bad network performance in 2.4.0-test1

• Next message: Andrew Sharp: "Re: Hot pluggable CPUs ( was Linux 2.5 / 2.6 TODO (preliminary) )"
• Previous message: Chris Lattner: "Problems getting user-mode-linux to work with 2.4.0-test1 kernel..."
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

I've looked through /proc/sys/net/ipv4 and I don't see anything terribly
wrong in 2.4.0-test1, yet there seems to be a serious performance
difference.

Both kernels were compiled with equivalent networking options, memory
mapped io, no packet filtering. Identical hardware. 2.2.15 is running
kpiod, but otherwise the process names in ps auxww are identical)

2.2.15:
# time ping -c 100000 -s 3500 -f 127.0.0.1 > /dev/null
2.50user 2.38system 0:04.88elapsed 99%CPU (0avgtext+0avgdata
0maxresident)k
0inputs+0outputs (106major+15minor)pagefaults 0swaps

2.4.0-test1
# time ping -c 100000 -s 3500 -f 127.0.0.1 > /dev/null
1.64user 8.12system 0:09.75elapsed 100%CPU (0avgtext+0avgdata
0maxresident)k
0inputs+0outputs (106major+17minor)pagefaults 0swaps

-

Linux test1 2.2.15 #1 Wed May 31 18:52:44 UTC 2000 i686 unknown
TCP REQUEST/RESPONSE TEST to localhost
Local /Remote
Socket Size Request Resp. Elapsed Trans.
Send Recv Size Size Time Rate
bytes Bytes bytes bytes secs. per sec

65535 65535 1 1 9.99 40920.28
65535 65535

Linux test2 2.4.0-test1 #5 Tue May 30 04:53:02 UTC 2000 i686 unknown
TCP REQUEST/RESPONSE TEST to localhost
Local /Remote
Socket Size Request Resp. Elapsed Trans.
Send Recv Size Size Time Rate
bytes Bytes bytes bytes secs. per sec

65535 65535 1 1 9.99 4361.66
65535 65535

-

Linux test1 2.2.15 #1 Wed May 31 18:52:44 UTC 2000 i686 unknown
TCP STREAM TEST to localhost
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec

 65535 65535 65535 9.99 1048.07

Linux test2 2.4.0-test1 #5 Tue May 30 04:53:02 UTC 2000 i686 unknown
TCP STREAM TEST to localhost
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec

 65535 65535 65535 9.99 732.44

This cannot be good.

Justin

-
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/

• Next message: Andrew Sharp: "Re: Hot pluggable CPUs ( was Linux 2.5 / 2.6 TODO (preliminary) )"
• Previous message: Chris Lattner: "Problems getting user-mode-linux to work with 2.4.0-test1 kernel..."
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

This archive was generated by hypermail 2b29 : Wed Jun 07 2000 - 21:00:17 EST