They said it's not faster than the desktop parts, which is almost a false statement since it actually is faster at multithreaded than both base desktop SKUs. And then they switch to providing a specific percentage when comparing the the 4080 (also based on an deceptively narrow comparison, one their other articles counter) for dramatic effect.

I don't think it was ever meant to be the most informative article: it seems written to serve the contrarians because that's profitable from a readership perspective for a publication like this.