Lewis Donzis
02/04/2010 8:37 PM
post46609
|
I don't think we would mind, but there are a few issues right now:
1. We did all of our development on the 82599, so we don't even have a '598 to test with right now,
2. For expediency and because we have a large code base of filters for testing, we initially developed the driver for io
-net, and figured we'd just re-do it for io-pkt later, and
3. The chip is not only an order of magnitude faster than the 1G chips, it's an order of magnitude more complicated and
loaded with "features." Getting a functional driver that sends and receives packets and generally works the way you
would expect wasn't too bad, but we've been spending a lot of time fooling with the various advanced features to see how
(and whether) they work, along with their respective benefit (which in some cases is negative!)
So in other words, we're not opposed to it, but we're not there yet.
lew
|
|
|