Optimize comparison kernels for inlined views #7621
Description
Is your feature request related to a problem or challenge? Please describe what you are trying to do.
There are places where arrays with only inlined views (<= 12 bytes) are not handled separately from view arrays with data buffers. One of them is the comparison kernel.
Describe the solution you'd like
Check upfront if the view holds no data buffer, use the fast path (only compare the u128 views), which should be much faster than checking the length individually.
Describe alternatives you've considered
Additional context
There are probable much more places that can benefit from this optimization.