jhorstmann opened a new issue #1182:
URL: https://github.com/apache/arrow-rs/issues/1182


   **Is your feature request related to a problem or challenge? Please describe 
what you are trying to do.**
   
   For simple arithmetic kernels (+,-,*), the compiler should be able to 
automatically vectorize the scalar code and even generate better code than our 
custom simd implementations. Our simd kernels currently process a specific 
number of lanes at the same time, dependent on the element type, an 
autovectorized implementation can possibly get unrolled multiple times so it 
only has to check the loop condition every n lanes.
   
   The checked division kernels probably still benefit from the custom simd 
implementation and should be kept.
   
   **Describe the solution you'd like**
   
   - add some more benchmarks in `arithmetic_kernels`
   - make those benchmarks process a larger amount of data, currently the 
arrays are of length 512 and the overhead of allocation or validity bitmap 
calculation might dominate the actual arithmetic computation
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to