From: Raymond Hettinger <python@rcn.com>
Date: Wed, 13 May 2015 04:40:50 +0000 (-0700)
Subject: More timings suggest that 2500 is closer to the break-even point.
X-Git-Tag: v3.5.0b1~176
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=636488043b490ab70413990b458443aa41f504b2;p=thirdparty%2FPython%2Fcpython.git

More timings suggest that 2500 is closer to the break-even point.
---

diff --git a/Modules/_heapqmodule.c b/Modules/_heapqmodule.c
index 01b35be8eba7..88c35cf32eec 100644
--- a/Modules/_heapqmodule.c
+++ b/Modules/_heapqmodule.c
@@ -335,7 +335,7 @@ heapify_internal(PyObject *heap, int siftup_func(PyListObject *, Py_ssize_t))
        in cache, we prefer the simpler algorithm with less branching.
     */
     n = PyList_GET_SIZE(heap);
-    if (n > 10000)
+    if (n > 2500)
         return cache_friendly_heapify(heap, siftup_func);
 
     /* Transform bottom-up.  The largest index there's any point to