"
after the recent days additions to the very core of the forwarding logic (not to mention that llvm15 arrived to debian sid lately) it just gets better and better compared to the last year results… btw http://mp.ls is still a magnitude easier to “route” than srv6 dataplane on this 1 of a mobile core i7 cpu… XDDDDDDDDDb from my side, its almost like *wdm but a layer above and with the g-mpls rfc series, these two could cooperate… XDDDDDD not to mention the encap overhead… from my side srv6 is a 1st class replacement citizen for the “overlay” isps and this imho remain, but well, change my mind?!.. XDDDDDDDDDDDDDDDDD
"
^^^^ a permaling, last year it had a bit less v6 imho…
once the p4interpreter is ready, one can run/trace/debug/reedit a p4 program in real time and have a per line hit count… and as it’s a per cpu performance imho its almost a toffino1 but on an notebook x86-64… XDDD
once again, the
p4interpreter------[ethernet]-----any-of-the-forwarders
is the simplest topology and the topology is also realtime and could be saved, reloaded, etc…
so you can/should freely flood to your p4 program and spot the weaknesses… normally… XDDDDDD
well the interpreter will be a tooons slower, but then heyy, this beast will do it’s best u know to “toss da pakkz”…
just some more popcorn time 4 u all, i have 2 other stuff to do before that… XDDDDDDDDDDD