bad performance using ndis

[John Kennedy]

On Wed, Jan 14, 2004 at 09:26:44AM +0800, Huang wen hui wrote:
> hi,
> I have R40 and try to get Intel 2100 wlan work.
> I download new driver from
> http://www-306.ibm.com/pc/support/site.wss/document.do?lndocid=TPAD-MATRIX
> basicly,wlan works, but performance is bad:
> 
> 64 bytes from 192.168.1.6: icmp_seq=92 ttl=64 time=6.718 ms
> 64 bytes from 192.168.1.6: icmp_seq=88 ttl=64 time=4267.039 ms (DUP!)
> 64 bytes from 192.168.1.6: icmp_seq=91 ttl=64 time=1544.590 ms (DUP!)
> 64 bytes from 192.168.1.6: icmp_seq=89 ttl=64 time=3765.145 ms (DUP!)
> 64 bytes from 192.168.1.6: icmp_seq=87 ttl=64 time=5986.673 ms (DUP!)
> 64 bytes from 192.168.1.6: icmp_seq=93 ttl=64 time=1.980 ms
> 64 bytes from 192.168.1.6: icmp_seq=94 ttl=64 time=6.855 ms
> 64 bytes from 192.168.1.6: icmp_seq=95 ttl=64 time=6.870 ms
> 64 bytes from 192.168.1.6: icmp_seq=96 ttl=64 time=7.209 ms
> 64 bytes from 192.168.1.6: icmp_seq=97 ttl=64 time=2.118 ms
> 64 bytes from 192.168.1.6: icmp_seq=98 ttl=64 time=2.929 ms
> 64 bytes from 192.168.1.6: icmp_seq=96 ttl=64 time=2223.507 ms (DUP!)
> ^C
> --- 192.168.1.6 ping statistics ---
> 99 packets transmitted, 99 packets received, +104 duplicates, 0% packet loss
> round-trip min/avg/max/stddev = 1.896/1928.298/10524.836/2269.834 ms

  That isn't bad performance (unless you consider <= 7ms bad), that is
duplicate packets that are taking a long time to make it back to you.

  Everything that is over 1000 ms is a duplicate.  #96, for example:

	64 bytes from 192.168.1.6: icmp_seq=96 ttl=64 time=7.209 ms
	64 bytes from 192.168.1.6: icmp_seq=96 ttl=64 time=2223.507 ms (DUP!)

  Whatever is causing the duplicates may be causing your performance to
tank as a side-effect, of course.