RE: Mainline kernel OLTP performance update

[Styner, Douglas W]

Working on it as we speak...

>-----Original Message-----
>From: James Bottomley [mailto:James.Bottomley@xxxxxxxxxxxxxxxxxxxxx]
>Sent: Tuesday, April 28, 2009 10:16 AM
>To: Chuck Ebbert
>Cc: Andi Kleen; Styner, Douglas W; linux-kernel@xxxxxxxxxxxxxxx; linux-
>driver@xxxxxxxxxx; linux-scsi@xxxxxxxxxxxxxxx
>Subject: Re: Mainline kernel OLTP performance update
>
>On Tue, 2009-04-28 at 12:57 -0400, Chuck Ebbert wrote:
>> On Mon, 27 Apr 2009 09:02:40 +0200
>> Andi Kleen <andi@xxxxxxxxxxxxxx> wrote:
>>
>> > "Styner, Douglas W" <douglas.w.styner@xxxxxxxxx> writes:
>> >
>> > >
>> > > ======oprofile 0.9.3 CPU_CLK_UNHALTED for top 30 functions
>> > > Cycles% 2.6.24.2                   Cycles% 2.6.30-rc2
>> > > 74.8578 <database>                   67.6966 <database>
>> >
>> > The dip in database cycles is indeed worrying.
>> >
>> > > 1.0500 qla24xx_start_scsi          1.1724 qla24xx_start_scsi
>> > > 0.8089 schedule                    1.0578 qla24xx_intr_handler
>> > > 0.5864 kmem_cache_alloc            0.8259 __schedule
>> > > 0.4989 __blockdev_direct_IO        0.7451 kmem_cache_alloc
>> > > 0.4357 __sigsetjmp                 0.4872 __blockdev_direct_IO
>> > > 0.4152 copy_user_generic_string    0.4390 task_rq_lock
>> > > 0.3953 qla24xx_intr_handler        0.4338 __sigsetjmp
>> >
>> > And also why the qla24xx_intr_handler became ~2.5x as expensive.
>> > Cc linux-scsi and qla24xx maintainers.
>> >
>>
>> They are getting 31000 interrupts/sec vs. 22000/sec on older kernels.
>
>Should be fixed by:
>
>http://marc.info/?l=linux-scsi&m=124093712114937
>
>If someone could verify, I'd be grateful.
>
>Thanks,
>
>James
>

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/