Big difference in file copy speeds from XP to 98SE depending on which PC initiates the copy

  • Thread starter Thread starter QuienEs
  • Start date Start date
Q

QuienEs

This has always made me curious.

If I initiate a copy at my XP-home laptop [it uses 11g wireless] to
networked 98SE PC I get around 400KB/sec xfer rate.

If I initiate the copy at the 98SE PC, still copying files from XP
laptop to 98SE PC, it runs at around 1.3 MB/sec.

The speed is measured with DU Meter on the 98SE PC.

The CPU speed of the XP laptop is Athlon 1600+, the 98SE is Athlon
2400+.

Same gross difference whether I use XXCOPY or Windows Explorer
although XXCOPY is about 10% faster than Windows Explorer in both
cases.

Questions are, why might this be and is there anything I can do
differently when initiating the copy on the XP laptop to make it
run at the higher speed?

TIA, QE in NJ
 
This has always made me curious.

If I initiate a copy at my XP-home laptop [it uses 11g wireless] to
networked 98SE PC I get around 400KB/sec xfer rate.

If I initiate the copy at the 98SE PC, still copying files from XP
laptop to 98SE PC, it runs at around 1.3 MB/sec.
[...]
Questions are, why might this be and is there anything I can do
differently when initiating the copy on the XP laptop to make it
run at the higher speed?
Click to expand...

My guess is that the difference is in the size of the TCP Receive
Window. The default receive window size on Win 95/98/ME is 8760 --
meaning that there must be an acknowledge handshake every 8760 bytes
when sending to a Win 95/98/ME machine. The default Receive Window on
a Win NT/2K/XP machine is 17520 resulting in half the amount of
handshaking when sending to one of these machines. The size of these
windows and your resulting performance can be changed.

For more info, checkout the FAQ at:

<http://www.dslreports.com/faq/578>

For a useful freeware tool to change the receive window on your
computer, checkout:

<http://www.dslreports.com/drtcp>

HTH,
John
 
On Wed, 04 Aug 2004 20:26:53 GMT, John Wunderlich

||
|> This has always made me curious.
|>
|> If I initiate a copy at my XP-home laptop [it uses 11g wireless] to
|> networked 98SE PC I get around 400KB/sec xfer rate.
|>
|> If I initiate the copy at the 98SE PC, still copying files from XP
|> laptop to 98SE PC, it runs at around 1.3 MB/sec.
|>[...]
|> Questions are, why might this be and is there anything I can do
|> differently when initiating the copy on the XP laptop to make it
|> run at the higher speed?
|>
|
|My guess is that the difference is in the size of the TCP Receive
|Window. The default receive window size on Win 95/98/ME is 8760 --
|meaning that there must be an acknowledge handshake every 8760 bytes
|when sending to a Win 95/98/ME machine. The default Receive Window
on
|a Win NT/2K/XP machine is 17520 resulting in half the amount of
|handshaking when sending to one of these machines. The size of these
|windows and your resulting performance can be changed.
|
|For more info, checkout the FAQ at:
|
|<http://www.dslreports.com/faq/578>
|
|For a useful freeware tool to change the receive window on your
|computer, checkout:
|
|<http://www.dslreports.com/drtcp>
|
|HTH,
| John
=======================
Thanks John for your quick answer. I'll download that registry tweak
tool, but before I use it I want to be sure I clearly presented my
situation, IOW:

The file copies in both cases I presented are FROM the XP machine TO
the W98SE machine. The big difference in transfer speeds depends
on which machine I use to initiate the xfer. If I sit at the laptop
[XP] and send the files to the desktop [W98SE] it runs at ~350KB/sec.
If I sit at the desktop and "pull" the files to the desktop, IOW
initiating the xfer from the desktop makes it run 3-4 times faster.

Thanks again, and in advance for any reply to this clarification.

PS - in any case I'll be having some fine with Dr TCP.

QE
 
The file copies in both cases I presented are FROM the XP machine TO
the W98SE machine.
Click to expand...

Ooops, My mind glossed over that point. I have a similar setup at
home. I'll have to try it and see if I get the same results...
-- John
 
On Thu, 05 Aug 2004 15:18:15 GMT, John Wunderlich

||
|> The file copies in both cases I presented are FROM the XP machine
TO
|> the W98SE machine.
|
|Ooops, My mind glossed over that point. I have a similar setup at
|home. I'll have to try it and see if I get the same results...
|-- John
===================
Thanks in advance, I'd like to hear [err see] the results. I upped
the number on the 98SE PC with DrTCP and it didn't cause any
appreciable change.

Cheers, QE
 
===================
Thanks in advance, I'd like to hear [err see] the results. I upped
the number on the 98SE PC with DrTCP and it didn't cause any
appreciable change.
Click to expand...

Interesting... Here are my results on a 100-Base-T connection via
wired router:

Win98 pull file from WinXP: 47% bandwidth utilization
Win98 push file to WinXP: 35% bandwidth utilization
WinXP pull file from Win98: 22% bandwidth utilization
WinXP push file to Win98: 15% bandwidth utilization

This means that there is a 3x time difference between Win98 pulling a
file from WinXP and WinXP pushing the file to Win98 (this matches your
results rather closely). Now I'm curious, too.

-- John
 
On Thu, 05 Aug 2004 21:33:45 -0700, John Wunderlich

||
|> ===================
|> Thanks in advance, I'd like to hear [err see] the results. I upped
|> the number on the 98SE PC with DrTCP and it didn't cause any
|> appreciable change.
|>
|
|Interesting... Here are my results on a 100-Base-T connection via
|wired router:
|
|Win98 pull file from WinXP: 47% bandwidth utilization
|Win98 push file to WinXP: 35% bandwidth utilization
|WinXP pull file from Win98: 22% bandwidth utilization
|WinXP push file to Win98: 15% bandwidth utilization
|
|This means that there is a 3x time difference between Win98 pulling a
|file from WinXP and WinXP pushing the file to Win98 (this matches
your
|results rather closely). Now I'm curious, too.
|
|-- John
================
Thanks John. Very interesting and very complete results. I'll be
away from home for a week and will be able to continue experiments
only when I return. I will bring laptop and try to follow this
thread if and when I get internet on the road.

Have done only 1 more experiment, and it did not make any difference,
I turned off ZoneAlarm firewall on both machines. BTW, I never have
CPU hogging Anti Virus running in the background.

QE
================
 
Back
Top