[sandman]

Alright this is pathetic. I'm used to getting over 100KB/sec at home on a cable modem and here I'm getting mostly less than 20KB/sec. This network is blazingly fast for PC users, there has to be SOMETHING I can do: (new ethernet card? change settings? ask my network admin something that will help?)

[Tsilou B.]

I had the same problem with my university network connection. My Windows PC got about 200KB/s while I got 60-70KB/s on my Mac.
I found out that the mtu was the problem. I changed it from 1500 to about 1200, IIRC and suddenly I got > 200KB/s on the Mac.

You could try

sudo ifconfig en0 mtu 1200

check with ifconfig, if it has really changed to 1200, and if it has, test the connection speed. If it's still slow, try other values between 1000 and 1500, and if you have the same problem I had, your download speed should match the PC download speed as soon as you found the correct value. Try many values, don't give up if the first 5 values don't speed it up.

[sandman]

Originally posted by Tsilou B.:
I had the same problem with my university network connection. My Windows PC got about 200KB/s while I got 60-70KB/s on my Mac.
I found out that the mtu was the problem. I changed it from 1500 to about 1200, IIRC and suddenly I got > 200KB/s on the Mac.

You could try

sudo ifconfig en0 mtu 1200

check with ifconfig, if it has really changed to 1200, and if it has, test the connection speed. If it's still slow, try other values between 1000 and 1500, and if you have the same problem I had, your download speed should match the PC download speed as soon as you found the correct value. Try many values, don't give up if the first 5 values don't speed it up.

Thanks for the tip, I will give this a try. I just found out that my Mac is getting 181 Kbps while the roommate's PC is getting 1121 Kbps, pretty significant difference.

[sandman]

I tried mtu values of 1050, 1100, 1150..... 1450, 1500 and the default 1500 gave me the best speed, albiet the same 180 Kbps while the PC gets 1120 Kbps. This is really discouraging, I can't believe I need a PC to use broadband properly.

[[APi]TheMan]

I download at 1.3MB/sec from Microsoft's servers... should I bother tweaking? ! I think sandman should tell us which file he's trying to download so we can all go clock ourselves too...

My network card is running full-duplex at 100BaseTX, by the way. I've tried to set my Pismo's ethernet card to full-duplex and it never stuck, some cards just can't do it.

What kind of machine are you running, sandman?

[Tsilou B.]

You could try to download one of the Gentoo Linux live CDs. They should both run without problems on the 17" PowerBook.

http://gentoo.oregonstate.edu/releas....4_rc7/livecd/

You can boot from the LiveCD (it doesn't change the hard drive) and run

dhcpcd eth0

to setup the Ethernet network. Measure the speed, e.g. with

curl -o /dev/null http://somefastserver/somefile

If you get higher speeds in Linux, you could be sure that it's no hardware failure, but only some configuration problem.

[sandman]

Originally posted by Tsilou B.:
You could try to download one of the Gentoo Linux live CDs. They should both run without problems on the 17" PowerBook.

http://gentoo.oregonstate.edu/releas....4_rc7/livecd/

You can boot from the LiveCD (it doesn't change the hard drive) and run

dhcpcd eth0

to setup the Ethernet network. Measure the speed, e.g. with

curl -o /dev/null http://somefastserver/somefile

If you get higher speeds in Linux, you could be sure that it's no hardware failure, but only some configuration problem.

I can only get about 20KB/sec downloading that file so I don't think I have the patience for that. APi check my sig for my config. Any other options?

[sandman]

APi, go to http://bandwidthplace.com/speedtest/ to test your bandwidth. On this site I get roughly 390 kilobits per sec. My roommate gets a staggering 2.4 megabits per sec.

[DBvader]

im no longer at school, but when i get back to UCSD, i will definetely give the suggestions a try.

" good point - try timing identical downloads on your mac and your roommate's pc to see if they really take the same amount of time. windows tends to exaggerate."

no, it was downloading at that speed, becuase the downloads would finish in time, etc... and as a windows user before my Dual 867, im fairly confident that windows doesnt exxagerate. its menumeters ( i use it, btw) that is actually a bit off

at home (my current situation) i have my PC and mac connected to my cable modem. the mac routes (using jags internet sharing) a connection to the PC. i havent noticed a significant difference in bandwidth (then again, the bandwidth is significantly more limited on a cable modem)

i really wonder what the issue here is, and if we can maybe resolve it, that would be great.

"A friend of mine works in the IT department at the regional schoolboard, and a year ago he brought up in a discussion that Apple has never been able to provide the schoolboard with an answer as to why the network (broadband) is slower on Apple hardware... He had said that in most cases it was almost half as slow."

thats seriously pretty fascinating, and its consistent with what i have seen with my and my roomates computer.

[Graymalkin]

This thread inspired me to clock my network performance. I used ifconfig to change media and mediaopt values and watched MenuMeters for my throughput. I transfered a 123MB file over a 100BaseT switch from a Windows share on the network.

10BaseT @ half-duplex: reported 60-80KB/s Tx throughput and 4-5KB/s Rx throughput, the transfer took less than five minutes.

10BaseT @ full-duplex: reported 4-7KB/s Tx throughput and 800-900KB/s Rx throughput, the transfer took about a minute minute.

100BaseT @ half-duplex: reported 60-80KB/s Tx throughput and 5-7KB/s Rx throughput, transfer took less than a minute.

100BaseT @ full-duplex: reported 5-8KB/s Tx throughput and 200-400KB/s Rx throughput, transfer took about five minutes.

My PC is set to force a 100BaseTX full-duplex connection. Apparently my 12" Powerbook's ethernet card is not a big fan of either operating in full duplex mode at 100Mbps or my switch is causing some sort of weird problem. Being as several models of Macs refuse to even run full-duplex mode I'm inclined to believe my Powerbook's ethernet card is the culprit. The best throughput was at 100BaseTX running half-duplex. Take into consideration however that these speeds are over a Samba/CIFS share and not over the internet, my cable modem is capped at 256Kb/s ? 32KB/s.

I've also begun to wonder how accurate the bandwidth reporting is on both OSX and Windows. I don't recall seeing a mention of what programs you've been using to download, nor the programs your roomate has been using. Depending on the application a bandwidth throughput reading can vary wildly. If you measure your bandwidth at sub-second intervals your measurements are going to look very strange, increasing the interval gives you a much steadier bandwidth reading. Is your roomate updating the bandwidth reading at sub-second intervals and calling out readings or is the bandwidth reading steady? Also is the actual file you're downloading taking measurably longer to download than on your roomate's PC? If your app is saying you're getting 1K/s yet the 100MB file is done downloading in twenty seconds you've got to wonder if you're even getting an accurate bandwidth reading.

Make sure you use large files as your test download and not tiny GIF images or something. A file that can fit into a handful of transmission units is not an accurate way to find your actual throughput. Files taking less than a second to transfer make throughput meters extrapolate outrageous throughput values.

[wadesworld]

sandman,

When you get back, run the following commands and post the results here:

netstat -id

netstat -s

Wade

[sandman]

Originally posted by wadesworld:
sandman,

When you get back, run the following commands and post the results here:

netstat -id

netstat -s

Wade

Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll Drop
lo0 16384 <Link#1> 7475 0 7475 0 0 0
lo0 16384 localhost ::1 7475 - 7475 - - -
lo0 16384 fe80:1::1 fe80:1::1 7475 - 7475 - - -
lo0 16384 127 localhost 7475 - 7475 - - -
gif0* 1280 <Link#2> 0 0 0 0 0 0
stf0* 1280 <Link#3> 0 0 0 0 0 0
en1 1500 <Link#4> 00:03:93:ec:ca:89 0 0 6 0 0 0
en1 1500 fe80:4::203 fe80:4::203:93ff: 0 - 6 - - -
en0 1500 <Link#5> 00:0a:95:a0:e6:06 2135 0 659 0 0 0
en0 1500 fe80:5::20a fe80:5::20a:95ff: 2135 - 659 - - -
en0 1500 128.227.190/2 x190-134.dhnet. 2135 - 659 - - -
en0 1500 (16)00:00:ff:18:98 2135 0 659 0 0 0

[sandman]

tcp:
8034 packets sent
3707 data packets (969542 bytes)
0 data packets (0 bytes) retransmitted
0 resends initiated by MTU discovery
4158 ack-only packets (0 delayed)
0 URG only packets
0 window probe packets
53 window update packets
116 control packets
7934 packets received
3819 acks (for 963786 bytes)
52 duplicate acks
0 acks for unsent data
3906 packets (1183660 bytes) received in-sequence
12 completely duplicate packets (13268 bytes)
0 old duplicate packets
0 packets with some dup. data (0 bytes duped)
141 out-of-order packets (188020 bytes)
0 packets (0 bytes) of data after window
0 window probes
20 window update packets
1 packet received after close
0 discarded for bad checksums
0 discarded for bad header offset fields
0 discarded because packet too short
49 connection requests
28 connection accepts
0 bad connection attempts
0 listen queue overflows
68 connections established (including accepts)
74 connections closed (including 8 drops)
6 connections updated cached RTT on close
6 connections updated cached RTT variance on close
0 connections updated cached ssthresh on close
4 embryonic connections dropped
3819 segments updated rtt (of 3818 attempts)
25 retransmit timeouts
0 connections dropped by rexmit timeout
0 persist timeouts
0 connections dropped by persist timeout
6 keepalive timeouts
0 keepalive probes sent
6 connections dropped by keepalive
4 correct ACK header predictions
627 correct data packet header predictions
udp:
1185 datagrams received
0 with incomplete header
0 with bad data length field
0 with bad checksum
1 dropped due to no socket
301 broadcast/multicast datagrams dropped due to no socket
0 dropped due to full socket buffers
0 not for hashed pcb
883 delivered
121 datagrams output
ip:
9341 total packets received
0 bad header checksums
0 with size smaller than minimum
0 with data size < data length
0 with ip length > max ip packet size
0 with header length < data size
0 with data length < header length
0 with bad options
0 with incorrect version number
0 fragments received
0 fragments dropped (dup or out of space)
0 fragments dropped after timeout
0 packets reassembled ok
9121 packets for this host
39 packets for unknown/unsupported protocol
0 packets forwarded (0 packets fast forwarded)
133 packets not forwardable
0 packets received for unknown multicast group
0 redirects sent
8184 packets sent from this host
0 packets sent with fabricated ip header
0 output packets dropped due to no bufs, etc.
0 output packets discarded due to no route
0 output datagrams fragmented
0 fragments created
0 datagrams that can't be fragmented
0 tunneling packets that can't find gif
0 datagrams with bad address in header
icmp:
1 call to icmp_error
0 errors not generated 'cuz old message was icmp
Output histogram:
echo reply: 2
destination unreachable: 1
0 messages with bad code fields
0 messages < minimum length
0 bad checksums
0 messages with bad length
0 multicast echo requests ignored
0 multicast timestamp requests ignored
Input histogram:
destination unreachable: 11
echo: 2
2 message responses generated
ICMP address mask responses are disabled
igmp:
28 messages received
0 messages received with too few bytes
0 messages received with bad checksum
13 membership queries received
0 membership queries received with invalid field(s)
15 membership reports received
0 membership reports received with invalid field(s)
15 membership reports received for groups to which we belong
14 membership reports sent
ipsec:
0 inbound packets processed successfully
0 inbound packets violated process security policy
0 inbound packets with no SA available
0 invalid inbound packets
0 inbound packets failed due to insufficient memory
0 inbound packets failed getting SPI
0 inbound packets failed on AH replay check
0 inbound packets failed on ESP replay check
0 inbound packets considered authentic
0 inbound packets failed on authentication
0 outbound packets processed successfully
0 outbound packets violated process security policy
0 outbound packets with no SA available
0 invalid outbound packets
0 outbound packets failed due to insufficient memory
0 outbound packets with no route
ip6:
0 total packets received
0 with size smaller than minimum
0 with data size < data length
0 with bad options
0 with incorrect version number
0 fragments received
0 fragments dropped (dup or out of space)
0 fragments dropped after timeout
0 fragments that exceeded limit
0 packets reassembled ok
0 packets for this host
0 packets forwarded
0 packets not forwardable
0 redirects sent
11 packets sent from this host
0 packets sent with fabricated ip header
0 output packets dropped due to no bufs, etc.
0 output packets discarded due to no route
0 output datagrams fragmented
0 fragments created
0 datagrams that can't be fragmented
0 packets that violated scope rules
0 multicast packets which we don't join
Mbuf statistics:
0 one mbuf
0 one ext mbuf
0 two or more ext mbuf
0 packets whose headers are not continuous
0 tunneling packets that can't find gif
0 packets discarded due to too may headers
0 failures of source address selection
0 forward cache hit
0 forward cache miss
icmp6:
0 calls to icmp_error
0 errors not generated because old message was icmp error or so
0 errors not generated because rate limitation
Output histogram:
multicast listener report: 9
neighbor solicitation: 2
0 messages with bad code fields
0 messages < minimum length
0 bad checksums
0 messages with bad length
Histogram of error messages to be generated:
0 no route
0 administratively prohibited
0 beyond scope
0 address unreachable
0 port unreachable
0 packet too big
0 time exceed transit
0 time exceed reassembly
0 erroneous header field
0 unrecognized next header
0 unrecognized option
0 redirect
0 unknown
0 message responses generated
0 messages with too many ND options
0 messages with bad ND options
0 bad neighbor solicitation messages
0 bad neighbor advertisement messages
0 bad router solicitation messages
0 bad router advertisement messages
0 bad redirect messages
0 path MTU changes
ipsec6:
0 inbound packets processed successfully
0 inbound packets violated process security policy
0 inbound packets with no SA available
0 invalid inbound packets
0 inbound packets failed due to insufficient memory
0 inbound packets failed getting SPI
0 inbound packets failed on AH replay check
0 inbound packets failed on ESP replay check
0 inbound packets considered authentic
0 inbound packets failed on authentication
0 outbound packets processed successfully
0 outbound packets violated process security policy
0 outbound packets with no SA available
0 invalid outbound packets
0 outbound packets failed due to insufficient memory
0 outbound packets with no route
pfkey:
0 requests sent to userland
0 bytes sent to userland
0 messages with invalid length field
0 messages with invalid version field
0 messages with invalid message type field
0 messages too short
0 messages with memory allocation failure
0 messages with duplicate extension
0 messages with invalid extension type
0 messages with invalid sa type
0 messages with invalid address extension
0 requests sent from userland
0 bytes sent from userland
0 messages toward single socket
0 messages toward all sockets
0 messages toward registered sockets
0 messages with memory allocation failure

[Spliffdaddy]

I tried that download speed link up there ^ and got 1.7Mb/sec on WinXP - and 970Kb/sec on WinME. Same machine, dual-booted...same connection (direct to cablemodem - connection autosets to 10Mb).

Almost twice as fast using XP. wtf? It looked real to me. One took 10 seconds and one took 5.

For those folks suggesting that their Macs are at the mercy of their hardware (NIC) - I suggest that's untrue. Apparently the OS can have a big impact on the speed of your network connection.

[Tsilou B.]

Your connection doesn't seem to have problems. Are you sure the configuration is correct on your Mac? What's the subnet mask?
Is it the same one as on the fast PC?

Update: I've just read one of your earlier posts. So your Subnet mask is 255.255.255.0, check the PC subnet mask by running "ipconfig" on the Windows command line.
(If the PC still runs Windows 98/ME, use "winipcfg" instead of "ipconfig".)

[Graymalkin]

Originally posted by Spliffdaddy:
I tried that download speed link up there ^ and got 1.7Mb/sec on WinXP - and 970Kb/sec on WinME. Same machine, dual-booted...same connection (direct to cablemodem - connection autosets to 10Mb).

Almost twice as fast using XP. wtf? It looked real to me. One took 10 seconds and one took 5.

For those folks suggesting that their Macs are at the mercy of their hardware (NIC) - I suggest that's untrue. Apparently the OS can have a big impact on the speed of your network connection.

If you're running the download off of a cable modem it is likely one system is hitting the cable provider's transparent proxy cache and the second time you download it isn't. One serious problem with web based bandwidth meters is the fact a lot of cable and DSL head ends run transparent proxies to reduce the load on the extneral network connection. Your cable link is say 1Mb up and down but only so far as the cable head end, from there you get whatever fraction of the T3 connection there is available. The providers run the transparent caches on the head end so less data needs to come over the external pipe repeatedly.

[sandman]

Originally posted by Tsilou B.:
Your connection doesn't seem to have problems. Are you sure the configuration is correct on your Mac? What's the subnet mask?
Is it the same one as on the fast PC?

Update: I've just read one of your earlier posts. So your Subnet mask is 255.255.255.0, check the PC subnet mask by running "ipconfig" on the Windows command line.
(If the PC still runs Windows 98/ME, use "winipcfg" instead of "ipconfig".)

The PC's subnet mask is the same 255.255.255.0 as mine. And to try to give more details I'm using Safair for downloads/browsing. He's using IE.

[Tsilou B.]

Ok, two more things:

1.) ping some server that's not too far away, e.g. your university's webserver.

ping -c 4 internetaddress

and on the Windows PC it's probably:

ping -n 4 internetaddress

Post the ping output from your Mac and the PC.

2.) Do a traceroute to the same server:

traceroute internetaddress

on the Windows PC it's:

tracert internetaddress
(who said windows supported long file names? )

Post that output, too.

[Spliffdaddy]

Originally posted by Graymalkin:
If you're running the download off of a cable modem it is likely one system is hitting the cable provider's transparent proxy cache and the second time you download it isn't. One serious problem with web based bandwidth meters is the fact a lot of cable and DSL head ends run transparent proxies to reduce the load on the extneral network connection. Your cable link is say 1Mb up and down but only so far as the cable head end, from there you get whatever fraction of the T3 connection there is available. The providers run the transparent caches on the head end so less data needs to come over the external pipe repeatedly.

No, the results are repeatable and predictible.

I tried XP first and got 1.7Mb/sec.

rebooted into WinME and got about half the bandwidth I had in WinXP.

Tried it several times with the same result.

I've always heard that NT-based Windows versions were faster on a network than 9x versions - I just didn't know they were almost twice as fast.

There might be some tweaks I could try with WinME, but I don't think I should have to manually adjust those network settings.

good luck on the Mac end of things. you are not alone.

[wadesworld]

Sandman,

It's starting to look like you're down to the point of needing someone to take a packet trace of both machines in action.

There's nothing in your netstat output that shows major network problems.

It sounds like the Ethernet switch and your machine don't like each other very much.

I'd see if you can get the administrator of the switch to look at the stats on your port and see if it sees any problems.

Wade

[sandman]

Originally posted by Tsilou B.:
Ok, two more things:

1.) ping some server that's not too far away, e.g. your university's webserver.

ping -c 4 internetaddress

and on the Windows PC it's probably:

ping -n 4 internetaddress

Post the ping output from your Mac and the PC.

2.) Do a traceroute to the same server:

traceroute internetaddress

on the Windows PC it's:

tracert internetaddress
(who said windows supported long file names? )

Post that output, too.

Where would I find my universities webserver address?

[wadesworld]

Where would I find my universities webserver address?

Just ping it by name:

ping -c www.myuniversity.edu

Wade

[Giv]

There has been a suggestion to reduce the MTU (a MTU set too high can cause packet fragmentation, although path MTU discovery should avoid that).

What hasn't been mentionned is that XP may be using jumbo packets. I don't know if 10.2 supports jumbo packets, however Panther will.

I don't know how to check that theory unfortunately.

[decursive]

have you tried BroadbandOptimizer?

http://www.enigmarelle.com/sw/BroadbandOptimizer/

[SMacTech]

Originally posted by Graymalkin:
This thread inspired me to clock my network performance. I used ifconfig to change media and mediaopt values and watched MenuMeters for my throughput. I transfered a 123MB file over a 100BaseT switch from a Windows share on the network.

10BaseT @ half-duplex: reported 60-80KB/s Tx throughput and 4-5KB/s Rx throughput, the transfer took less than five minutes.

10BaseT @ full-duplex: reported 4-7KB/s Tx throughput and 800-900KB/s Rx throughput, the transfer took about a minute minute.

100BaseT @ half-duplex: reported 60-80KB/s Tx throughput and 5-7KB/s Rx throughput, transfer took less than a minute.

100BaseT @ full-duplex: reported 5-8KB/s Tx throughput and 200-400KB/s Rx throughput, transfer took about five minutes.


Make sure you use large files as your test download and not tiny GIF images or something.

Something is quite wrong with your setup to take this long transferring a 100mB file over 100bT full/duplex.
Definitely do the test with a single large file. On the Win2k domain I admin, it took approximately 12 seconds to transfer 100MB over 100bt FD from my 450 cube using OS X 10.2.6. About 10 seconds on my P4 2.6g Dell Dimension.
When PCs and Macs are properly configured, the PC will have a slight edge, as the Mac is going to a non-native filesystem and has an additional layer in the network stack. Going to Win2k and using SFM, for example, or 3rd party AFP/IP server. SMB/CIFS support in OS X is still buggy and not too optimized at this time.
The biggest single problem is auto-negotiation and configuration of switches and NICs. Also having something as simple as Tree spanning turned on in a switch can diminish performance considerably on a Mac.
I would also never overlook a bad Cat-5 cable as well. It may seem to work, but it can still be bad.

[sandman]

Well Panther speeds up my connection with the same ifconfig en0 settings so it has to be something else causing the slowdown. (Panther gives me 600 kbps, still not NEAR what the network is giving PC users.)

[sandman]

I sent someone in the tech dept. an email regarding my problem and this is what they said: "What OS are you running? It's important to set the ethernet card speed
to 10MB (not auto or 100MB) and to set the duplex to full (not auto)."

[chabig]

Two comments on the off-topic "meaning of MAC" debate:

1. A google search for "media access control" yields 11,000 matches, while a google search for "machine address code" yields 70.

2. It's called a MAC address, which make sense logically if MAC stands for media access control, since it's a protocol address. On the other hand "machine address code address" makes no sense.

Chris

[sandman]

Ok I finally got the ethernet card to run full-duplex, and then I couldn't connect, so that is clearly not the problem, even though the network tech said it had to be full-duplex.

[Charles Reader]

I just tried the bandwith test from bandwithplace.com and it said that my bandwith was 3.4 megabits per second. I consider this to be much higher than it ought to be given my network connection is through a 802.11b network connection.

So I went over to 2wire.com and they said my network connection is 3.12 mbs. Who am I to argue in this case?

[wadesworld]

Ok I finally got the ethernet card to run full-duplex, and then I couldn't connect, so that is clearly not the problem, even though the network tech said it had to be full-duplex.

Um, your assumption is incorrect.

If the tech said it had to be full duplex and it's in half-duplex, that would definitely explain the poor performance you're seeing.

As for why it didn't connect, more troubleshooting would be required.

Wade

[sandman]

Originally posted by wadesworld:
Um, your assumption is incorrect.

If the tech said it had to be full duplex and it's in half-duplex, that would definitely explain the poor performance you're seeing.

As for why it didn't connect, more troubleshooting would be required.

Wade

I agree. At full-duplex I can't get a web connection but packets are being received from the server as seen through the Network Utility.

I'm in contact with the web administrator and we are trying to figure this out. Any ideas on other settings besides speed and duplex? Those are both at the correct setting. Now I need to establish a connection.

[videian28]

i used bandwidth optimizer and went and did some tests, I got exactly the same rate as my xp machine did

worth a shot