Rachelint commented on code in PR #6155:
URL: https://github.com/apache/arrow-rs/pull/6155#discussion_r1711707696
##########
arrow-buffer/src/buffer/null.rs:
##########
@@ -131,9 +176,20 @@ impl NullBuffer {
}
/// Returns the null count for this [`NullBuffer`]
- #[inline]
pub fn null_count(&self) -> usize {
- self.null_count
+ match &self.null_count {
+ NullCount::Eager(v) => *v,
+ NullCount::Lazy(v) => {
+ let cached_null_count = v.load(Ordering::Acquire);
+ if cached_null_count != UNINITIALIZED_NULL_COUNT {
+ return cached_null_count as usize;
+ }
+
+ let computed_null_count = self.buffer.len() -
self.buffer.count_set_bits();
+ v.store(computed_null_count as i64, Ordering::Release);
Review Comment:
> I wonder if using `Ordering::Relaxed`
https://doc.rust-lang.org/std/sync/atomic/enum.Ordering.html#variant.Relaxed
would make this PR potentially faster
>
> Specifically, it might make the generated code simpler
>
> Also, was there a reason to remove `#[inline]` - maybe that could account
for the slow down 🤔
I am not sure should we use `Relaxed` or `Aquire + Release`, I switch to
`Aquire + Release` because I am worried about the situation that:
`ArrayRef` is hold by multiple threads, and they call the `null_count`
function concurrently, if we use `Relaxed`, the `count_set_bits` computation is
possible to be performed many times...
Another reason is that I found the strongest `SeqCst` used in `arrow cpp`
https://github.com/apache/arrow/blob/187197c369058f7d1377c1b161c469a9e4542caf/cpp/src/arrow/array/data.cc#L206-L218
--
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]
