Ah yes, I think that Mike misunderstand my first post (that was probably not very clear): the slight performance improvement we have with this last version is versus the fastest freemarker jar (2.3.10), the one that your tests shown to be 4 times faster.
Jacopo On Apr 3, 2012, at 8:00 AM, Hans Bakker wrote: > Mike had some concerns....and seemed to be resolved now..... > > On 04/03/2012 12:52 PM, Jacopo Cappellato wrote: >> Hi Hans, >> >> yes, please wait until I fix the issue with the include; as regards >> performance, what is the problem exactly? >> >> Thanks, >> >> Jacopo >> >> On Apr 3, 2012, at 5:57 AM, Hans Bakker wrote: >> >>> Hi Jacopo, >>> >>> it is just a single commit: r986577 >>> >>> now the footer has a problem in the trunk, and the speed problems seems >>> still a problem, >>> lets first wait until that is fixed? >>> >>> >>> Regards, >>> Hans >>> >>> On 04/02/2012 09:16 PM, Jacopo Cappellato wrote: >>>> Thank you Hans, >>>> >>>> I had to actually spend a lot of time to fix it. >>>> But at the end it was not an issue in Freemarker but instead the problem >>>> was that we were not disabling debug messages in freemarker classes. >>>> I have fixed this by adding proper Log4J configuration and then I have >>>> upgraded to the latest stable release of freemarker (2.3.19). >>>> The performance looks good; I have executed (using JMeter) the main >>>> ecommerce screen with 50 threads hitting the page 100 times each: >>>> >>>> with 2.3.10 >>>> >>>> Generate Summary Results = 5000 in 195.2s = 25.6/s Avg: 1925 Min: >>>> 74 Max: 9307 Err: 0 (0.00%) >>>> >>>> with 2.3.19 >>>> >>>> Generate Summary Results = 5000 in 187.6s = 26.6/s Avg: 1842 Min: >>>> 88 Max: 8483 Err: 0 (0.00%) >>>> >>>> The new version is slightly faster now. >>>> >>>> Hans, could you now please revert the changes you did when you switched >>>> back to 2.3.10? I am especially interested in having back the catalog >>>> Directive (instead of the Transform) because in future versions of >>>> Freemarker Transforms will not be supported and we will have to start the >>>> migration to directives: having one in place will definitely help. >>>> >>>> Thanks, >>>> >>>> Jacopo >>>> >>>> >>>> On Apr 2, 2012, at 9:49 AM, Hans Bakker wrote: >>>> >>>>> Very impressive Jacopo: good Job. >>>>> >>>>> see what the developers of freemarker say..... >>>>> >>>>> Regards, >>>>> Hans >>>>> >>>>> On 04/02/2012 02:43 PM, Jacopo Cappellato wrote: >>>>>> An update on this: I have found and resolved the issue that Hans >>>>>> reported on Freemarker bug tracker: >>>>>> >>>>>> https://sourceforge.net/tracker/index.php?func=detail&aid=3046069&group_id=794&atid=100794 >>>>>> >>>>>> Now I am working with them to see if my change can be made official and >>>>>> released. >>>>>> >>>>>> Jacopo >>>>>> >>>>>> On Aug 22, 2010, at 2:52 PM, Hans Bakker wrote: >>>>>> >>>>>>> We are sill working on this to optimize the cache so the newest >>>>>>> freemarker will work again. >>>>>>> >>>>>>> Regards >>>>>>> Hans >>>>>>> >>>>>>> On Sun, 2010-08-22 at 02:38 -0700, Mike Z wrote: >>>>>>>> Any update on this? Thanks. >>>>>>>> >>>>>>>> On Mon, Aug 16, 2010 at 1:58 AM, Hans Bakker >>>>>>>> <[email protected]> wrote: >>>>>>>>> We tried the newer version but is also slow. >>>>>>>>> we are are now looking into moving back to the 2.3.10 version. >>>>>>>>> >>>>>>>>> Regards, >>>>>>>>> Hans >>>>>>>>> >>>>>>>>> On Mon, 2010-08-16 at 09:22 +0200, Jacques Le Roux wrote: >>>>>>>>>> I found only that >>>>>>>>>> http://freemarker.sourceforge.net/docs/app_faq.html#faq_question_25 >>>>>>>>>> So nothing special. >>>>>>>>>> We could try to upgrade to newer versions but 2.3.16 does not seem >>>>>>>>>> to bring much and 2.4 is not ready yet. >>>>>>>>>> >>>>>>>>>> Jacques >>>>>>>>>> >>>>>>>>>> Hans Bakker wrote: >>>>>>>>>>> this problem about the speed is caused in rev: 731487: >>>>>>>>>>> Upgrade Freemarker jar from 2.3.10 to 2.3.15 >>>>>>>>>>> >>>>>>>>>>> Regards, >>>>>>>>>>> Hans >>>>>>>>>>> >>>>>>>>>>> On Sun, 2010-08-15 at 08:45 +0700, Hans Bakker wrote: >>>>>>>>>>>> currently we are helping a customer to upgrade their ofbiz version. >>>>>>>>>>>> We are using "Apache bench" to test the ecommerce frontend. >>>>>>>>>>>> >>>>>>>>>>>> We now found that the current trunk is dramatically slower than rev >>>>>>>>>>>> 728596 from about 1.5 years ago. >>>>>>>>>>>> >>>>>>>>>>>> the test results: >>>>>>>>>>>> rev 728596: >>>>>>>>>>>> total time : 34.667 seconds >>>>>>>>>>>> Requests per second: 14.42 [#/sec] >>>>>>>>>>>> >>>>>>>>>>>> trunk rev 959845 >>>>>>>>>>>> total time : 152.698 seconds >>>>>>>>>>>> Requests per second: 3.27 [#/sec] >>>>>>>>>>>> >>>>>>>>>>>> below the complete test results. >>>>>>>>>>>> >>>>>>>>>>>> Help appreciated. >>>>>>>>>>>> >>>>>>>>>>>> Regards, >>>>>>>>>>>> Hans. >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> Test results, local compute ubuntu 10.04 java memory allocation max >>>>>>>>>>>> 2048Mb >>>>>>>>>>>> running in background, cache enabled. (cache.properties) >>>>>>>>>>>> >>>>>>>>>>>> command line: >>>>>>>>>>>> ab -n 500 -c 50 http://localhost:8080/ecommerce/control/main >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> rev 728596 >>>>>>>>>>>> ----------- >>>>>>>>>>>> Server Software: Apache-Coyote/1.1 >>>>>>>>>>>> Server Hostname: localhost >>>>>>>>>>>> Server Port: 8080 >>>>>>>>>>>> >>>>>>>>>>>> Document Path: /ecommerce/control/main >>>>>>>>>>>> Document Length: 60531 bytes >>>>>>>>>>>> >>>>>>>>>>>> Concurrency Level: 50 >>>>>>>>>>>> Time taken for tests: 34.667 seconds >>>>>>>>>>>> Complete requests: 500 >>>>>>>>>>>> Failed requests: 496 >>>>>>>>>>>> (Connect: 0, Receive: 0, Length: 496, Exceptions: 0) >>>>>>>>>>>> Write errors: 0 >>>>>>>>>>>> Total transferred: 30412908 bytes >>>>>>>>>>>> HTML transferred: 30262908 bytes >>>>>>>>>>>> Requests per second: 14.42 [#/sec] (mean) >>>>>>>>>>>> Time per request: 3466.737 [ms] (mean) >>>>>>>>>>>> Time per request: 69.335 [ms] (mean, across all concurrent >>>>>>>>>>>> requests) >>>>>>>>>>>> Transfer rate: 856.72 [Kbytes/sec] received >>>>>>>>>>>> >>>>>>>>>>>> Connection Times (ms) >>>>>>>>>>>> min mean[+/-sd] median max >>>>>>>>>>>> Connect: 0 198 745.6 0 3000 >>>>>>>>>>>> Processing: 1173 3239 778.2 3228 6287 >>>>>>>>>>>> Waiting: 40 433 318.7 357 1879 >>>>>>>>>>>> Total: 1173 3437 1287.7 3228 9287 >>>>>>>>>>>> >>>>>>>>>>>> rev 959845 >>>>>>>>>>>> =========== >>>>>>>>>>>> Server Software: Apache-Coyote/1.1 >>>>>>>>>>>> Server Hostname: localhost >>>>>>>>>>>> Server Port: 8080 >>>>>>>>>>>> >>>>>>>>>>>> Document Path: /ecommerce/control/main >>>>>>>>>>>> Document Length: 65788 bytes >>>>>>>>>>>> >>>>>>>>>>>> Concurrency Level: 50 >>>>>>>>>>>> Time taken for tests: 152.698 seconds >>>>>>>>>>>> Complete requests: 500 >>>>>>>>>>>> Failed requests: 499 >>>>>>>>>>>> (Connect: 0, Receive: 0, Length: 499, Exceptions: 0) >>>>>>>>>>>> Write errors: 0 >>>>>>>>>>>> Total transferred: 33189975 bytes >>>>>>>>>>>> HTML transferred: 33039975 bytes >>>>>>>>>>>> Requests per second: 3.27 [#/sec] (mean) >>>>>>>>>>>> Time per request: 15269.837 [ms] (mean) >>>>>>>>>>>> Time per request: 305.397 [ms] (mean, across all concurrent >>>>>>>>>>>> requests) >>>>>>>>>>>> Transfer rate: 212.26 [Kbytes/sec] received >>>>>>>>>>>> >>>>>>>>>>>> Connection Times (ms) >>>>>>>>>>>> min mean[+/-sd] median max >>>>>>>>>>>> Connect: 0 2 6.6 0 30 >>>>>>>>>>>> Processing: 11504 15083 1122.0 15065 19496 >>>>>>>>>>>> Waiting: 214 1312 355.2 1294 2751 >>>>>>>>>>>> Total: 11504 15086 1123.5 15065 19497 >>>>>>>>> -- >>>>>>>>> Ofbiz on twitter: http://twitter.com/apache_ofbiz >>>>>>>>> Myself on twitter: http://twitter.com/hansbak >>>>>>>>> Antwebsystems.com: Quality services for competitive rates. >>>>>>>>> >>>>>>>>> >>>>>>> -- >>>>>>> Ofbiz on twitter: http://twitter.com/apache_ofbiz >>>>>>> Myself on twitter: http://twitter.com/hansbak >>>>>>> Antwebsystems.com: Quality services for competitive rates. >>>>>>> >
