Messages in this thread | | | Date | Tue, 2 Mar 1999 18:19:23 +0100 (CET) | From | Andrea Arcangeli <> | Subject | Re: [patch] acks delayed and then merged in data-segment sent a bitafter |
| |
On Tue, 2 Mar 1999, Josip Loncaric wrote:
>I did not try Andrea's latest patch yet, but her earlier patch helped.
You are been very right not trying yesterday night one ;)
>This does seem to be a delayed ACK problem with small message sizes >(<100 bytes or so). A carefully thought out fix would really help in >this situation.
My not too much careful thought is that we need to waste some time before sending the ack in order to optimize TCP behavior. But for raw iteractive performance we need to disable this timeout. My latest patch (the one send some minutes ago) does this disabling completly delayed acks if TCP_NODELAY is set (at least this was my intention, I hope to havn't done mistakes, I have a little time to play with these high-tech issues...). So it would be interesting (at least to me) if you could give a try to my last patch and setting TCP_NODELAY.
BTW, cool and very productive report ;). Thanks.
Andrea Arcangeli
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |