Hi Michael
Well the old "default" interface "tlp0" delivered a better =20
throughput.. and that's just plain old 10Mb Ethernet, whereas this =20
one is 100Mb
and when i try to "ftp" from the NetBSD box to a MacOSX box... the =20
thoughput in receiving is pretty much the same....

so.. I _think_ the PCCard do have a problem...(and it does provide =20
much higher throughput with MacOSX)

BTW: the switch in between is 100Mb as well....
and another thing... the NetBSD box did grind to a halt last night =20
with a ton of the following:

rtk0: unable to allocate Rx mbuf
rtk0: unable to allocate Tx mbuf



Den 04/02/2008 kl. 14.22 skrev Michael Lorenz:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hello,
>
> On Feb 4, 2008, at 05:26, Niels S. Eliasen wrote:
>
>> After having upgraded to NetBSD 4.0, from a mix of 3.1 kernel and =20
>> 2.x userland, I now have a trusty PowerBook 3500 running fairly =20
>> smoothly...
>>
>> I do however have a performance problem with the Belkin network =20
>> PCCARD, which is Accton based...
>> (Besides I am using the lower of the two PCCARD slots, as I =20
>> believe it is the only one that is useable...)
>>
>> and from dmesg:
>> rtk0 at cardbus0 function 0: Accton MPX 5030/5038 10/100BaseTX
>> rtk0: Ethernet address 00:30:bd:11:51:85
>> rlphy0 at rtk0 phy 7: Realtek internal PHY
>> rlphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
>>
>>
>> ifconfig rtk0
>> rtk0: flags=3D8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
>>         address: 00:30:bd:11:51:85
>>         media: Ethernet autoselect (100baseTX full-duplex)
>>         status: active
>>         inet x.x.x.3 netmask 0xffffff00 broadcast x.x.x.255
>>         inet6 fe80::230:bdff:fe11:5185%rtk0 prefixlen 64 scopeid 0x3
>>
>>
>> and doing a ftp transfer:
>>
>> ftp> put 10Mtestb.rnd
>> local: 10Mtestb.rnd remote: 10Mtestb.rnd
>> 229 Entering Extended Passive Mode (|||62822|)
>> 150 Opening BINARY mode data connection for '10Mtestb.rnd'.
>> 100% |=20
>> ************************************************************| =20
>> 10240 KB    1.18 MB/s    00:08
>> 226 Transfer complete.
>> 10485760 bytes sent in 00:08 (1.18 MB/s)
>> ftp> get 10Mtestb.rnd
>> local: 10Mtestb.rnd remote: 10Mtestb.rnd
>> 229 Entering Extended Passive Mode (|||62809|)
>> 150 Opening BINARY mode data connection for =20
>> '10Mtestb.rnd' (10485760 bytes).
>> 100% |=20
>> ************************************************************| =20
>> 10240 KB    2.74 MB/s    00:03
>> 226 Transfer complete.
>> 10485760 bytes received in 00:03 (2.73 MB/s)
>>
>> and the ifconfig.rtk0
>>
>> cat /etc/ifconfig.rtk0
>> # x.x.x.3 netmask 0xffffff00 media 100baseTX-FDX
>> x.x.x.3 netmask 0xffffff00
>>
>> (have tried both the 100BaseTX-FDX and (now) with auto.... nothing =20=

>> very different performance wise...)
>>
>> Network load is next to nothing.....
>>
>> I would expect somewhat more throughput.....
>
> Can your disk can keep up with that? Those 2.5" driver used in old =20
> powerbooks aren't exactly fast by today's standard. If you want to =20
> measure pure network throughput use something like ttcp that =20
> doesn't hammer the disk.
>
> have fun
> Michael
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (Darwin)
>
> iQEVAwUBR6cRs8pnzkX8Yg2nAQIrfQf/YU53coRet/1OFWkDEs6A1VhpTBWDrPCw
> 3rl3jI/QixvtS+b2A6W+VGzMrs8WgtkwK3HiEQ7e8SN2JzkuF4fcvgUG9RbLeiVr
> tPAyKovQ+VLQ9PZWy3aykoR7LnmcnKH/Qv9w7K/zZf1j0cNgqfgU2+fgllKdPr8H
> g4mzMw96dZbPgzG9pgNHLavNZfGxjhT3NzzqaM/c6Avpo2Zoua+hgXKnrR0R62J8
> d55PrNxXGiufKmXBq90efQVpW4tEkEIFrn9Yhp+7PH5l7+LXkmMR7lv6A1klKE6g
> dLlxSGx8fAHO0EdFMSgMVvyDFy5+BtI1K2glwfBVVWyIWulcHZCvAA=3D=3D
> =3Db02J
> -----END PGP SIGNATURE-----

mvh/kind regards
Niels S. Eliasen
H=F8rhavevej 1
DK-4250, Fuglebjerg
Tel/Cell: +45 46 32 85 27 +45 21 77 95 90
mailto:nse_rEmOvE_tHiS@delfi-konsult.com