On Fri, Jul 15, 2016 at 11:18 AM, Roman Penyaev
wrote:
> On Wed, Jul 13, 2016 at 1:45 PM, Kevin Wolf wrote:
>> Am 13.07.2016 um 13:33 hat Roman Penyaev geschrieben:
>>> Just to be sure that we are on the same page:
>>>
>>> 1. We have this commit "linux-aio: Cancel BH if not needed" which
>>>
>>>
On 15/07/2016 13:35, Roman Penyaev wrote:
> On Fri, Jul 15, 2016 at 12:17 PM, Roman Penyaev
> wrote:
>> On Fri, Jul 15, 2016 at 11:58 AM, Paolo Bonzini wrote:
>>>
>>>
>>> On 15/07/2016 11:18, Roman Penyaev wrote:
Those 3 red spikes and a blue hill is what we have to focus on. The
blu
On Fri, Jul 15, 2016 at 12:17 PM, Roman Penyaev
wrote:
> On Fri, Jul 15, 2016 at 11:58 AM, Paolo Bonzini wrote:
>>
>>
>> On 15/07/2016 11:18, Roman Penyaev wrote:
>>> Those 3 red spikes and a blue hill is what we have to focus on. The
>>> blue hill at the right corner of the chart means that alm
On 15/07/2016 12:17, Roman Penyaev wrote:
> On Fri, Jul 15, 2016 at 11:58 AM, Paolo Bonzini wrote:
>>
>>
>> On 15/07/2016 11:18, Roman Penyaev wrote:
>>> Those 3 red spikes and a blue hill is what we have to focus on. The
>>> blue hill at the right corner of the chart means that almost always t
On Fri, Jul 15, 2016 at 11:58 AM, Paolo Bonzini wrote:
>
>
> On 15/07/2016 11:18, Roman Penyaev wrote:
>> Those 3 red spikes and a blue hill is what we have to focus on. The
>> blue hill at the right corner of the chart means that almost always the
>> ring buffer was observed as full, i.e. qemu_l
On 15/07/2016 11:18, Roman Penyaev wrote:
> Those 3 red spikes and a blue hill is what we have to focus on. The
> blue hill at the right corner of the chart means that almost always the
> ring buffer was observed as full, i.e. qemu_laio_completion_bh() got
> a chance to reap completions not very
On Wed, Jul 13, 2016 at 1:45 PM, Kevin Wolf wrote:
> Am 13.07.2016 um 13:33 hat Roman Penyaev geschrieben:
>> Just to be sure that we are on the same page:
>>
>> 1. We have this commit "linux-aio: Cancel BH if not needed" which
>>
>>a) introduces performance regression on my fio workloads on t
On Wed, Jul 13, 2016 at 09:57:09AM +0200, Roman Pen wrote:
Please send each new revision of a patch series as a separate email
thread. Do not thread revisions with Reply-To:, References:.
See http://qemu-project.org/Contribute/SubmitAPatch for all the
guidelines on submitting patches.
> v1..v2:
On Wed, Jul 13, 2016 at 1:45 PM, Kevin Wolf wrote:
> Am 13.07.2016 um 13:33 hat Roman Penyaev geschrieben:
>> Just to be sure that we are on the same page:
>>
>> 1. We have this commit "linux-aio: Cancel BH if not needed" which
>>
>>a) introduces performance regression on my fio workloads on t
On Wed, Jul 13, 2016 at 2:22 PM, Eric Blake wrote:
> On 07/13/2016 01:57 AM, Roman Pen wrote:
>> v1..v2:
>>
>> o comment tweaks.
>> o fix QEMU coding style.
>
> The above comments should be delayed...
>
>>
>> Invoking io_setup(MAX_EVENTS) we ask kernel to create ring buffer for us
>> with spec
On 07/13/2016 01:57 AM, Roman Pen wrote:
> v1..v2:
>
> o comment tweaks.
> o fix QEMU coding style.
The above comments should be delayed...
>
> Invoking io_setup(MAX_EVENTS) we ask kernel to create ring buffer for us
> with specified number of events. But kernel ring buffer allocation logi
Am 13.07.2016 um 13:33 hat Roman Penyaev geschrieben:
> Just to be sure that we are on the same page:
>
> 1. We have this commit "linux-aio: Cancel BH if not needed" which
>
>a) introduces performance regression on my fio workloads on the
> following config: "iothread=1, VCPU=8, MQ=8".
On Wed, Jul 13, 2016 at 12:31 PM, Paolo Bonzini wrote:
>
>
> On 13/07/2016 09:57, Roman Pen wrote:
>> v1..v2:
>>
>> o comment tweaks.
>> o fix QEMU coding style.
>>
>> Invoking io_setup(MAX_EVENTS) we ask kernel to create ring buffer for us
>> with specified number of events. But kernel ring
On 13/07/2016 09:57, Roman Pen wrote:
> v1..v2:
>
> o comment tweaks.
> o fix QEMU coding style.
>
> Invoking io_setup(MAX_EVENTS) we ask kernel to create ring buffer for us
> with specified number of events. But kernel ring buffer allocation logic
> is a bit tricky (ring buffer is page si
v1..v2:
o comment tweaks.
o fix QEMU coding style.
Invoking io_setup(MAX_EVENTS) we ask kernel to create ring buffer for us
with specified number of events. But kernel ring buffer allocation logic
is a bit tricky (ring buffer is page size aligned + some percpu allocation
are required) so eve
15 matches
Mail list logo
