diff --git a/baselibs.conf b/baselibs.conf
deleted file mode 100644
index e4a7a4d..0000000
--- a/baselibs.conf
+++ /dev/null
@@ -1 +0,0 @@
-libxxhash0
diff --git a/xxhash.changes b/xxhash.changes
index d8ee869..6ad13f8 100644
--- a/xxhash.changes
+++ b/xxhash.changes
@@ -1,9 +1,3 @@
--------------------------------------------------------------------
-Wed Aug 14 15:06:38 UTC 2024 - Jan Engelhardt <jengelh@inai.de>
-
-- Build hwcap_libs (x86-64-v3) for xxhash, apparently there is
-  something to gain.
-
 -------------------------------------------------------------------
 Sun May  5 19:35:03 UTC 2024 - Bruno Pitrus <brunopitrus@hotmail.com>
 
diff --git a/xxhash.spec b/xxhash.spec
index 1b9ece8..09142bf 100644
--- a/xxhash.spec
+++ b/xxhash.spec
@@ -24,7 +24,6 @@ License:        BSD-2-Clause AND GPL-2.0-only
 Group:          Productivity/Security
 URL:            https://github.com/Cyan4973/xxHash
 Source0:        https://github.com/Cyan4973/xxHash/archive/v%{version}.tar.gz#/xxHash-%{version}.tar.gz
-Source9:        baselibs.conf
 Patch1:         test-tools-do-not-override-cflags.patch
 BuildRequires:  gcc-c++
 BuildRequires:  pkg-config
@@ -63,6 +62,10 @@ Headers and other development files for xxHash.
 export CFLAGS="%{optflags} -DXXH_X86DISPATCH_ALLOW_AVX=1"
 export CXXFLAGS="$CFLAGS"
 export LDFLAGS="%{?build_ldflags}"
+# DISPATCH=1 if you want AVX2/AVX512. But it does not seem to perform any
+# better than the lowest-denomimation code on at least the 1135G7 and 5950X
+# CPUs, and for both LP64 as well as ILP32 — it seems to be all within margin
+# of error.
 %make_build prefix=%{_prefix} libdir=%{_libdir}
 
 %install