New benchmarking script

Useful for examining how a code change affects performance

chunk.py

@@ -924,7 +924,8 @@ class ChunkRenderer(object):
         # check to see if there are any signs in the persistentData list that are from this chunk.
         # if so, remove them from the persistentData list (since they're have been added to the world.POI
         # list above.
-        self.queue.put(['removePOI', (self.chunkX, self.chunkY)])
+        if self.queue:
+            self.queue.put(['removePOI', (self.chunkX, self.chunkY)])
 
         return img
 

contrib/benchmark.py

@@ -0,0 +1,51 @@
+import chunk
+import world
+import tempfile
+import glob
+import time
+import cProfile
+import os
+import sys
+import shutil
+
+# Simple Benchmarking script.  Usage and example:
+
+#  $ python contrib/benchmark.py World4/
+#  Rendering 50 chunks...
+#  Took 20.290062 seconds or 0.405801 seconds per chunk, or 2.464261 chunks per second
+
+
+# create a new, empty, cache dir
+cachedir = tempfile.mkdtemp(prefix="benchmark_cache", dir=".")
+if os.path.exists("benchmark.prof"): os.unlink("benchmark.prof")
+
+w = world.WorldRenderer("World4", cachedir)
+
+numchunks = 50
+chunklist = w._find_chunkfiles()[:numchunks]
+
+print "Rendering %d chunks..." % (numchunks)
+def go():
+    for f in chunklist:
+        chunk.render_and_save(f[2], w.cachedir, w, (None,None), None)
+start = time.time()
+if "-profile" in sys.argv:
+    cProfile.run("go()", 'benchmark.prof')
+else:
+    go()
+stop = time.time()
+
+delta = stop - start
+
+print "Took %f seconds or %f seconds per chunk, or %f chunks per second" % (delta, delta/numchunks, numchunks/delta)
+
+if "-profile" in sys.argv:
+    print "Profile is below:\n----\n"
+
+    import pstats
+    p = pstats.Stats('benchmark.prof')
+
+    p.strip_dirs().sort_stats("cumulative").print_stats(20)
+
+
+shutil.rmtree(cachedir)