Re: [PATCH v4 00/17] Wind River Systems AVP PMD vs virtio?

On Mar 15, 2017, at 10:02 PM, Vincent JARDIN [off-list ref] wrote:

Le 15/03/2017 à 05:10, O'Driscoll, Tim a écrit :

quoted

quoted

so, still an nack because:
  - no performance data of avp vs virtio,

I don't think it should be a requirement for Allain to provide performance data in order to justify getting this accepted into DPDK. Keith pointed out in a previous comment on this patch set that even if performance is the same as virtio, there might still be other reasons why people would want to use it.

quoted

  - 2013 is gone,
  - it unfocuses from virtio.

Tim,

you get it wrong, it IS the major point: if AVP is good in performance, then an alternative to virtio is needed.

Please, stop turning around the topic, and send facts/numbers that demonstrate that there is a value in having alternative to virtio. Currently, the only argument is a code developed in 2013 needs to  be upstreamed because vhost-user was not available in 2013.

I am sorry Vincent, but no one else provided performance data on any of the PMDs I did not have to provide performance data on TAP compared to AF_PACKET. Then why would you need to require performance data on this PMD, just because it uses something similar to what we have in some other PMDs is not a reason. The core code of DPKD is not PMDs is very important part, but they are just drivers for data I/O and DPDK provides the framework to run these drivers.

Tim is not turning around the topic here, only pointing out the real questions and statements on this topic.

Best regards,
 Vincent

Regards,
Keith