promise, this is last one from me on this issue...:-)

the parameters which control the actual window sizes on XP SP2 are limited by:

HKLM\system\currentcontrolset\services\afd\parameters\DefaultReceiveWindow
HKLM\system\currentcontrolset\services\afd\parameters\DefaultSendWindow

gotten from:

http://technet2.microsoft.com/WindowsServer/en/Library/94d21089-411b-4bce-a823-49a77a46e7661033.mspx

-Sunil


Sunil <funtoos@yahoo.com> wrote:
the reason why physical interface gives higher thruput with 64K compared to loopback is that the loopback has rtt of 15ms when using pcap-bridge while the physical one has an rtt of 0.45ms. If the loopback is using tuntap driver instead of winpcap in a bridged setup, the rtt is down to 0.45ms.

so its the additional delay (introduced by winpcap??) which is bringing down the thruput for loopback. High b/w and high rtt is a disaster for thruput if the window size is small. A higher TCP window size compensates for this delay and corrects the thruput. Only if I could force xp to accept that...:(

-Sunil

Sunil <funtoos@yahoo.com> wrote:
I have read the wiki a couple of times in full. I don't see any reason why, if I am able to configure two interfaces to make colinux and windows talk to each other and net, when people there have problem with even one, would it be network driver issue.

as I have proven from the iperf data the problem lies in the part where TCP options negotiations happen between windows and colinux. When windows is initializing the connection, the negotiation is for a high window size, but when colinux is initializing the connection, the window size is not optimal but the default. Either colinux is not asking the right questions or windows is not responding with right answers. I have a feeling that it is the latter, and could be a bug with auto-tuning on windows side.

With proper window sizes, both upload (550 Mbps) and download (350 Mbps) speeds are very much in the ballpark, at least not order of magnitude different.

Can someone with XP SP1 try setting the TcpWindowSize (255500) and Tcp1323Opts (1) and confirm if the thruput improves on loopback?

We are talking about 40Mbytes per second transfer rate which is close to many HDs of today. That is exciting, isn't it? If only XP will let us...:(



Matt Dockerty <matthew.dockerty@186k.co.uk> wrote:
In my experience, my Intel card performs a lot slower than an a loopback driver. From what I've read here it's largely dependant on your NIC and it's driver software. Have you read the Wiki, there are some interesting points on networking there?


From: colinux-users-admin@lists.sourceforge.net [mailto:colinux-users-admin@lists.sourceforge.net] On Behalf Of Sunil
Sent: 30 March 2006 03:20
To: Sunil; Matt Dockerty; 'colinux'
Subject: RE: [coLinux-users] loopback and some question marks on speed

partial good news. The throughput is related to window size in a wierd way for colinux and loopback connection.
1. It doesn't matter how high I set the TCP window size on colinux, 192.168.0.0 network is much faster (220 vs 30) than 10.3.75.0 loopback network.
2. High values of TCP windows size on windows XP side give the boost to 10.3.75.0 network beyond 192.168.0.0 network. This also requires a high window size on colinux, which is understandable.
 
C:\Program Files\IPerf>iperf -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 64.2 KByte (default)
------------------------------------------------------------
[1920] local 10.3.75.1 port 5001 connected with 10.3.75.2 port 46177
[ ID] Interval       Transfer     Bandwidth
[1920]  0.0-10.1 sec  39.7 MBytes  31.5 Mbits/sec
[1920] local 192.168.0.3 port 5001 connected with 192.168.0.11 port 41019
[ ID] Interval       Transfer     Bandwidth
[1920]  0.0-10.0 sec   272 MBytes   217 Mbits/sec
close failed: Successful WSAStartup not yet performed.
C:\Program Files\IPerf>iperf -s -w 256K
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size:  256 KByte
------------------------------------------------------------
[1920] local 10.3.75.1 port 5001 connected with 10.3.75.2 port 46174
[ ID] Interval       Transfer     Bandwidth
[1920]  0.0-10.0 sec   157 MBytes   126 Mbits/sec
[1920] local 192.168.0.3 port 5001 connected with 192.168.0.11 port 41016
[ ID] Interval       Transfer     Bandwidth
[1920]  0.0-10.0 sec   278 MBytes   223 Mbits/sec
C:\Program Files\IPerf>iperf -s -w 640K
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size:  640 KByte
------------------------------------------------------------
[1920] local 10.3.75.1 port 5001 connected with 10.3.75.2 port 46179
[ ID] Interval       Transfer     Bandwidth
[1920]  0.0-10.0 sec   356 MBytes   285 Mbits/sec
[1920] local 192.168.0.3 port 5001 connected with 192.168.0.11 port 41021
[ ID] Interval       Transfer     Bandwidth
[1920]  0.0-10.0 sec   278 MBytes   222 Mbits/sec
close failed: Successful WSAStartup not yet performed.
 
At 1280K, it increased further but iperf died abnormally on ctrl-c...:)
 
I think this is interesting. Why would increasing the tcp window size on xp give boost to one but not the other interface?
 
Thanks,
-Sunil

Sunil <funtoos@yahoo.com> wrote:
How can the direction make order of magnitude difference? I notice in the data that the windows uses TCP window size of 64KB for both client and server, while colinux uses 16KB for client and 85KB for server iperf. Are these configured in the kernel?

look at these:

root@co-amd-pc ~
$ iperf -c 10.3.75.1
------------------------------------------------------------
Client connecting to 10.3.75.1, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[  3] local 10.3.75.2 port 51361 connected with 10.3.75.1 port 5001
[  3]  0.0-10.0 sec  39.1 MBytes  32.8 Mbits/sec

root@co-amd-pc ~
$ iperf -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 85.3 KByte (default)
------------------------------------------------------------
[  4] local 10.3.75.2 port 5001 connected with 10.3.75.1 port 3215
[  4]  0.0-10.0 sec    687 MBytes    576 Mbits/sec

Thanks,
Sunil


Matt Dockerty <matthew.dockerty@186k.co.uk> wrote:
I should have mentioned YMMV. Networking doesn't seem like an exact science
in Colinux at this alpha stage and I only got these levels of performance
through blind trial and error.

My setup is a MS Loopback Adapter on a 192.168 subnet and a TAP adapter on
same. My physical LAN is on a 10.10 subnet and Internet Connection Sharing
is enabled. My network config in Colinux is simply this:



Like Holger says 10Mbps means nothing since the loopback driver is only
capped by the limitations of the software. Here are the results I'm getting:

colinux ~ # iperf -c 192.168.0.1
------------------------------------------------------------
Client connecting to 192.168.0.1, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.0.100 port 1767 connected with 192.168.0.1 port 5001
[ 3] 0.0-10.7 sec 69.1 MBytes 54.0 Mbits/sec
colinux ~ # iperf -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 85.3 KByte (default)
------------------------------------------------------------
[ 4] local 192.168.0.100 port 5001 connected with 192.168.0.1 port 2146
[ 4] 0.0-10.0 sec 125 MBytes 105 Mbits/sec

--
Matt

> -----Original Message-----
> From: colinux-users-admin@lists.sourceforge.net
> [mailto:colinux-users-admin@lists.sourceforge.net] On Behalf
> Of Holger Krull
> Sent: 29 March 2006 07:59
> To: colinux
> Subject: Re: [coLinux-users] loopback and some question marks on speed
>
> Sunil schrieb:
>
> > way? Why are loopback adapters shown as 10Mbps? Is there a
> way to change
> > that?
>
> That is just a label without meaning.
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking
> scripting language
> that extends applications into web and mobile media. Attend
> the live webcast
> and join the prime developer group breaking into this new
> coding territory!
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&
dat=121642
> _______________________________________________
> coLinux-users mailing list
> coLinux-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/colinux-users
>



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
coLinux-users mailing list
coLinux-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/colinux-users


New Yahoo! Messenger with Voice. Call regular phones from your PC for low, low rates.


New Yahoo! Messenger with Voice. Call regular phones from your PC and save big.


New Yahoo! Messenger with Voice. Call regular phones from your PC and save big.


Blab-away for as little as 1¢/min. Make PC-to-Phone Calls using Yahoo! Messenger with Voice.


New Yahoo! Messenger with Voice. Call regular phones from your PC for low, low rates.