Chester (lamchester.delete@this.gmail.com) on November 23, 2020 2:26 pm wrote:
>
> I'm not sure where you're going with this. Why buy a benchmark when it's not telling you anything
> useful, and you can present more relevant data by directly testing programs people use?

Yeah, well, if "present relevant data by directly testing programs people use" was really the primary motivation for reviewers choosing benchmark X or Y, then it would he highly unlikely for them to converge on the exact same benchmarks (Cinebench, C-Ray, etc.) despite the extremely varied performance-sensitive workloads found in the real world.

The only reasonable explanation for their choice of benchmarks is not that they try to test useful, relevant real-world stuff. Rather, as Doug S pointed out, they test what is easiest to test for them and to show interesting-looking numbers to their readers (who are probably not very demanding in the first place, except perhaps in the gaming area).

Which is why, when a new CPU gets released, you get Cinebench performance numbers from who knows how many review sites. Useful, isn't it?

(ok, things got better; it used to be SuperPi and Sisoft Sandra)