> On Feb 27, 2017, at 4:15 AM, Legacy, Allain <allain.leg...@windriver.com> > wrote: > >> -----Original Message----- >> From: Vincent JARDIN [mailto:vincent.jar...@6wind.com] > ... >> So, before spending too much time in reviewing your long serie, I think that >> proper statements are needed. In the past, other NICs using Qemu have >> been sent, but they were avoided since virtio solved all the issues. >> > Ok, I'll put together some additional information. Should I resubmit the > patch series with an updated cover letter, update the NIC guide doc, or just > reply to this thread? > > So is the intent to only have a single qemu based device/driver then? If > so, why?
Allain, I think the best information is not really performance, big is a good way to prove its worth. If you can list the advantages and differences with AVP would help understand why someone would like to use AVP. Even if the performance was same it may have many other advantages that would be good to include in DPDK. All that said I do not think we should be limiting PMDs in DPDK just because they look similar to some other PMD. They are taking ownership of the maintenance of the PMD and if for some reason the PMD becomes not maintained then we can discuss removing the PMD later. We have a number of PMDs now that perform the same function only in a different way, which seems just fine with everyone. Just my $0.02 worth. > Regards, Keith
