etseidl commented on PR #6220:
URL: https://github.com/apache/arrow-rs/pull/6220#issuecomment-2278593775

   While performance for decoding is still abominable, oddly this changes 
improves it somewhat.
   
   On my laptop (old macbook pro w/ 2.2GHz core i7)
   ```
   % critcmp master set_from
   group                                                                        
        master                                 set_from
   -----                                                                        
        ------                                 --------
   decoding: dtype=FixedLenByteArray(16), encoding=BYTE_STREAM_SPLIT            
        1.09   564.8±15.15µs        ? ?/sec    1.00   517.2±10.01µs        ? 
?/sec
   decoding: dtype=FixedLenByteArray(2), encoding=BYTE_STREAM_SPLIT             
        1.05    384.5±6.32µs        ? ?/sec    1.00    365.4±8.88µs        ? 
?/sec
   decoding: dtype=f32, encoding=BYTE_STREAM_SPLIT                              
        1.01     38.8±0.86µs        ? ?/sec    1.00     38.6±0.85µs        ? 
?/sec
   decoding: dtype=f64, encoding=BYTE_STREAM_SPLIT                              
        1.00     82.1±1.48µs        ? ?/sec    1.01     82.6±2.22µs        ? 
?/sec
   encoding: dtype=FixedLenByteArray(16), encoding=BYTE_STREAM_SPLIT            
        1.02   282.4±15.78µs        ? ?/sec    1.00   277.1±14.91µs        ? 
?/sec
   encoding: dtype=FixedLenByteArray(2), encoding=BYTE_STREAM_SPLIT             
        1.05     50.6±4.60µs        ? ?/sec    1.00     48.2±1.91µs        ? 
?/sec
   encoding: dtype=f32, encoding=BYTE_STREAM_SPLIT                              
        1.00     44.4±1.42µs        ? ?/sec    1.01     44.9±1.69µs        ? 
?/sec
   encoding: dtype=f64, encoding=BYTE_STREAM_SPLIT                              
        1.00    108.3±2.07µs        ? ?/sec    1.00    108.6±2.72µs        ? 
?/sec
   ```
   
   On my workstation (3.6 GHz Core i7-12700K)
   ```
   group                                                                        
        master                                 set_from
   -----                                                                        
        ------                                 --------
   decoding: dtype=FixedLenByteArray(16), encoding=BYTE_STREAM_SPLIT            
        1.19    357.0±2.69µs        ? ?/sec    1.00    299.7±1.98µs        ? 
?/sec
   decoding: dtype=FixedLenByteArray(2), encoding=BYTE_STREAM_SPLIT             
        1.23    300.2±1.84µs        ? ?/sec    1.00    244.4±1.98µs        ? 
?/sec
   decoding: dtype=f32, encoding=BYTE_STREAM_SPLIT                              
        1.00     15.1±0.31µs        ? ?/sec    1.00     15.1±0.13µs        ? 
?/sec
   decoding: dtype=f64, encoding=BYTE_STREAM_SPLIT                              
        1.02     33.5±0.64µs        ? ?/sec    1.00     32.9±0.57µs        ? 
?/sec
   encoding: dtype=FixedLenByteArray(16), encoding=BYTE_STREAM_SPLIT            
        1.01    120.8±2.49µs        ? ?/sec    1.00    119.6±0.52µs        ? 
?/sec
   encoding: dtype=FixedLenByteArray(2), encoding=BYTE_STREAM_SPLIT             
        1.00     24.1±0.24µs        ? ?/sec    1.01     24.2±0.17µs        ? 
?/sec
   encoding: dtype=f32, encoding=BYTE_STREAM_SPLIT                              
        1.01     16.5±0.20µs        ? ?/sec    1.00     16.4±0.12µs        ? 
?/sec
   encoding: dtype=f64, encoding=BYTE_STREAM_SPLIT                              
        1.00     36.1±0.26µs        ? ?/sec    1.01     36.3±0.39µs        ? 
?/sec
   ```


-- 
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