With local MariaDB 10.0.13. out put from SQLyog "Messages"in erface: http://faq.webyog.com/content/8/147/en/how-shall-i-understand-the-_query-exe... *1 queries executed, 1 success, 0 errors, 0 warnings* *Query: SELECT BENCHMARK(50000000, (SELECT 1))* *1 row(s) returned* *Execution Time : 0.267 sec* *Transfer Time : 0 sec* *Total Time : 0.267 sec* Yes, I recently had a terrific machine. Intel socket 1155 (2 threads per core) 4 cores 3.5 Ghz,32 MB RAM. :-) GPU is nothing particlar. It is a 512 MB DDR2-RAMGeforce card (used in order to achive 2560*1440 screen resolution , what the intel CPU graphics does not provide. -- Peter On Tue, Mar 3, 2015 at 2:28 PM, Justin Swanhart <greenlion@gmail.com> wrote:
You probably have either a faster CPU or bigger cache on your cpu. How long does benchmark select 1 take? You should find it is faster on your system too, right?
--Justin
Sent from my iPhone
On Mar 3, 2015, at 6:18 AM, Peter Laursen <peter_laursen@webyog.com> wrote:
Your 3rd test case takes 2:02 in MariaDB 10.1 and 2:13 in MySQL 5.6 on my system (when otherwise idle) .
-- Peter
On Tue, Mar 3, 2015 at 1:58 PM, Federico Razzoli <federico_raz@yahoo.it> wrote:
I made some quick test to show what I mean by "performance problem". Note that I'm not saying the the first and the second test cases should perform equally. But there is too much difference - see the conclusion.
1)
MariaDB [test]> SELECT BENCHMARK(50000000, (SELECT 1)); +---------------------------------+ | BENCHMARK(50000000, (SELECT 1)) | +---------------------------------+ | 0 | +---------------------------------+ 1 row in set (0.42 sec)
2)
DELIMITER || CREATE FUNCTION f() RETURNS TINYINT BEGIN RETURN 1; END || DELIMITER ;
MariaDB [test]> SELECT BENCHMARK(50000000, (SELECT f())); +-----------------------------------+ | BENCHMARK(50000000, (SELECT f())) | +-----------------------------------+ | 0 | +-----------------------------------+ 1 row in set (2 min 5.70 sec)
3)
DELIMITER || CREATE FUNCTION f(x TINYINT) RETURNS TINYINT BEGIN RETURN x; END || DELIMITER ;
MariaDB [test]> SELECT BENCHMARK(50000000, (SELECT f(1))); +------------------------------------+ | BENCHMARK(50000000, (SELECT f(1))) | +------------------------------------+ | 0 | +------------------------------------+ 1 row in set (3 min 35.20 sec)
-- Conclusions:
Times in seconds:
0.42 125.70 215.20
I don't know which ratio would be acceptable, but the difference betweem a trivial query and a trivial function is too high. Also, the difference between 2) and 3) is that the function in 3) accepts and returns a parameter. Again, the performance difference seems to me too high.
Regards Federico
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp