Wednesday 02 November 2005 2:14:43 pm - 6 replies
Hi there,
We seem to be running into some speed issues with 3.6.x builds, and I suspect we're overlooking something pretty basic.
We have template compile and caching all enabled, but the results I'm getting back are:
Time accumulators: Accumulator Elapsed Percent Count Average ini_load Load cache 0.2459 sec 2.6363% 11 0.0224 sec Mysql Total Mysql_queries 0.0289 sec 0.3104% 16 0.0018 sec Looping result 0.0013 sec 0.0142% 10 0.0001 sec Template Total 0.5434 sec 5.8% 3 0.1811 sec Template load 0.1106 sec 1.1856% 3 0.0369 sec Template processing 0.4317 sec 4.6295% 3 0.1439 sec override Cache load 0.0777 sec 0.8327% 3 0.0259 sec Total script time: 9.3256 sec
Pretty horrible ![]()
In the admin we're seeing >1.5 secs on the template processing.
I've check ini overrides in case there's a setting turning off the caching, but cant find it.
I've been looking at threads like http://ez.no/community/forum/inst...configuration/ez_publish_performance and we really should be getting better results.
The machine is a brand spanking new dual P4 (about 3.5 Mhz), loads of RAM. Its running on a windows 2003 server, PHP has something like 128mb memory alloc - and this is the only site running on the box!
Interestingly, we're seeing the same speed problems on development sites we have on our production box (linus based), so I suspect we're overlooking something.
If anyone has some pointers to debug this I'd be very grateful.
Regards, Steve.
Wednesday 02 November 2005 2:18:18 pm
Hi,
Realised that time info wasnt that bad - here's page loads after clearing cache, site has reloaded, and you start visiting other pages:
Time accumulators: Accumulator Elapsed Percent Count Average ini_load Load cache 0.1722 sec 0.6652% 13 0.0132 sec Mysql Total Mysql_queries 3.2601 sec 12.5904% 343 0.0095 sec Looping result 0.1270 sec 0.4906% 253 0.0005 sec Template Total 20.1504 sec 77.8% 3 6.7168 sec Template load 8.4687 sec 32.7060% 3 2.8229 sec Template parser: create text elements 0.1363 sec 0.5264% 35 0.0039 sec Template parser: remove whitespace 0.0250 sec 0.0965% 35 0.0007 sec Template parser: construct tree 0.8069 sec 3.1163% 35 0.0231 sec Template load and register function 0.0525 sec 0.2027% 6 0.0087 sec Template processing 11.6806 sec 45.1104% 3 3.8935 sec override Cache load 0.0843 sec 0.3255% 4 0.0211 sec Sytem overhead Fetch class attribute name 0.0000 sec 0.0000% 0 0.0000 sec class_abstraction Instantiating content class attribute 0.0005 sec 0.0020% 6 0.0001 sec General String conversion 0.0081 sec 0.0312% 3 0.0027 sec String conversion w/ codepage reverse 0.0075 sec 0.0289% 3 0.0025 sec Total script time: 25.8933 sec
Wednesday 02 November 2005 2:21:50 pm
This is navigating back to the same page, no cache clearing:
Time accumulators: Accumulator Elapsed Percent Count Average ini_load Load cache 0.1518 sec 4.3682% 12 0.0126 sec Mysql Total Mysql_queries 0.0534 sec 1.5363% 24 0.0022 sec Looping result 0.0025 sec 0.0712% 17 0.0001 sec Template Total 1.2465 sec 35.9% 3 0.4155 sec Template load 0.1060 sec 3.0497% 3 0.0353 sec Template processing 1.1394 sec 32.7880% 3 0.3798 sec override Cache load 0.0793 sec 2.2830% 4 0.0198 sec Sytem overhead Fetch class attribute name 0.0000 sec 0.0000% 0 0.0000 sec class_abstraction Instantiating content class attribute 0.0005 sec 0.0136% 6 0.0001 sec General String conversion 0.0095 sec 0.2735% 3 0.0032 sec String conversion w/ codepage reverse 0.0089 sec 0.2550% 3 0.0030 sec Total script time: 3.4749 sec
Gets back to the more social, but unacceptable 3.4 secs load time!
Steve
Thursday 03 November 2005 6:41:25 pm
EzCrew,
Any chance of some input here. We've got cache enabled sites on machines that have plenty of muscle performing terribly.
It looks like the machine always has to chug away when prcessing templates which suggests its not compiling/caching etc, but we dont see how to debug at this point? Its only a problem we seem to have on 3.6.x builds.
Could you give some pointers please?
Thanks, Steve.
Monday 07 November 2005 4:31:28 pm
Steve- I would also implicate a missing PHP accelerator. I regularly see a 5-10 fold improvement in page load times when using an accelerator. We are using eaccelerator on windows as well, and are getting page loads on pages with complex templates as quick as .25 seconds on a 1.8ghz dual proc. With the accelerator, it was 2.5-3 seconds.
You must be logged in to post messages in this topic!