Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 4 Nov 2007 13:22:08 +0100
From:      Peter Schuller <peter.schuller@infidyne.com>
To:        Ivan Voras <ivoras@freebsd.org>
Cc:        freebsd-current@freebsd.org
Subject:   Re: Status of MySQL on 7 w/o patches applied
Message-ID:  <20071104122206.GA84737@hyperion.scode.org>
In-Reply-To: <ff3f05$29a$1@ger.gmane.org>
References:  <20071016110642.GB54457@hyperion.scode.org> <20071016212307.GA8850@hyperion.scode.org> <ff3f05$29a$1@ger.gmane.org>

next in thread | previous in thread | raw e-mail | index | archive | help

--qDbXVdCdHGoSgWSk
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

> > So to follow-up, with RELENG_7 the performance is now significantly
> > better than 6.2 for me too. I guess it was the debugging options in
> > userland. If someone wants details anyway I'll provide them, but since
> > it is no longer that interesting I'm leaving it as is if no one speaks
> > up.
>=20
> It's always interesting to get "third party" confirmations on such
> benchmarks so please post them :)

Better late than never...

So to recap, these tests were done on single-processer AMD64 machines
with 512 MB of memory. So nothing as interesting as SMP scalability
and such, but still interesting to see an increase in absolute speed
in the simple case. They were also performed maintaly with the 4BSD
scheduler.

Due to the reason for running these tests, they were also made with
MyISAM rather than InnoDB, prepared as such:

sysbench --test=3Doltp \
         --mysql-table-engine=3Dmyisam \
         --oltp-table-size=3D1000000 \
         prepare

And run as such:

sysbench --num-threads=3D16 \
         --max-requests=3D100000 \
         --test=3Doltp \
         --oltp-table-size=3D10000 \
         --mysql-socket=3D/tmp/mysql.sock \
         --oltp-read-only \
         run

The MySQL version is 5.0.45_1 in both cases, compiled from ports with
default build settings, and (on purpose) default MySQL runtime
settings (empty my.cnf).

Nothing in loader.conf on either machine. On the RELENG_7 one I tried
the following, one at a time, but none of them caused a different time
source to be chosen (probably because I was doing it the wrong way):

kern.timecounter.tc.TSC.quality=3D"1000"
kern.timecounter.hardware=3D"TSC"
kern.timecounter.choice=3D"TSC(1200) ACPI-fast(1000) i8254(0) dummy(-100000=
0)"
kern.timecounter.smp_tsc=3D"1"

CPU details (6.2 machine, then RELENG_7 machine):


CPU: AMD Athlon(tm) 64 Processor 3200+ (1989.81-MHz K8-class CPU)
  Origin =3D "AuthenticAMD"  Id =3D 0x40ff2  Stepping =3D 2
  Features=3D0x78bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE=
,MCA,CMOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2>
  Features2=3D0x2001<SSE3,CX16>
  AMD Features=3D0xea500800<SYSCALL,NX,MMX+,FFXSR,RDTSCP,LM,3DNow+,3DNow>
  AMD Features2=3D0x1d<LAHF,<b2>,<b3>,CR8>

CPU: AMD Athlon(tm) 64 Processor 3200+ (1989.82-MHz K8-class CPU)
  Origin =3D "AuthenticAMD"  Id =3D 0x40ff2  Stepping =3D 2
  Features=3D0x78bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE=
,MCA,CMOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2>
  Features2=3D0x2001<SSE3,CX16>
  AMD Features=3D0xea500800<SYSCALL,NX,MMX+,FFXSR,RDTSCP,LM,3DNow!+,3DNow!>
  AMD Features2=3D0x1d<LAHF,SVM,ExtAPIC,CR8>

And the sysbench results:

6.2:

OLTP test statistics:
    queries performed:
        read:                            1400000
        write:                           0
        other:                           200000
        total:                           1600000
    transactions:                        100000 (261.10 per sec.)
    deadlocks:                           0      (0.00 per sec.)
    read/write requests:                 1400000 (3655.45 per sec.)
    other operations:                    200000 (522.21 per sec.)

Test execution summary:
    total time:                          382.9903s
    total number of events:              100000
    total time taken by event execution: 6125.6312
    per-request statistics:
         min:                            0.0028s
         avg:                            0.0613s
         max:                            0.8482s
         approx.  95 percentile:         0.0697s

Threads fairness:
    events (avg/stddev):           6250.0000/94.01
    execution time (avg/stddev):   382.8520/0.01

RELENG_7 from October 16:

OLTP test statistics:
    queries performed:
        read:                            1400000
        write:                           0
        other:                           200000
        total:                           1600000
    transactions:                        100000 (292.50 per sec.)
    deadlocks:                           0      (0.00 per sec.)
    read/write requests:                 1400000 (4095.01 per sec.)
    other operations:                    200000 (585.00 per sec.)

Test execution summary:
    total time:                          341.8797s
    total number of events:              100000
    total time taken by event execution: 5468.4663
    per-request statistics:
         min:                            0.0028s
         avg:                            0.0547s
         max:                            0.8707s
         approx.  95 percentile:         0.0561s

Threads fairness:
    events (avg/stddev):           6250.0000/124.43
    execution time (avg/stddev):   341.7791/0.01

--=20
/ Peter Schuller

PGP userID: 0xE9758B7D or 'Peter Schuller <peter.schuller@infidyne.com>'
Key retrieval: Send an E-Mail to getpgpkey@scode.org
E-Mail: peter.schuller@infidyne.com Web: http://www.scode.org


--qDbXVdCdHGoSgWSk
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.4 (FreeBSD)

iD8DBQFHLbluDNor2+l1i30RAu+QAJ9CN2vbEykxyz3mtJvL4GZUT6uM1wCfc8Xo
i12yQ/2PIlQq9w+Ue0giho0=
=4fx8
-----END PGP SIGNATURE-----

--qDbXVdCdHGoSgWSk--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20071104122206.GA84737>