aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorWarner Losh <imp@FreeBSD.org>2021-03-14 00:30:37 +0000
committerWarner Losh <imp@FreeBSD.org>2021-03-14 00:30:37 +0000
commitc65a1c7ddf4ff56ec1782943fced0d5eb5902107 (patch)
tree7822315adebe83e4a7f57bcab7b50825a21bb809
parent3841b8b3e839127dfbeaf8cf66e48318bcc1f47b (diff)
-rw-r--r--website/content/en/releases/13.0R/relnotes.adoc2
1 files changed, 1 insertions, 1 deletions
diff --git a/website/content/en/releases/13.0R/relnotes.adoc b/website/content/en/releases/13.0R/relnotes.adoc
index fae57527b3..5a250195b2 100644
--- a/website/content/en/releases/13.0R/relnotes.adoc
+++ b/website/content/en/releases/13.0R/relnotes.adoc
@@ -345,7 +345,7 @@ While those operations can be emulated in the kernel on i486, they cannot be emu
Updating the default allows compiler generated code to select the right atomics in those cases, allow better optimizations and produce better error messages when necessary.
The older library and/or include file approaches are much less optimal in resulting code and diagnostics.
Current compiler technology produces better, faster, and/or smaller binaries for i686 than for i486.
-Several bugs in compiler support for i486 code generation attest to its lessor use in the wider ecosystem.
+Several bugs in compiler support for i486 code generation attest to its lesser use in the wider ecosystem.
In the wider ecosystem, i686 has been the default for many years so has received more testing and more optimization.
Finally, the 32-bit amd64 libraries have been i686 since their inception.
These factors strongly suggest that a i686 default will provide such an improved enough user experience to offset the minor pain for those few users of the older technology.