Jeremy Zawodny wrote:

>On Fri, Oct 03, 2003 at 06:23:24PM -0400, Ware Adams wrote:
>>Peer Reiser wrote:
>>
>>>Next week I will have access to a new PomerMac G5 with Dual 2GHZ
>>>processors, and i want to do some indexing. Does anyone know if
>>>MySQL will take advantage of dual processors if the only process
>>>running is the indexing process??
>>
>>No, it won't directly.  However, other processes going on will use
>>the 2nd CPU (non-mysql processes) and if you run other queries they
>>will use it.
>
>Really?  About a year ago, when I asked an Apple engineer about theith
>SMP and threading support, he was able to convince me that it didn't
>suffer from the FreeBSD 4.x limitations.
>
>Have you seen documentation that really describes OS X's
>implementation?  I'd love to know the truth. :-)

I haven't seen any documentation, and I'm not sure I'm explaining things
properly, but here's what I've observed running MySQL on a decent sized
data set over almost a year on OS X:

When only one query is active in MySQL (observed via show processlist, all
connection IDs show 'sleep' except one) the mysqld process in top never
shows more than 100% (or never more than 105-110% to be absolutely
truthful)

When multiple queries are active in MySQL the mysqld process frequently
approaches 200% (assuming each can hit 100% when run on it's own)

When a single MySQL query is active and another heavy load process is
running on the machine (e.g. running rsync on a big directory) mysqld will
go to 100% and the other process will approach the level it would hit
without mysqld running

This is on a G4 1.42GHz dual proc running OS X and hooked up to an XServe
RAID.

My conclusions from this were that MySQL on OS X cannot use more than one
processor for a single query, but it uses multiple ones fine when it has
multiple queries to process.  Also, it performs fine sharing the two
processors with other applications.

Can MySQL use multiple processors for a single query on other OS's?  It's
pretty key for us as we tend to run relatively few, long duration queries
as opposed to lots of quick ones.

Thanks,
Ware

-- 
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe:    http://lists.mysql.com/[EMAIL PROTECTED]

Reply via email to