On Tue, 15 Oct 2024, liuhongt wrote:
> +<ul>
> + <li>The default vectorizer cost model at <code>-O2</code> has been enhanced
> + to handle unknown tripcount. But it still disables vectorization of loops
> + when any runtime check for data dependence or alignment is required,
> + it also disables vectorization of epilogue loops but otherwise is equal
> + to the cheap cost model.
> + </li>
> +</ul>
Thank you! I did a bit (if late) copy editing; see below.
Pushed.
Gerald
commit 1845c708ee69271cb04af7ef7e639cebbdab6f17
Author: Gerald Pfeifer <ger...@pfeifer.com>
Date: Wed Aug 20 14:25:54 2025 +0200
gcc-15: Copy edit note on default vectorizer cost model
Avoid one "but", break up a sentence, fix spelling of "run-time" as
an adjective, and improve language.
diff --git a/htdocs/gcc-15/changes.html b/htdocs/gcc-15/changes.html
index c94fa1bd..47f2cb1d 100644
--- a/htdocs/gcc-15/changes.html
+++ b/htdocs/gcc-15/changes.html
@@ -62,9 +62,9 @@ You may also want to check out our
<ul>
<li>The default vectorizer cost model at <code>-O2</code> has been enhanced
- to handle an unknown tripcount. But it still disables vectorization of
loops
- when any runtime check for data dependence or alignment is required,
- it also disables vectorization of epilogue loops but otherwise is equal
+ to handle an unknown tripcount. It still disables vectorization of loops
+ when any run-time check for data dependence or alignment is required.
+ It also disables vectorization of epilogue loops, but is otherwise equal
to the cheap cost model.
</li>
<li>The vectorizer now supports vectorizing loops with early exits where
