PDA

View Full Version : 2 laptops, same network - different speakeasy results


high_revs
01-06-2010, 6:15 PM
Been noticing 1 of my laptops is slow(ish). I did some speed tests via speakeasy on both, using same browser (FF) and versions of it. The only difference is the OS (1 Vista, 1 XP Pro). Isn't it strange that on the same wireless network (wpa enabled), same browser, same locations on speakeasy, the XP machine would only be 1/3 or 1/4 the results of my vista? Granted the Vista machine is more powerful, h/w wise (newer process, more ram, etc). They're also on the same location (my desk) so distance doesn't play a differing factor.

I ran spybot and malwarebytes and there are no trojans or anything on the XP one. Using AVG anti-virus and it's updated too.

But for pure internet speed tests, shouldn't they be remotely the same? At least comparable?

Turo
01-06-2010, 6:23 PM
But for pure internet speed tests, shouldn't they be remotely the same? At least comparable?

Nope. There's way more factors in internet speed tests than can be negated by using the same network. I don't know all the details of how it works, but I do know that it can depend on how many people are using that ISP in the area at that specific time. I'm sure dropping packets can affect it, and a million other things that you or I have no control over.

Speedtests like speakeasy are just there for rough estimates and I would suggest 3-5 of them in a row just to get an average if you want to be more accurate.

Also, the different OSs can play a large role as well. The way xp handles the internet is different than vista.

Another thing, you mention one is 1/3 or 1/4 of the other. Are we talking 1000 as compared to 750? Because I can see how that would fit in a normal deviation. If we're talking 10,000 as compared to 6600, then that may be significant. (I forgot which units they use, forgive me for that.)

So, long story short, just having all the variables that 2 different machines are capable of having is definitely enough to explain a difference in speeds, but there are also many other things that can be different.

Corbin Dallas
01-06-2010, 6:42 PM
Flush the DNS on both first, then run the test. Sounds like the vista machine could be caching items for a perceived increase in speed.

start>run>cmd

In the CMD window:

ipconfig /flushdns

high_revs
01-06-2010, 10:58 PM
hm... ok.. i'll flush the dns, clear all cookies and history on both machines.

oh.. when i said network, it's my personal home network/wifi. not work in the traditional sense. it's intenet line -> modem -> wifi router then to both laptops. I also reset the wifi router and the modem too. (also did remote router reboot via dd-wrt). didn't seem to make a difference.



Speedtests like speakeasy are just there for rough estimates and I would suggest 3-5 of them in a row just to get an average if you want to be more accurate.

Another thing, you mention one is 1/3 or 1/4 of the other. Are we talking 1000 as compared to 750? Because I can see how that would fit in a normal deviation. If we're talking 10,000 as compared to 6600, then that may be significant. (I forgot which units they use, forgive me for that.)


just selecting a couple of comments..

i did about 2 each (dallas, seattle, washington and NY). later one, los angeles too and SF. one laptop after the other (immediate once the upload speed test finished). it's good that speakeasy isn't limiting the tests unlike dslreports where one can only do 2 or 3 tests years back.

the indicated speeds from speakeasy and only selecting one city's results (rounding the numbers a tad)...

vista... ~14k megabit/sec kpbs
xp... ~3k or 4, megabit/sec kpbs

i'll have to do some test once the wife logs off since she's fond of streaming movies. i don't have any torrents running either or anything that would fetch from the internet (i closed windows side-bar that might get stock prices or local temperature).

the only other test i can do to eliminate the wifi (both should be on same protocol - 802.11g). i can hardwire it to router and/or direct to modem too. of course, it'll be hard to do back to back test immediately but it seconds in between might be close enough.

will try again and see.

djbooya
01-07-2010, 12:11 AM
Some factors between the 2 operating systems:

1) Network driver - do they both use the same 802.11 wireless card? Check to make sure the XP machine has the latest version of the browser on it.

2) CPU Utilization - open task manager (Ctrl-Shift-Escape), choose the performance tab and look at that while running on XP.. CPU shouldn't be very high...if it is that could be contributing to low results

3) Disk IO - if you have a slower disk on the XP machine the results could be effected. If you're memory gets swapped to disk and generates a bunch of disk operations that could effect the performance numbers as well.

I would suggest testing via LAN connections just to narrow the variables as well.

another test would be to transfer a large file between your 2 machines using the wireless using something like FTP w/filezilla since it will report bandwidth usage...if you can get at least 15Mbps on the XP machine in this test then there's something going on between the XP machine and the path to the internet, whether the wifi router or the lan modem or something.

nick
01-07-2010, 12:44 AM
Been noticing 1 of my laptops is slow(ish). I did some speed tests via speakeasy on both, using same browser (FF) and versions of it. The only difference is the OS (1 Vista, 1 XP Pro). Isn't it strange that on the same wireless network (wpa enabled), same browser, same locations on speakeasy, the XP machine would only be 1/3 or 1/4 the results of my vista? Granted the Vista machine is more powerful, h/w wise (newer process, more ram, etc). They're also on the same location (my desk) so distance doesn't play a differing factor.

I ran spybot and malwarebytes and there are no trojans or anything on the XP one. Using AVG anti-virus and it's updated too.

But for pure internet speed tests, shouldn't they be remotely the same? At least comparable?

It would also depend on your computer(s). The other load they were under at the time, hardware, the parameters of each laptop's TCP/IP stack, browser version and add-ons, etc. etc.

high_revs
01-07-2010, 12:21 PM
both laptops don't have the same wifi nic (or hadwire nic). as for h/w performance, vista one has 4gb ram vs. 2gb for the xp. hdd-wise, vista has 160gb vs. 60gb on xp. i don't think hdd performance (leading to i/o) would have much impact as browsing/downloading doesn't require heavy i/o activity (unlike a database where that's more critical with i/o, hdd performance and ram vs. cpu). if it was, i'd have a desktop with striping! :)

i didn't go the comparison tests yet. instead, i decided to just image the xp laptop back to original state - os, acronis imaging software and just wifi configured. nothing else. i even uninstalled the trial norton av just in case (which by the way was just going thru the prompts for setup, registration, etc. that i didn't go thru). i did the speed tests via wifi and they're more or less the same from the same cities used - seattle, dallas, washington dc, sf. well, at least i'm on a clean-slate on the laptop but i can restore the last image prior to wiping it out if need be (and where my data is backed up).

so i hardwired it to the router and the speeds dramatically increased! i was reaching almost 17,000 kpbs even for one city, 14-16000 kpbs for some cities. [i've seen up to 20,000 kpbs if i hardwire directly to the modem]. slowest was around 10,5000 kpbs for SF. just to further remove variables, i removed the bluetooth antenna (forgot about that since the xp laptop doesn't have built-in bluetooth), and it didn't have any impact. i thought it might (remotely but just to further remove variables) but i doubt it causes interference since a lot of laptops have built in bluetooth with their 802.11a/b/c/n.

so it might be a "dying" wifi nic card. i'll have to open it up to see if it's intel or broadcomm wifi nic, and upgrade the driver.

well, this is my current theory anyway.

Sinixstar
01-13-2010, 1:43 AM
Been noticing 1 of my laptops is slow(ish). I did some speed tests via speakeasy on both, using same browser (FF) and versions of it. The only difference is the OS (1 Vista, 1 XP Pro). Isn't it strange that on the same wireless network (wpa enabled), same browser, same locations on speakeasy, the XP machine would only be 1/3 or 1/4 the results of my vista? Granted the Vista machine is more powerful, h/w wise (newer process, more ram, etc). They're also on the same location (my desk) so distance doesn't play a differing factor.

I ran spybot and malwarebytes and there are no trojans or anything on the XP one. Using AVG anti-virus and it's updated too.

But for pure internet speed tests, shouldn't they be remotely the same? At least comparable?

Not necessarily. Especially if you're talking about a wireless network.

Try using the Ethernet port to direct wire both laptops to the router and try the results again - I bet they're at least closer.

Bottom line though - throughput speeds have to do with a lot more than just what the rated speed of the adapter is. The rated speed on the adapter is maximum throughput of that device, and that device alone. The rest of the computers hardware still has to deal with that data as it's coming in - and is only going to be able to process so much at a time.

The hardware differences between a factory new Vista machine today (Core2Duo, x64) is going to be able to handle a whole heck of a lot more than an XP machine that's probably a few years old (Pentium, no Dual Core, no x64).

I guess what i'm trying to say is - no, actually - that doesn't surprise me in the slightest.

bigcalidave
01-14-2010, 1:14 AM
If you have a connection that fast you should upgrade your wireless to N. You shouldn't have that big of a gap from your wired speed to your wireless speed. Get new nics for both laptops, some nice new N cards, and a good router. That should fix your speed problems.

high_revs
01-14-2010, 1:22 PM
Try using the Ethernet port to direct wire both laptops to the router and try the results again - I bet they're at least closer.

Bottom line though - throughput speeds have to do with a lot more than just what the rated speed of the adapter is. The rated speed on the adapter is maximum throughput of that device, and that device alone. The rest of the computers hardware still has to deal with that data as it's coming in - and is only going to be able to process so much at a time.

The hardware differences between a factory new Vista machine today (Core2Duo, x64) is going to be able to handle a whole heck of a lot more than an XP machine that's probably a few years old (Pentium, no Dual Core, no x64).



I did not do a comparison test but I plugged the older laptop straight to the router above and it was much, much faster. I put some data a few posts above. So while I do believe h/w upgrades are going to improve performance, the only h/w upgrade that would help out a lot in a browsing experience are (in order): nic (wifi or not), graphics adapter, cpu and that's it. I say this because i/o is not high when browsing (not like I'm using CPU's floating point calculation or anything like that that is CPU intensive). I narrowed it down to the card at this point. Hopefully it's just the driver. If not, time for a new wifi nic. hopefully one that'll fit in stock place and not use a USB or pc card slot.

Note: I put vista on and it's the same. It *might* have just been those moments... I did hit much faster speeds a few times on some different cities. I'll have to test-run different times of the data for same city to see my new averages.

If you have a connection that fast you should upgrade your wireless to N. You shouldn't have that big of a gap from your wired speed to your wireless speed. Get new nics for both laptops, some nice new N cards, and a good router. That should fix your speed problems.
Appreciate the reply, but I'd have to disagree with you. If I'm capped around (being generous here) at 25megabit/s (highest I got straight to modem was only 20 or 21 megabit, ever!), there is no need for 802.11N. N might get you 160+ megabit, but the wifi nics are not my bottleneck. My ISP is. 801.11G is theoritically can reach 54megabit/s, and that's still much above my ISP speed. Even if I'm at an avg working speed, I suspect it'll still be higher than my ISP limit, or worse, higher than my daily avg speed. Just would be throwing money away. Not to mention a new router, new nics for 4 machines.

Now if I transferred a lot of data between my home machines in a home network, 802.11g improvement will definitely be felt.

P.S. I was shown a demo of a "smart home" one time. they were running 802.11N. this was 2.5 years ago and N wasn't quite commercial yet. But that's because they had fat pipes (multi-fiber channel) and were streaming a lot of HD from the HD provider over internet to the HDTV. that definitely needed 802.11n since the ISP pipe was much larger for HD content delivery. 802.11g won't can't handle that and meet the quality needed for HD.

:)

sfwdiy
01-14-2010, 5:02 PM
Have you tried running a traceroute to a site like Google recently? Give that a shot and see if there are any hops that take an inordinate amount of time. Also, have each machine just ping google about 40-50 times and see if you end up with any dropped packets.

I noticed a huge slowdown on my network a while back and I realized that one of my routers started dropping packets like they were hot turds, I'm talking 50 to 60% packet loss. The router was failing. It could be a bad WiFi card in the XP machine doing something similar.

--B

shooterfpga
01-14-2010, 5:52 PM
edit your registry settings on xp and check the bandwidth again. the easiest way to edit them would be to use cablenut and the updated bandwidth settings. vista already has a built in bandwidth optimizer which on some (rare) networks is better left disabled.