From 398ac1affc5d64ff842987206a6ed2c7b115d6c6 Mon Sep 17 00:00:00 2001 From: Julien Muchembled <jm@nexedi.com> Date: Tue, 9 Feb 2010 18:52:51 +0000 Subject: [PATCH] testPerformance: update thresholds git-svn-id: https://svn.erp5.org/repos/public/erp5/trunk@32379 20353a03-c40f-0410-a6d1-a30d3c3de9de --- product/ERP5Type/tests/testPerformance.py | 46 ++++++++++++++++------- 1 file changed, 33 insertions(+), 13 deletions(-) diff --git a/product/ERP5Type/tests/testPerformance.py b/product/ERP5Type/tests/testPerformance.py index 2135e8cc02..522e05cb8b 100644 --- a/product/ERP5Type/tests/testPerformance.py +++ b/product/ERP5Type/tests/testPerformance.py @@ -29,6 +29,7 @@ import unittest from time import time import gc +import subprocess import transaction from DateTime import DateTime @@ -44,22 +45,37 @@ import os, hotshot # specific testing environment. We must always try to stay below max # historical values. # Historical values -MIN_OBJECT_VIEW=0.142 # 0.112 -MAX_OBJECT_VIEW=0.144 # 0.120 -MIN_OBJECT_MANY_LINES_VIEW=0.288 # 0.274 -MAX_OBJECT_MANY_LINES_VIEW=0.292 # 0.294 -MIN_OBJECT_PROXYFIELD_VIEW=0.225 # 0.199 -MAX_OBJECT_PROXYFIELD_VIEW=0.228 # 0.220 +MIN_OBJECT_VIEW=0.144 # 0.112 +MAX_OBJECT_VIEW=0.147 # 0.120 +MIN_OBJECT_MANY_LINES_VIEW=0.281 # 0.274 +MAX_OBJECT_MANY_LINES_VIEW=0.286 # 0.294 +MIN_OBJECT_PROXYFIELD_VIEW=0.213 # 0.199 +MAX_OBJECT_PROXYFIELD_VIEW=0.217 # 0.220 #CURRENT_MIN_OBJECT_VIEW=0.1220 #CURRENT_MAX_OBJECT_VIEW=0.1280 -MIN_MODULE_VIEW=0.147 # 0.125 -MAX_MODULE_VIEW=0.150 # 0.175 -MIN_TIC=0.0329 # 0.260 -MAX_TIC=0.0350 # 0.343 +MIN_MODULE_VIEW=0.149 # 0.125 +MAX_MODULE_VIEW=0.151 # 0.175 +MIN_TIC=0.0323 # 0.260 +MAX_TIC=0.0344 # 0.343 MIN_OBJECT_CREATION=0.0068 # 0.0070 MAX_OBJECT_CREATION=0.0073 # 0.0082 LISTBOX_COEF=0.00169 # 0.02472 # Change history +# 2010-02-09 +# the bot is slightly slower since 2009-11-29 +# MIN_OBJECT_VIEW : 0.142 -> 0.144 +# MAX_OBJECT_VIEW : 0.144 -> 0.147 +# MIN_MODULE_VIEW : 0.147 -> 0.149 +# MAX_MODULE_VIEW : 0.150 -> 0.151 +# too fast by the result of optimisation +# MIN_TIC : 0.0329 -> 0.0323 +# MAX_TIC : 0.0350 -> 0.0344 +# MIN_OBJECT_MANY_LINES_VIEW : 0.288 -> 0.281 +# MAX_OBJECT_MANY_LINES_VIEW : 0.292 -> 0.286 +# MIN_OBJECT_PROXYFIELD_VIEW : 0.225 -> 0.213 +# MAX_OBJECT_PROXYFIELD_VIEW : 0.228 -> 0.217 +# XXX test_02_viewFooObjectWithManyLines became slower with [31650] +# due to a new field in Foo_view. # 2009-11-16 # MIN_OBJECT_CREATION : 0.0071 -> 0.0068 # MAX_OBJECT_CREATION : 0.0077 -> 0.0073 @@ -69,7 +85,7 @@ LISTBOX_COEF=0.00169 # 0.02472 # MIN_OBJECT_MANY_LINES_VIEW : 0.289 -> 0.288 # MIN_OBJECT_MANY_LINES_VIEW : 0.293 -> 0.292 # 2009-11-12 -# temporary increase threashold for view to notice future regressions +# temporary increase threshold for view to notice future regressions # MIN_OBJECT_VIEW : 0.132 -> 0.142 # MAX_OBJECT_VIEW : 0.138 -> 0.144 # too fast by the result of optimisation @@ -119,8 +135,12 @@ class TestPerformance(ERP5TypeTestCase, LogInterceptor): """ Executed before each test_*. """ - # Prevent GC from happening. - # We don't want cpu time to be spent outside of python code if possible. + # We don't want cpu time to be spent by random external sources: + # - Bot should have its SQL database in a tmpfs storage. + # - As bot delete all '*.pyc' files before updating the working copy, + # all '*.pyc' files have just been recreated. They should be synced: + subprocess.call('sync') + # - Prevent GC from happening. # It would increase the "crosstalk" between using more ram and using more cpu. # Another problem is that it makes result even less reproductible on another # machine where memory use does not evolve identicaly (ie. x86_64 arch, -- 2.30.9