I have some information now. First, I apologize for taking so long- I 
had the misfortune of getting sick during this analysis.

So comparing builds from 1/18 and 2/2 I can see approximately a 15% 
slowdown in Frame Creation time, going from 0.086 sec. to 0.102 sec. 
Also, Style Resolution slowed from approximately 0.223 sec. to 0.248 
sec, or about 9%.

Total layout time went from .559 sec to 0.592, or about a 6% slowdown.

This does not seem to account for the total slowdown we are seeing - 
clearly there is something else going on. In fact, the general noise 
factor in the Gecko Performance Tests are about 5%, so the change I am 
seeing could be noise, although I believe that there is a slight 
performance degradation worth looking into in frame creation and 
probably style resolution.

These results are all from Viewer. It could be that the problem is 
actually in Mozilla itself, and Pierre has produced some interesting 
data on the impact of skins on performance[1].


- marc

1. news://news.mozilla.org/3A814068.A8E21741%40netscape.com

Chris Waterson wrote:

> We have still made zero progress on this:. A week has gone by, and page
> load is still down by 13%. Here are the people that are on the hook
> during the time that the slowdown occurred. If I made a quick look
> through the deltas
> 
>  
> http://bonsai.mozilla.org/showcheckins.cgi?&treeid=SeaMonkey&batchid=453
> 
> If had to guess, I'd say that darin and ftang should be reviewing their
> changes...
> 
> chris
> 
> Chris Hofmann wrote:
> 
>> ----------------------------------------------------------------------
>> The following people are on "the hook", since they have made checkins
>> to the tree since it last opened:
>> 
>>               Who                            What
>>   Peter.VanderBeken%pandora.be2 changes
>>   av%netscape.com             1 change
>>   axel%pike.org               7 changes
>>   beard%netscape.com          1 change (1 while tree closed!)
>>   ben%netscape.com            1 change
>>   bienvenu%netscape.com       2 changes
>>   blakeross%telocity.com      10 changes
>>   brade%netscape.com          1 change
>>   ccarlen%netscape.com        14 changes
>>   cls%seawood.org             1 change
>>   darin%netscape.com          34 changes (2 while tree closed!)
>>   dbaron%fas.harvard.edu      5 changes (4 while tree closed!)
>>   dmose%mozilla.org           1 change
>>   dr%netscape.com             1 change
>>   jj%netscape.com             1 change (1 while tree closed!)
>>   kin%netscape.com            19 changes
>>   mcafee%netscape.com         8 changes (8 while tree closed!)
>>   mkaply%us.ibm.com           4 changes (2 while tree closed!)
>>   mscott%netscape.com         2 changes
>>   neeti%netscape.com          1 change
>>   nhotta%netscape.com         1 change (1 while tree closed!)
>>   racham%netscape.com         15 changes
>>   radha%netscape.com          1 change
>>   sspitzer%netscape.com       1 change
>>   timeless%mac.com            2 changes
>>   yokoyama%netscape.com       1 change
>> 137 checkins.
> 

Performance History and Trending Table


Wednesday 2/7/2001 17:40:08

Pull-ID Build-ID Parsing Content Creation Frame Creation Style Resolution Reflow Total Layout Total Time
Sec%
Sec%
Sec%
Sec%
Sec%
Sec%
Sec
Jan18-a
0.11419.294
 
98.37 %
0.10619.353
 
95.60 %
0.088 16.223
 
85.04 %
0.22240.174
 
89.11 %
0.0314.956
 
89.68 %
0.56048.221
 
93.87 %
1.160
 
95.78 %
Jan18-b
0.10418.377
 
89.51 %
0.11119.854
 
100.00 %
0.084 15.204
 
81.17 %
0.22440.784
 
90.08 %
0.0345.781
 
100.00 %
0.55748.071
 
93.29 %
1.162
 
95.91 %
Feb02-a
0.11618.733
 
100.00 %
0.10418.249
 
94.38 %
0.099 17.485
 
95.25 %
0.24941.079
 
100.00 %
0.0294.454
 
84.92 %
0.59749.414
 
100.00 %
1.211
 
100.00 %
Feb02-b
0.10918.295
 
93.94 %
0.10818.702
 
97.31 %
0.104 18.017
 
100.00 %
0.23640.202
 
94.78 %
0.0314.784
 
89.68 %
0.58649.690
 
98.28 %
1.188
 
98.03 %

Reply via email to