SG Broadband Community

I've run the tcp analyzer and tried to change rwin value to 128480 but each time I rerun the test it shows it at 8192. I'm running a cable modem into a router and then to 3 pcs. I'm showing 600kps downstream usually. Thanks for any help. Gary

Did you reboot after you made the changes?

What OS? Are you seeing a Windows Scaling value? Can you Cut&Paste your Analyzer output?

Here's my info. I'm running win98se on a 750 duron w/576mb ram and a cable modem from AT&T.
Tue Aug 27 22:18:25 2002.



SpeedGuide.net TCP/IP Analyzer
TCP properties for IP = 12.255.129.90 (12-255-129-90.client.attbi.com)
Browser/OS = Mozilla/4.0 (compatible; MSIE 6.0; Windows 98)
Note: Read the FAQ if the above is not your IP address.

TCP options string = 020405b401010402

MTU = 1500
MTU is fully optimized for broadband.

MSS = 1460
Maximum useful data in each packet = 1460, which equals MSS.

Default Receive Window (RWIN) = 8192
RWIN Scaling (RFC1323) = 0 bits
Unscaled Receive Window = 8192
For optimum performance, consider changing RWIN to a multiple of MSS.
Other values for RWIN that might work well with your current MTU/MSS:
513920 (MSS x 44 * scale factor of 8)
256960 (MSS x 44 * scale factor of 4)
128480 (MSS x 44 * scale factor of 2)
64240 (MSS x 44)

bandwidth * delay product:
Your RcvWindow limits you to: 327.68 kbps (40.96 KBytes/s) @ 200ms
Your RcvWindow limits you to: 131.072 kbps (16.384 KBytes/s) @ 500ms
Consider increasing your RWIN value to optimize TCP/IP for broadband.

MTU Discovery (RFC1191) = ON

Time to live left = 49 hops
TTL value is ok.

Timestamps (RFC1323) = OFF

Selective Acknowledgements (RFC2018) = ON

IP type of service field (RFC1349)= 00000000




1 connections tested since 03.10.2001.
Analyzer version: 1.01 - updated 05.16.2002.
Thanks, Gary

What do you have GlobalMaxTcpWindowSize set to? In Windows 98SE, I have it set to the same value as DefaultRcvWindow. I use 256960, and the correct value shows up during the TCP/IP Analyzer test.

Frank's Windows 95/98 Tips

Is scaling on or off?

Try the DSLR Tweak Test and see what it says your RWIN is, sometimes the TCP/IP Analyzer is incorrect.

Just to answer a few questions... Yes, windows scaling is on, and the dslreports tweak test also reports the RWIN as 8192. The tweaks test recommends to: turn on selective acks, increase ttl to 64, choose rwin between 14600 and 37960, and download and use drtcp(already have installed). Let me know which items to change or what to try next. Thanks, Gary

8192 is the default RWIN for 98, by chance do you have GlobalMaxTcpWindoSize set to 8192? Try downloading Cablenut from my signature and try the following settings (Cablenut will override any previous settings).

BcastNameQueryCount - 1
BcastQueryTimeout - 100
BSDUrgent - 1
CacheTimeout - 600000
DefaultRecvWindow - 37960 or 52560 or 64240
DefaultTTL - 64
EnableDNS - 0
GlobalMaxTcpWindoSize - 37960 or 52560 or 64240
KeepAliveInterval - 500
KeepAliveTime - 14400000
Lanabase - 0
LocalCopyMade - 1
MaxConnections - 99
MaxConnectRetries - 5
MaxDataRetries - 64
NameTableSize - 255
NameSrvQueryTimeout - 100
PMTUBlackHoleDetect - 0
PMTUDiscovery - 1
RoutingBufSize - 146432
RoutingPackets - 100
SackOpts - 1
SessionKeepAlive - 7200
SessionTableSize - 255
Size/Small/Medium/Large - 3
Tcp1323Opts - 0
TcpTimedWaitDelay - 30
MaxDupAcks - 2
DefaultTOS - 92
IGMPLevel - 2
MaxConnectionsPer1_0Server - 20
MaxConnectionsPerServer - 10

Sounds like an out-of-date vtcp.386 file...

You cannot use Windows Scaling in Win98 until you upgrade vtcp.386.

What type of router are U running behind?

Default Receive Window (RWIN) = 8192
RWIN Scaling (RFC1323) = 0 bits
Unscaled Receive Window = 8192


According to that, scaling is off. I would try the updated vtcp.386 file.

The VTCP.386 file for Win98 is downloadable from here

Please take a tweak analyzer test from Here and post the results to see what it says.

He already used that test and got the same results. It reports the same as SG. I'm sure once he updates VTCP.386 file, he'll be all set. Also change Tcp1323Opts to = 1, and repost your analyzer results. Windows will not use scaling if this is set to 0. Steele, you told him to set it to 0, which does not use scaling. Set it to 1, and any RWIN you use, large or small, will be enabled. This setting is not just for RWINS over 655**

NOTE: The valid range is 0,1,2 or 3 where:
0 (disable RFC 1323 options)
1 (window scale enabled only)
2 (timestamps enabled only)
3 (both options enabled)

I just experimented with my 98 machine, and with scaling off, my RWIN does not show up as entered, when I use the analyzer. It actually shows as 65535, instead of 372300, which is what it is set at.

Your proxy is off?

Comcast doesn't use proxies

Scaling should be ON if number is above 65535 and you may need this patch on Win SE

http://www.speedguide.net/files/vtcp386/236926USA8.EXE

I know LObo, I have the patch. For some reason though, when scaling is off, the analyzer does not report my RWIN correctly. I noticed this years ago, when I first installed cablenut and used the "98 cable normal" patch, which does not use scaling. Then, I would manually turn scaling on, and then my RWIN would show up as 49640, which what was entered.

I know this doesn't make logical sense, but I bet if wb7sdxgary turned scaling on, his RWIN would report correctly. ( as long as he has the patch in of course)

Scaling should be ON if number is above 65535

Like I said earlier, I am fully aware of this. What I am saying is the analyzer does not report my rwin, that is over 65535, correctly on my 98 machine, which is set to 372300, with scaling off. I also understand scaling must be on for large recieve window support.
It reports my RWIN to be 65535, even though the patch is in, but scaling is off. As soon as I turn scaling on, the correctly entered RWIn shows up in the analyzer. It sounds to me, like a analyzer problem, not a windows problem. Also, no matter what my RWIN is set too, with scaling off it reports it to be 65535. Even if it is set less then 65535. So if I set it to 32670, with scaling off, the analyzer shows it to be 65535

The Analyzer is reporting correctly. It is not a Windows or an Analyzer problem. Everything is working as it should.

If you TURN OFF scaling, then it is NOT possible for your ReceiveWINdow to be over 65535. Sure, you can set the number higher in your registry (using regedit, the Optimizer, or Cablenut), but that number is MEANINGLESS unless you turn on scaling.

If scaling is OFF, your RWIN cannot REALLY be over 65535 -- and the Analyzer is correctly telling you this fact.
_____

If you are setting your RWIN to under 65535 (with scaling on or off), and the Analyzer is still showing you 65535, then I would have to guess you are not setting your RWIN correctly or you did not reboot.

I dont think your understanding. Just like I told Lobo, I am aware of what scaling is. How do you explain that the analyzer reports 65536 when my rwin is set at 32470, with scaling off?? Scaling only has to be on if you set rwin to over 65536, just like you said

You are correct! Sorry. I was looking at 32470 and seeing 324700!! My mistake -- I stand corrected.

You should not see 65535 if your RWIN is set lower than this. I have to suspect something is interfering. I suspect you have rebooted, but I always have to say that again just to be sure.

Then, are you using a GlobalMax setting? This may interfere, so if you have that, delete it.

Then, where exactly are you setting your RWIN? What program are you using, and where exactly in the registry is the change taking place?

All the settings are correct. I use cablenut. Anyhow, I only mentioned this because in the original thread, he claimed the analyzer was reporting his RWIN wrong, and I used to have this same problem and what fixed it was turning scaling on, reagrdless of what size RWIN I had.

I am also aware that scaling is not neccessary with small rwins under 65536, but it did fix the problem for me. Before I found cablenut, I used regedit, and all settings were correct.

I would like to see if this is the same case with this guy(wb7sdxgary).

Got it. But it remains illogical that scaling would be helpful if the RWIN is under 65535. Turning on scaling has *nothing* to do with the "window" field in the SYN packet. Therefore, there is no logical "cause and effect relationship". But I guess one can never exclude random luck or voodoo.