Slashdot Mirror

← Back to Stories (view on slashdot.org)

986MB/s With BSD And Gigabit Ethernet

Posted by Nik on from the free-for-all dept.

WasterDave wrote in with this link to information about zero copy sockets on FreeBSD. Some hunting turned up more detailed information about NetBSD and Gigabit networking. Pointers to similar information for OpenBSD are appreciated...

3 of 34 comments (clear)

  1. MB/s != Mbps by scott4000 · · Score: 4

    The units in this case would be in megabits per second, not megabytes per second. 986MB/s would be 7888Mbps, or 7.888Gbps.

    -------
    FreeBSD: The power to serve.

  2. no, Linux does not have zero-copy sockets by cpeterso · · Score: 4

    From the linux-kernel list:

    David Miller: "make no mistake, for transmit we will at some point have a zero copy scheme available. And you can be certain that when it does happen, the end to end latency will not suffer like it does on other systems for the cases where zero copy makes no sense at all."

    Linus Torvalds: "So don't fall into the trap of thinking that zero-copy is always obviously a win. It isn't."

    Ingo Molnar: "zero-copy makes a RL difference only in a small part of those uses. I very much dislike zero-copy-maniac designs which give up just about everything to get nice bandwith numbers."


    --
    cpeterso
  3. Re:Porting to Linux? by Ingo+Molnar · · Score: 5
    As part of TUX i've implemented zero-copy TCP xmit. It turned out to be a minimal change (barring driver changes), less hassle than we initially thought. Obviously we couldnt have gotten those SPECweb99 numbers without zero-copy TCP xmit.

    One important question is, what MTU have they used. If it's the 9000 byte MTU jumbo gigabit frames then these BSD numbers havent got too practical relevance (i can saturate 8 gigabit cards with TUX, ie. 900MB/sec with 9000 byte MTU). If it's the standard 1500 byte MTU then it's nicer. (hm, i just found it, they indeed used 9000 byte MTU...)