>>> On 11.06.15 at 17:28, wrote:
> At 15:51 +0100 on 05 Jun (1433519478), Jan Beulich wrote:
>> Iirc before both of these changes, and the v10 ones imo should
>> have invalidated it. Tim, I'm particularly trying to understand
>> whether you're okay with the original's (potentially even heavier)
>>
At 15:51 +0100 on 05 Jun (1433519478), Jan Beulich wrote:
> >>> On 02.06.15 at 18:26, wrote:
> > Performance analysis of aggregate network throughput with many VMs
> > shows that performance is signficantly limited by contention on the
> > maptrack lock when obtaining/releasing maptrack handles fr
>>> On 05.06.15 at 18:42, wrote:
> On 05/06/15 17:11, Jan Beulich wrote:
> On 05.06.15 at 17:55, wrote:
>>> On 05/06/15 15:51, Jan Beulich wrote:
>>> On 02.06.15 at 18:26, wrote:
> +/*
> + * max_maptrack_frames is per domain so each VCPU gets a share of
> + * the
On 05/06/15 17:11, Jan Beulich wrote:
On 05.06.15 at 17:55, wrote:
>> On 05/06/15 15:51, Jan Beulich wrote:
>> On 02.06.15 at 18:26, wrote:
+/*
+ * max_maptrack_frames is per domain so each VCPU gets a share of
+ * the maximum, but allow at least one frame per
>>> On 05.06.15 at 17:55, wrote:
> On 05/06/15 15:51, Jan Beulich wrote:
> On 02.06.15 at 18:26, wrote:
>>> +/*
>>> + * max_maptrack_frames is per domain so each VCPU gets a share of
>>> + * the maximum, but allow at least one frame per VCPU.
>>> + */
>>> +if ( v->maptrack
On 05/06/15 15:51, Jan Beulich wrote:
On 02.06.15 at 18:26, wrote:
>> Performance analysis of aggregate network throughput with many VMs
>> shows that performance is signficantly limited by contention on the
>> maptrack lock when obtaining/releasing maptrack handles from the free
>> list.
>>
>>> On 02.06.15 at 18:26, wrote:
> Performance analysis of aggregate network throughput with many VMs
> shows that performance is signficantly limited by contention on the
> maptrack lock when obtaining/releasing maptrack handles from the free
> list.
>
> Instead of a single free list use a per-V
On 02/06/15 17:26, David Vrabel wrote:
> Performance analysis of aggregate network throughput with many VMs
> shows that performance is signficantly limited by contention on the
> maptrack lock when obtaining/releasing maptrack handles from the free
> list.
Sometime during all this refactoring, I
Performance analysis of aggregate network throughput with many VMs
shows that performance is signficantly limited by contention on the
maptrack lock when obtaining/releasing maptrack handles from the free
list.
Instead of a single free list use a per-VCPU list. This avoids any
contention when obta
