4 ms·
> But it doesn't actually correspond to any kind of real-world performance. But it may correlate with performance, and that might be sufficient -- in the same
by beambot 7y ago
> But it doesn't actually correspond to any kind of real-world performance.
But it may correlate with performance, and that might be sufficient -- in the same way that Moore's law did for so long too
- klyrs 7y ago> But it may correlate with performance... Can't really argue with that sort of hedge. I'd even go so far as to say it does correlate with performance on a completely artificial benchmark. But that's a far cry from "it faithfully describes computational power" that would justify the headline in question.