If memory serves me right, Matthew Dillon wrote: > I've noticed that -current has much lower TCP performance. I haven't > had time to investigate it but I presume there is some overhead > somewhere that is killing it.
Here's a data point but I'm not sure how useful it is. At the start of December I was using tcpreplay to spew packets from a stored trace into a testbed network as fast as the CPU could go, and I saw: 5-CURRENT (11/19): 9244 pps, 35.6 Mbps 4-STABLE (late November): 21827 pps, 84 Mbps These measurements were on the same machine, which is a 1.7 GHz P4, 512MB RAM, ATA disk. Network interface was a four-port dc-type card. GENERIC kernels. These are "typical" numbers, but fairly repeatable over various trace files I was using. At the time I was more interested in being able to get packets on the network quickly than in why there was a performance difference, so I just plopped 4-STABLE on the machine without doing any other investigation. Bruce.
msg33018/pgp00000.pgp
Description: PGP signature