pitrou commented on code in PR #13394:
URL: https://github.com/apache/arrow/pull/13394#discussion_r909337414
##########
cpp/src/arrow/csv/writer.cc:
##########
@@ -59,6 +59,25 @@ namespace csv {
namespace {
+// This function is to improve performance. It copies csv delimiter and eol
+// without calling `memcpy`.
+// Each csv field is followed by a delimiter or eol, which is often only one
+// or two chars. If copying both the field and delimiter with `memcpy`, cpu
+// may suffer from high branch misprediction as we are tripping `memcpy` with
+// interleaved (normal/tiny/normal/tiny/...) buffer sizes, which are handled
+// separately inside `memcpy`. This function goes fast path if the buffer
+// size is one or two chars to leave `memcpy` only for copyig csv fields.
Review Comment:
```suggestion
// This function is to improve performance. It copies CSV delimiter and eol
// without calling `memcpy`.
// Each CSV field is followed by a delimiter or eol, which is often only one
// or two chars. If copying both the field and delimiter with `memcpy`, CPU
// may suffer from high branch misprediction as we are tripping `memcpy` with
// interleaved (normal/tiny/normal/tiny/...) buffer sizes, which are handled
// separately inside `memcpy`. This function goes fast path if the buffer
// size is one or two chars to leave `memcpy` only for copying CSV fields.
```
--
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]
