Patrol: moved Walter is a wanker 12/Memory/Usage Analysis to Development Team/Memory/Usage Analysis over redirect: revert

2010-02-24T01:33:13Z

moved Walter is a wanker 12/Memory/Usage Analysis to Development Team/Memory/Usage Analysis over redirect: revert

← Older revision	Revision as of 21:33, 23 February 2010
(No difference)

Page move vandal: moved Development Team/Memory/Usage Analysis to Walter is a wanker 12/Memory/Usage Analysis: Walter is a wanker

2010-02-24T00:21:33Z

moved Development Team/Memory/Usage Analysis to Walter is a wanker 12/Memory/Usage Analysis: Walter is a wanker

← Older revision	Revision as of 20:21, 23 February 2010
(No difference)

Patrol: moved Anal bleaching 15/Memory/Usage Analysis to Development Team/Memory/Usage Analysis over redirect: revert

2010-02-21T22:42:05Z

moved Anal bleaching 15/Memory/Usage Analysis to Development Team/Memory/Usage Analysis over redirect: revert

← Older revision	Revision as of 18:42, 21 February 2010
(No difference)

Anal bleaching: moved Development Team/Memory/Usage Analysis to Anal bleaching 15/Memory/Usage Analysis: Anal bleaching

2010-02-21T22:22:37Z

moved Development Team/Memory/Usage Analysis to Anal bleaching 15/Memory/Usage Analysis: Anal bleaching

← Older revision	Revision as of 18:22, 21 February 2010
(No difference)

RafaelOrtiz: Created page with 'Guppy-pe can give us detailed statistics about which python objects are taking space in the heap, the relationships between them and some more useful data. Particularly useful is…'

2009-10-10T04:45:19Z

Created page with 'Guppy-pe can give us detailed statistics about which python objects are taking space in the heap, the relationships between them and some more useful data. Particularly useful is…'

New page

Guppy-pe can give us detailed statistics about which python objects are taking space in the heap, the relationships between them and some more useful data. Particularly useful is the ability to take a snapshot of the current state and later print the diff between both points.

See http://guppy-pe.sourceforge.net/heapy_Use.html#heapykinds.Use.monitor.

The following patch to sugar-shell will periodically print some interesting information about memory usage. It may not apply cleanly, but anybody interested in using it will need anyway to read and understand this code.

<pre>
diff --git a/bin/sugar-shell b/bin/sugar-shell
index b0f66b4..29f130a 100755
--- a/bin/sugar-shell
+++ b/bin/sugar-shell
@@ -85,7 +85,117 @@ def _shell_started_cb():
hw_manager = hardwaremanager.get_manager()
hw_manager.set_dcon_freeze(0)

+global mem_used, objects, garbage
+mem_used = 0
+objects = 0
+garbage = 0
+
+def _log_mem_usage():
+ import analysis, os, gc, logging
+ for i in xrange(3):
+ total = len(gc.get_objects())
+ unreachable = gc.collect()
+ logging.debug('After collection - total before: %r total after: %r unreachable: %r' % (total, len(gc.get_objects()), unreachable))
+
+ global mem_used, objects, garbage
+
+ current_mem = analysis.Analysis(os.getpid()).ApproxRealMemoryUsage()
+ current_objects = len(gc.get_objects())
+ current_garbage = len(gc.garbage)
+ logging.debug('mem: %i increment: %i' % (current_mem, current_mem - mem_used))
+ logging.debug('total objects: %i increment: %i' % (current_objects, current_objects - objects))
+ logging.debug('garbage: %i increment: %i\n' % (current_garbage, current_garbage - garbage))
+
+ mem_used = current_mem
+ objects = current_objects
+ garbage = current_garbage
+
+ return True
+
+global prev_heap
+prev_heap = None
+
+def _log_objects():
+ import gc
+ gc.collect()
+
+ global prev_heap
+ import guppy, logging
+ import view.BuddyIcon
+ h = guppy.hpy()
+
+ heap = h.heap()
+
+ if prev_heap is not None:
+ diff = heap.diff(prev_heap)
+ for i in xrange(6):
+ logging.debug(diff)
+ diff = diff.more
+
+ prev_heap = heap
+
+ return True
+
+global obj_occurrences
+obj_occurrences = {}
+
+def _get_obj_dict():
+ import gc
+
+ obj_dict = {}
+ for o in gc.get_objects():
+ t = str(type(o))
+ if t in obj_dict:
+ obj_dict[t] += 1
+ else:
+ obj_dict[t] = 1
+
+ return obj_dict
+
+def _calc_diff(dict_a, dict_b):
+ types_a = set(dict_a.keys())
+ types_b = set(dict_b.keys())
+ types = types_a | types_b
+ result = {}
+
+ for t in types:
+ if t in dict_a and t in dict_b:
+ result[t] = dict_a[t] - dict_b[t]
+ elif t in dict_a:
+ result[t] = dict_a[t]
+ else:
+ result[t] = - dict_b[t]
+ if result[t] == 0:
+ del result[t]
+ return result
+
+def _log_diff():
+ import logging
+ logging.debug('Heap diff:')
+
+ current_objs = _get_obj_dict()
+
+ global obj_occurrences
+ if obj_occurrences:
+ diff = _calc_diff(current_objs, obj_occurrences)
+ for t, freq in diff.iteritems():
+ logging.debug('%s\t%d' % (t, freq))
+
+ obj_occurrences = current_objs
+
+ logging.debug('-----------------------------------')
+
+ return True
+
def main():
+ import gc
+ gc.set_debug(gc.DEBUG_LEAK)
+
+ #_log_objects()
+ #gobject.timeout_add(30000, _log_objects)
+ gobject.timeout_add(5000, _log_mem_usage)
+ #gobject.timeout_add(5000, _log_diff)
+
gobject.idle_add(_shell_started_cb)

logsmanager.setup()
</pre>

Development Team/Memory/Usage Analysis - Revision history

Patrol: moved Walter is a wanker 12/Memory/Usage Analysis to Development Team/Memory/Usage Analysis over redirect: revert

Page move vandal: moved Development Team/Memory/Usage Analysis to Walter is a wanker 12/Memory/Usage Analysis: Walter is a wanker

Patrol: moved Anal bleaching 15/Memory/Usage Analysis to Development Team/Memory/Usage Analysis over redirect: revert

Anal bleaching: moved Development Team/Memory/Usage Analysis to Anal bleaching 15/Memory/Usage Analysis: Anal bleaching

RafaelOrtiz: Created page with 'Guppy-pe can give us detailed statistics about which python objects are taking space in the heap, the relationships between them and some more useful data. Particularly useful is…'