Dandandan commented on code in PR #9542:
URL: https://github.com/apache/arrow-rs/pull/9542#discussion_r2929185034
##########
arrow-select/src/interleave.rs:
##########
@@ -154,13 +154,51 @@ fn interleave_primitive<T: ArrowPrimitiveType>(
data_type: &DataType,
) -> Result<ArrayRef, ArrowError> {
let interleaved = Interleave::<'_, PrimitiveArray<T>>::new(values,
indices);
+ let arrays = &interleaved.arrays;
+ let len = indices.len();
+
+ let mut output = Vec::with_capacity(len);
+ let dst: *mut T::Native = output.as_mut_ptr();
+ let mut base = 0;
+
+ // Process 8 elements at a time to issue multiple independent loads
+ // and increase memory-level parallelism for random access patterns.
+ let chunks = indices.chunks_exact(8);
+ let remainder = chunks.remainder();
+ for chunk in chunks {
+ let v0 = arrays[chunk[0].0].value(chunk[0].1);
Review Comment:
Ah like that - yeah that should be possible (would require some hardware
setup) although I think it would not be really to a user setting compile time
constants.
On the bencmarks it looks to be beneficial for both aarch64 and x86 CPUs -
getting ~23% on M1 (my laptop) and -30% on intel (the @alamb-ghbot runner).
##########
arrow-select/src/interleave.rs:
##########
@@ -154,13 +154,51 @@ fn interleave_primitive<T: ArrowPrimitiveType>(
data_type: &DataType,
) -> Result<ArrayRef, ArrowError> {
let interleaved = Interleave::<'_, PrimitiveArray<T>>::new(values,
indices);
+ let arrays = &interleaved.arrays;
+ let len = indices.len();
+
+ let mut output = Vec::with_capacity(len);
+ let dst: *mut T::Native = output.as_mut_ptr();
+ let mut base = 0;
+
+ // Process 8 elements at a time to issue multiple independent loads
+ // and increase memory-level parallelism for random access patterns.
+ let chunks = indices.chunks_exact(8);
+ let remainder = chunks.remainder();
+ for chunk in chunks {
+ let v0 = arrays[chunk[0].0].value(chunk[0].1);
Review Comment:
Ah like that - yeah that should be possible (would require some hardware
setup) although I think it would not be really to a user setting compile time
constants.
On the benchmarks it looks to be beneficial for both aarch64 and x86 CPUs -
getting ~23% on M1 (my laptop) and -30% on intel (the @alamb-ghbot runner).
--
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]
