Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-30 Thread Jan Scheurich
gt;; Ilya Maximets > <i.maxim...@samsung.com>; ovs-dev@openvswitch.org; Bhanuprakash > Bodireddy <bhanuprakash.bodire...@intel.com> > Cc: Heetae Ahn <heetae82....@samsung.com> > Subject: Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output > batching. > &

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-30 Thread Ilya Maximets
Hi Darrell, Yes, I'm still interested in this series. I had a lot of other work last few weeks. I hope that I'll have enough time to reply to the questions and concerns in a near future. Best regards, Ilya Maximets. On 30.08.2017 01:45, Darrell Ball wrote: > Hi Jan/Ilya/Bhanu > > Just

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-29 Thread Darrell Ball
Hi Jan/Ilya/Bhanu Just wondering if we are still pursuing this patch series ? Thanks Darrell On 8/14/17, 8:33 AM, "ovs-dev-boun...@openvswitch.org on behalf of Jan Scheurich" wrote: > > We have tested the effect

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-14 Thread Jan Scheurich
> > We have tested the effect of turbo mode on TSC and there is none. The > TSC frequency remains at the nominal clock speed, no matter if the core is > clocked down or up. So, I believe for PMD threads (where performance > matters) TSC would be an adequate and efficient clock. > > It's highly

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-14 Thread Ilya Maximets
On 14.08.2017 16:12, Jan Scheurich wrote: >>> >From earlier in-house trials we know we need to target flush times of 50 >> us or less, so we clearly need better time resolution. Sub-ms timing in PMD >> should be based on TSC cycles, which are already kept in the pmd struct. >> Could you provide a

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-14 Thread Jan Scheurich
> >>From earlier in-house trials we know we need to target flush times of 50 > us or less, so we clearly need better time resolution. Sub-ms timing in PMD > should be based on TSC cycles, which are already kept in the pmd struct. > Could you provide a corresponding patch for performance testing? >

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-14 Thread Ilya Maximets
On 14.08.2017 02:33, Jan Scheurich wrote: >> This allows to collect packets from more than one RX burst >> and send them together with a configurable maximum latency. >> >> 'other_config:output-max-latency' can be used to configure >> time that a packet can wait in output batch for sending. >> >>

Re: [ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-13 Thread Jan Scheurich
> This allows to collect packets from more than one RX burst > and send them together with a configurable maximum latency. > > 'other_config:output-max-latency' can be used to configure > time that a packet can wait in output batch for sending. > > Signed-off-by: Ilya Maximets

[ovs-dev] [PATCH RFC v3 4/4] dpif-netdev: Time based output batching.

2017-08-10 Thread Ilya Maximets
This allows to collect packets from more than one RX burst and send them together with a configurable maximum latency. 'other_config:output-max-latency' can be used to configure time that a packet can wait in output batch for sending. Signed-off-by: Ilya Maximets ---