Surely this isn't the first time you see software that claims high performance and speed without any hard benchmarks attached.
You can benchmark an algorithm or a small piece of code, but for something like a game library, the definition of what is "fast" and what isn't is too dependent on context to be meaningful.
That doesn't mean that the statement is useless: it tells me that this is, at least partially, one of the goals of the library.
For most folks I think they use an fps score while not sacrificing aesthetic quality or game play experience.
i.e. The engine is considered lower quality if it is laggy, generates DLSS chowder, or glitches up on some hardware.
The people responsible for shader cache performance get a lot of grief given the performance hit in unpredictable play contexts is often very noticeable.
You can benchmark an algorithm or a small piece of code, but for something like a game library, the definition of what is "fast" and what isn't is too dependent on context to be meaningful.
That doesn't mean that the statement is useless: it tells me that this is, at least partially, one of the goals of the library.