Some 8 years ago, Anandtech produced a comprehensive comparison review of just about every cooler on the market. Such things can be updated; even different test-beds can be "normalized" to make comparisons to old platforms, but the main factor in a nutshell is the processor load wattage.
A review of some nine items doesn't provide thorough enough information to assure that the best-performing coolers are even in the sample mix. Other times, I'd seen enough to raise suspicion of a sort of "payola-in-kind," when a major computer magazine compared some seven or eight coolers, picked a "Kickass 9" winner, and then presented a two-page color advertisement for the winning cooler on the pages just following the review. "You buy my advertising; I'll scratch your back." Coolers that had been part of a discussion frenzy among enthusiasts touting their phenomenal performance weren't included.
So it pays to consult several comparison reviews. The results are rank-ordered by cooler; the test-bed specs are published; apply deductive logic to how three coolers are treated in two reviews when each review only covers two coolers. You would then be able to build your own comprehensive list of statistics: Ambient, Idle and Load temperatures; bundled fan dBA ratings; change in C between either of the first two and Load temperatures. Some tests report Ambient that can vary over a range of 5 or 10C. You can normalize for Ambient to rank-order the performance of different coolers tested in different ambient conditions.
The test-beds would need to match in thermal wattage. Some reviews test at stock settings for idle and load comparisons; they all usually report the performance statistics for overclock settings. Most of them report thermal wattage. You could even extrapolate for comparison, because thermal wattage affects magnitudes of improvement. For instance, a TIM improves load temperature by X-degrees C on one test-bed @ 100W. You might expect an improvement of only X - n degrees for a wattage of 130.
And if you have tried a handful of coolers over the years, you can throw your own observations into the mix. Maybe you didn't save log-files for spreadsheet analysis -- an easy thing to do if you chose to do it. But you'd have a memory of things like a lead or highest-core sensor reading, perhaps a vague average, the range of values across the cores at a point in time and so on.
Facebook
Twitter