Re: Question related to MySQL and Threads and "Context_voluntary"

While I would like to have my own octo-core Mac, I suspect that it might be part of the problem. Do you have a Linux box on which you can run this test?

On 7/19/07, Mariella Petrini <mariellapetrini@yahoo.com> wrote:
> The same problem happens with only MyISAM table
> storage engine as well.
>
> The same problem occurs with 5.0.41 (32 bits compiled)
> and 5.1.20 (64 bits compiled)
>
>
> The platform is the following:
> Mac Pro (MacPro2,1)
> Quad-Core Intel Xeon, 3GHz and 2 processors.
> Total number of cores: 8
> L2 Cache (per processor): 8 MB
> Memory: 8 GB
> Bus Speed: 1.33 GHz
> Boot ROM Version: MP21.007F.B02
>
> (Kernel: 8.10.1 Darwin Kernel Version 8.10.1: Wed May
> 23 16:33:00 PDT 2007; root:xnu-792.22.5~1/RELEASE_I386
> i386 i386)
>
>
> The high increase in number of voluntary context
> switches and messages exchanged when the parallel
> increases might might be a clue multicore could not be
> exploited properly.
>
>
> Thanks in advance for your help,
>
> Mariella
>
>
> --- Mark Callaghan <mcallaghan@google.com> wrote:
>
> > As you have stated, there is no IO activity during
> > your test, so the
> > disk config doesn't matter. The dataset uses fewer
> > than 1000 pages in
> > the InnoDB buffer cache.
> >
> > From the iostat output, the CPU usage for system
> > time is huge (24% to
> > 29%) while user time is low (0% to 1%).
> >
> > I have benchmarks that do concurrent queries from
> > multiple InnoDB
> > tables on 4 CPU core servers. The tests are
> > different in that they
> > select all rows, so there is some network IO where
> > yours only return 1
> > row. But the system time never exceeds 5% for 4, 8
> > or 16 concurrent
> > clients.
> >
> > Do you also get high system time from queries on
> > MyISAM tables? InnoDB
> > had some issues (mostly fixed) with multi-core
> > servers, but it
> > shouldn't be this bad. I am curious if your platform
> > (MacOS X) might
> > be part of the problem. My tests used Linux 2.6 on
> > x86.
> >
> > On 7/19/07, Mariella Petrini
> > <mariellapetrini@yahoo.com> wrote:
> > >
> > > --- Mark Callaghan <mcallaghan@google.com> wrote:
> > >
> > > > On 7/19/07, Mariella Petrini
> > > > <mariellapetrini@yahoo.com> wrote:
> > > > > Hi,
> > > > >
> > > > >
> > > > > I have been working with MySQL 5.1.20 (64 bits
> > > > > compiled on Mac OS for MacPro Intel Xeon
> > Quad-Core
> > > > > qith 2 processors and 8GB of RAM) and MySQL
> > 5.0.41
> > > > (32
> > > > > bits).
> > > >
> > > > Does that mean the host has 8 cores?
> > > Correct
> > >
> > > > What is the disk configuration?
> > >
> > > Do you mean
> > > df -k
> > > Filesystem 1K-blocks Used
> > Avail
> > > Capacity Mounted on
> > > /dev/disk1s2 488050672 157171460
> > 330623212
> > > 32% /
> > > devfs 97 97
> > 0
> > > 100% /dev
> > > fdesc 1 1
> > 0
> > > 100% /dev
> > > <volfs> 512 512
> > 0
> > > 100% /.vol
> > > /dev/disk0s2 488050672 70520
> > 487980152
> > > 0% /Volumes/NotQuiteMaxDisk
> > > automount -nsl [206] 0 0
> > 0
> > > 100% /Network
> > > automount -fstab [223] 0 0
> > 0
> > > 100% /automount/Servers
> > > automount -static [223] 0 0
> > 0
> > > 100% /automount/static
> > >
> > >
> > > or what type of RAID level is used ?
> > >
> > >
> > >
> > >
> > >
> > > >
> > > > >
> > > > > While running some bnchmarks of the select
> > > > statements
> > > > > (independently of the storage engine used,
> > either
> > > > > InnoDB or MyISAM) I have noticed that
> > > > > the response time of a query degrades badly
> > > > (double
> > > > > degrades) as soon as another query is run.
> > > >
> > > > Is the server IO or CPU bound during the query?
> > > > vmstat and iostat will
> > > > help you determine that.
> > > >
> > > > >
> > > > > E.g. This is a case for a very simple query
> > that
> > > > faces
> > > > > the problem.
> > > > >
> > > > > SELECT SQL_NO_CACHE count(DISTINCT user_id)
> > as
> > > > > res_count FROM t1 WHERE (acc > 0 AND cat=34
> > AND
> > > > thr
> > > > > between 42 and 52) AND genre='SF';
> > > > >
> > > > > Using explain
> > > > >
> > > > >
> > > >
> > >
> >
> +----+-------------+---------------------------+------+-------------------+---------------+---------+-------+-------+-------------+
> > > > > | id | select_type | table
> > |
> > > > type
> > > > > | possible_keys | key | key_len
> > |
> > > > ref
> > > > > | rows | Extra |
> > > > >
> > > >
> > >
> >
> +----+-------------+---------------------------+------+-------------------+---------------+---------+-------+-------+-------------+
> > > > > | 1 | SIMPLE | t1 | ref | uag,genre_idx
> > |
> > > > > genre_idx | 6 | const | 61324 | Using
> > where
> > > > |
> > > > >
> > > >
> > >
> >
> +----+-------------+---------------------------+------+-------------------+---------------+---------+-------+-------+-------------+
> > > > >
> > > > >
> > > > > show index from t1;
> > > > >
> > > >
> > >
> >
> +---------------------------+------------+---------------+--------------+-----------------+-----------+-------------+----------+--------+------+------------+---------+
> > > > > | Table | Non_unique | Key_name |
> > > > Seq_in_index |
> > > > > Column_name | Collation | Cardinality |
> > > > Sub_part |
> > > > > Packed | Null | Index_type | Comment |
> > > > >
> > > >
> > >
> >
> +---------------------------+------------+---------------+--------------+-----------------+-----------+-------------+----------+--------+------+------------+---------+
> > > > > | t1 | 1 | user_id_idx |
> > > > 1 |
> > > > > user_id | A | 178219 |
> > > > NULL |
> > > > > NULL | | BTREE | |
> > > > > | t1 | 1 | uag |
> > > > 1 |
> > > > > acc | A | 8 |
> > > > NULL |
> > > > > NULL | | BTREE | |
> > > > > | t1 | 1 | uag |
> > > > 2 |
> > > > > cat | A | 8 |
> > > > NULL |
> > > > > NULL | | BTREE | |
> > > > > | t1 | 1 | uag |
> > > > 3 |
> > > > > thr | A | 538 |
> > > > NULL |
> > > > > NULL | | BTREE | |
> > > > > | t1 | 1 | genre_idx |
> > > > 1 |
> > > > > genre | A | 6 |
> > > > NULL |
> > > > > NULL | | BTREE | |
> > > > >
> > > >
> > >
> >
> +---------------------------+------------+---------------+--------------+-----------------+-----------+-------------+----------+--------+------+------------+---------+
> > > > >
> > > > >
> > > > > If there is only one query running the
> > response
> > > > time
> > > > > is x, if there two queries running at the same
> > > > time
> > > > > the response time of each select is
> > approximately
> > > > 2x,
> > > > > if there 3 queries running the response time
> > is
> > > > > approximately 4x and if there are 4 queries
> >
> === message truncated ===
>
>
>
> ____________________________________________________________________________________
> Luggage? GPS? Comic books?
> Check out fitting gifts for grads at Yahoo! Search
> http://search.yahoo.com/search?fr=oni_on_mail&p=graduation+gifts&cs=bz
>

-- 
Mark Callaghan
mcallaghan@google.com

-- 
MySQL Server Benchmarks Mailing List
For list archives: 
http://lists.mysql.com/benchmarks
To unsubscribe:    
http://lists.mysql.com/benchmarks?unsub=lists@pantek.com