Menu
▾
▴
Re: [WWdevel] relative speed of sql_single and gdbm
|
From: Arnold P. <ap...@ma...> - 2005-03-16 20:30:53
|
At 02:57 PM 3/16/2005, John Jones wrote: Hi John, I made the change using the original version of SQL.pm, but removed the word binary from it. Here are the results: [Wed Mar 16 15:10:29 2005] 57113 1111003829 - [/webwork2/mth162/7/1/] [runTime = 0.0 sec sql_single] [Wed Mar 16 15:11:17 2005] 53856 1111003877 - [/webwork2/mth162/7/1/] [runTime = 5.0 sec sql_single] [Wed Mar 16 15:11:27 2005] 53856 1111003887 - [/webwork2/mth162/7/1/] [runTime = 6.0 sec sql_single] [Wed Mar 16 15:11:34 2005] 53856 1111003894 - [/webwork2/mth162/7/1/] [runTime = 5.0 sec sql_single] Not much difference. The 0.0 was viewing the problem for the first time and the next three were submitting answers. I also decided to try viewing other problems for the first time with the following results: [Wed Mar 16 15:17:10 2005] 54980 1111004230 - [/webwork2/mth162/7/2/] [runTime = 0.0 sec sql_single] [Wed Mar 16 15:17:14 2005] 53856 1111004234 - [/webwork2/mth162/7/3/] [runTime = 2.0 sec sql_single] [Wed Mar 16 15:17:15 2005] 54980 1111004235 - [/webwork2/mth162/7/4/] [runTime = 0.0 sec sql_single] [Wed Mar 16 15:17:18 2005] 53856 1111004238 - [/webwork2/mth162/7/5/] [runTime = 2.0 sec sql_single] [Wed Mar 16 15:17:22 2005] 53856 1111004242 - [/webwork2/mth162/7/6/] [runTime = 3.0 sec sql_single] Then I just submitted blank answers with the following results: [Wed Mar 16 15:17:42 2005] 55114 1111004262 - [/webwork2/mth162/7/6/] [runTime = 3.0 sec sql_single] [Wed Mar 16 15:17:46 2005] 55114 1111004266 - [/webwork2/mth162/7/6/] [runTime = 2.0 sec sql_single] [Wed Mar 16 15:17:50 2005] 55114 1111004270 - [/webwork2/mth162/7/6/] [runTime = 3.0 sec sql_single] [Wed Mar 16 15:17:53 2005] 55114 1111004273 - [/webwork2/mth162/7/6/] [runTime = 2.0 sec sql_single] and now with simple wrong answers: [Wed Mar 16 15:26:23 2005] 56205 1111004783 - [/webwork2/mth162/7/6/] [runTime = 2.0 sec sql_single] [Wed Mar 16 15:26:30 2005] 53856 1111004790 - [/webwork2/mth162/7/6/] [runTime = 6.0 sec sql_single] [Wed Mar 16 15:26:37 2005] 53856 1111004797 - [/webwork2/mth162/7/6/] [runTime = 6.0 sec sql_single] [Wed Mar 16 15:26:43 2005] 56205 1111004803 - [/webwork2/mth162/7/6/] [runTime = 3.0 sec sql_single] [Wed Mar 16 15:26:47 2005] 56205 1111004807 - [/webwork2/mth162/7/6/] [runTime = 2.0 sec sql_single] [Wed Mar 16 15:26:51 2005] 54980 1111004811 - [/webwork2/mth162/7/] [runTime = 1.0 sec sql_single] I don't know if I can tell anything from the above or not. I seems submitting an answer does slow stuff down somewhat. How many BD calls are involved in that process? Arnie >Arnold Pizer wrote: > >>At 02:06 PM 3/10/2005, John Jones wrote: >> >>Hi John, >> >>Thanks very much for the patch. I put it in and tested it a little bit. I >>haven't seen any problems but also I don't see a great deal of >>improvement in speed. I just tested it on one problem inserting correct >>and incorrect answers. With the original SQL.pm it took 8 seconds (I >>only did one attempt) and with the new SQL.pm between 2 and 8 seconds >>(details below). The load avg. on the server was about .35 and mysql >>cpu usage was about 30 or 40%. I guess we will have to wait until my >>next assignment is due to see if this makes a difference but the times >>below are still way too long. > >Hi Arnie, > >One thing you can do for comparison is use the old version of SQL.pm, but >remove the word binary from it (it appears in one place). This should be >ok provided you do not have 2 set names or 2 user id's which differ only >in case. This will show the maximum speed gain you can get from working >around this indexing problem. When assigning a set to a class, changing >this was something like a factor of 10 in speed between the original and >the original with "binary" removed. > >John > Prof. Arnold K. Pizer Dept. of Mathematics University of Rochester Rochester, NY 14627 (585) 275-7767 |
