cameronmc88,
You haven't mentioned, and no one has thought to ask; what is your ambient temperature? All computer temperatures rise and fall with ambient, which plays a major role in determining thermal performance.
What was you ambient temperature when you ran Prime95 Version 26.6 Small FFT's?
Guys,
For those who haven't read the Temp Guide just yet, here's a few details:
There's a difference between
stability testing and
thermal testing. For
thermal testing, the goal is to replicate Intel's TDP values as closely as possible, so as to establish a
valid thermal baseline. This is explained in Section 13.
While I agree there are several utilities superior to Prime95 for system
stability testing such as Asus RealBench, there are none more ideal for CPU
thermal testing than Prime95 Version 26.6 Small FFT's, because it's a
steady-state 100% workload which yields
steady-state Core temperatures, and is as close as you can get to replicating Intel's Thermal Design Power (TDP) values within 3%.
This means that after your CPU warms up and stabilizes about 4 to 6 minutes into the test, the Core temperatures are
very steady, which is critical to analyzing thermal performance. How can you determine thermal performance if you run a utility with fluctuating workloads which causes fluctuating Core temperatures that look like a bad day on the stock market?
Intel's TDP values are a calculation AND a measurement. In order to achieve accurate results which agree between calculated and measured values, Intel must use a
steady-state 100% workload. Further, 100%
usage doesn't necessarily equate to 100%
workload.
AIDA64 is OK for
stability testing, but it's not
steady-state, and is confusing due to having too many test combinations. AIDA64's only
steady-state test is "Stress CPU" when selected individually. Even though it shows 100%
usage, it's far below TDP and 100%
workload, which gives the user a false sense of security because of misleadingly low temperatures. Conversely, the "Stress FPU" test when selected individually, may instantly slam Core temperatures to Throttle at TjMax, way above TDP, so that doesn't work for
thermal testing either.
OCCT? Which test? CPU: Linpack? With or without AVX? Neither is
steady-state, nor are they close to TDP. CPU: OCCT? Notice the selection for "Small Data Set"? Sound familiar? Maybe like Prime95 "Small FFT's"? Core temperatures similar to AVX Version 27.7?
Steady-state but too far above TDP. And OCCT already includes HWMonitor.
Note: It's never a good idea to run more than one monitoring utility concurrently, because they may interfere with the accuracy of one another.
If you're going to talk about testing and specifications, then you need to be
very specific. I've tested every utility, including several most of you have never heard of, and I can tell you without any doubts, that Prime95 Version 26.6 is the best utility for
thermal testing. If you want to get a
valid thermal baseline, then that's the way to go. Run it with Core Temp or Real Temp, which was developed specifically for Intel processors.
CT