Comment #13 on issue 6356 by albertod.zgz: TCP receive/send windows are too  
small (32KB) causing slower transfers
http://code.google.com/p/chromium/issues/detail?id=6356

I disagree :)

I just re-tested everything. Method was to use Wireshark to log TCP packets  
locally. For additional
verification, I used an online checker (  
http://www.speedguide.net/analyzer.php ) which sniffs incoming
TCP packets and tells you their parameters, including TCP window size.  
These tests were conducted with the
build at http://build.chromium.org/buildbot/snapshots/chromium-rel-xp/10983/

Scenario 1 - XP SP2
IE/Firefox: 17520 bytes
Chrome: 32768 bytes

Scenario 2 - XP SP2 + registry change (TCPWindowSize = 256000)
IE/Firefox: 256960 bytes
Chrome: 32768 bytes

Scenario 3 - Vista SP1
IE/Firefox: Variable (5840-256960)
Chrome: 32768

Some explanations: both XP and vista round to the next multiple of MSS (in  
my case 1460) for maximum
efficiency, hence 256960 when requested 256000, unless overridden (32768 is  
suboptimal for this). On
Vista, due to autotunning, the initial window starts small (sometimes it's  
8760 or even 5840) but quickly
grows to an optimal size (in my case, up to 256960) during transfers. In  
the case of Chrome, it stays
fixed at 32768 no matter what.

In practice, this limits speed just as expected. This is a table of  
download speed tests that I registered
(under scenario 3, with a connection limit of about 600KB/s):

PING (ms)   CHROME SPEED   FIREFOX SPEED
     44-56            589             591
     87-92            346             590
   172-179            182             579

You might need to choose a server that is far away to see the effect  
(USA<->Europe should give about
150ms). The maximum speed Chrome will ever get is 32*(1/ping), where ping  
is the RTT in seconds.

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

--~--~---------~--~----~------------~-------~--~----~
Automated mail from issue updates at http://crbug.com/
Subscription options: http://groups.google.com/group/chromium-bugs
-~----------~----~----~----~------~----~------~--~---

Reply via email to