cjl :
CompuTronix: SmallFFT may be the "standard", but it is not representative of a maximum load case on an i7. LargeFFT is a better choice, since it also stresses the onboard memory controller, testing it for stability and resulting in an overall higher heat loading on an i7 based CPU.
cjl,
You are absolutely correct, but the difference is almost negligible. As you might expect, I've extensively tested the thermal signatures of Prime95 Large FFT's. Small FFT's and Blend on i7's, many months ago. Large FFT's provides an
overall higher heat load distibuted among the processor and memory components, which is also evidenced when observing Watts consumed at the power cord. Large FFT's maximum heat signature exceeds Small FFT's by just 1c on Core i variants, but does not exceed Small FFT's on Core 2 variants, where it's 2 to 3c lower during certain test segments.
Some latter-day utilities such as OCCT and Burn Test (reminiscent of TAT) use LinPack, which shows thermal signatures that resemble a bad day at the stock market, and cycle between light workloads, through test segments which spray all processor registers with all one's, (100%
thermal load, which equates to 115%
workload), and in turn can drive a high-end-air cooled, overclocked i7 at Vcore Max 1.375 with HT on, right through the ceiling to ring the Tjunction Max bell like a fire alarm!
😱
Since there are very few applications or games that will spike, let alone sustain processor workloads beyond 70% to 85%, utilities which load all registers with all one's are not representative of real-world computing. While these utilities are certainly very useful for stability testing, the author of Real Temp, Kevin Glynn, uses Prime95 Small FFT's unilaterally for thermal testing, and as the difference between Large and Small FFT's is only 1c on Core i's, it might not be worth throwing another variable at the n00bs.
Your point, however, is well taken, and is a very astute observation.
Comp
