In my experience, every computer is faster with Linux than with Windows. But if this measures just the processor performance on similar tasks I guess it’s news.
I think it comes down to the culture. A minuscule improvement to a file system is big news in the Linux community. There’s also lots of academic interest in the performance critical parts of the kernel that you just can’t emulate with a closed source model. Is anyone writing papers on how to obtain a 2% improvement in the task scheduler on Windows?
Linux dominates the server market, so even small improvements matter when you’re talking about a server farm with thousands of machines or the latest supercomputer. Many, many people care about the scalability of Linux. On Windows, we say: NTFS? It’s good enough. The user won’t notice on modern SSDs.
Original Phoronix article which has all the individual benchmarks—weird that they didn’t link to it