB speed and then Vcore (only if necessary).
Before testing for idle and max temperatures Orthos was run for 1 hour to make sure that it was stable at a given overclock and voltage. If both finished with no errors SuperPi set to 32m was run twice. After the stability testing was accomplished the given system was allowed to sit idle for 30minutes before starting the official tests. IF both of the above stated guidelines were not broken then testing continued with an increased overclock. These steps were then repeated until 1 or both of the general guidelines were broken.
As they have no bearing on these tests the RAM’s voltage and timings are not recorded, the RAM was set to run at or as close to as possible PC-6400 speeds by running various CPU : memory dividers. Please do not consider this a full “how to” review on overclocking or “safe guidelines” for overclocking nor even an indicator on how well a given CPU will overclock. IF you are interested in OC’ing your system, and use these guidelines we at HWC take no responsibility for the results. Bad Things can happen if you are not careful.
Complete Test System:
AMD: Phenom 9750
Intel: Gigabyte p35 DS4 &
AMD: DFI LANPARTY UT 790FX-M2R
Intel: 4GB G.Skill PC6400
AMD: 4GB Mushkin PC6400
Graphics card: Asus 8800GT TOP
Hard Drive: 1x Western Digital Se16 320GB (single platter)
Power Supply: Seasonic S12 600W
|Latest Reviews in Cooling|