Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[kpiersci]

I’ve benchmarked the following performance numbers.  The systems tested were 2 Core2Duo/2GB machines running Vista Ultimate: a desktop with Gigabit and an HP dv9000 Laptop with dual wireless adapters.  2 different routers were used for comparison purposes (see below).  The test consisted of copying 8 files (2GB Total) from laptop to desktop.  The first set of numbers refers to perfmon's average; the second, in (), to Vista's copy-dialog-box estimate -- both are substantially identical.


                    WCG200                                               DIR-655
WirlessG:     1,160,000 Bytes/sec (.99MB/sec)         1,470,000 Bytes/sec (1.47 MB/sec)
draft N:       307,000 Bytes/sec (261KB/sec)            5,900,000 Bytes/sec (5.7 MB/sec)
Gigabit:       12,000,000 Bytes/sec (9.3 MB/sec)       13,000,000 Bytes/sec (19.5 MB/sec)

Wireless G tested using:  Intel Pro Wireless 3945 ABG Wireless Adapter
draft N tested using:        dLink DWA-643 Xtreme N ExpressCard Adapter
Gigabit tested using:        Intel Pro 1000_PL Network Connection

WCG200 refers to:          Linksys Wireless G Cable Gateway w/100-baseT
DIR-655 refers to:           dLink Xtreme N Gigabit Router            

+ In the case of pairing the DWA-643 with the WCG200 using Wireless-G, the result is worse than using 2 ‘G’ devices by a factor of +/- 3X.

+ Using DIR-655’s wired-gigabit capability (with gigabit link confirmed by driver) is only marginally faster than a 100 base-T connection.  Cat6 cabling used.

+ Maximum measured ‘N’ throughput was 5.7MB/sec with adapter and router <1m apart.  No noticeable change was observed between N-mode and mixed-mode.


These numbers fall far short of dLink’s advertised performance.  The maximum throughput I was able to measure was approximately 15% of 300Mbit (wireless) and (also) 15% of 1Gbit (wired) advertised maximum for these products.


2 Questions:

1.) Is this the expected real-world performance of dLink’s devices?
2.) Are there any settings which can  _significantly_ boost performance in a simple file-transfer scenario?

[crion]

I usually work with LAN/SAN backups, and one thing I immediately noticed. You are benchmarking network performance with a laptop! Second thing, you don't state what performance you achieve without any router involved.

I would advise you to do a point to point crossed TP cable between your laptop and desktop to establish what they can accomplish without any router involved. Without this, your benchmark is pretty useless..

[Rad Spencer]

Using windows copy utility is not a real benchmark of anything except how slow your laptop is. If you want to see realistic benchamrking numbers, use a utility that is designed to do that. Laptops are known for having slow harddrives to save on space and heat.
Your "benchmark" is completely invalid. 

I own a DIR-655_a with 1.11 firmware and I use a utility called Chariot, it reports acurate numbers becuase the harddrive is not involved.

[Rara Avis]

I second the above.

If we wanted to make this experience worthwhile, since I know this thread is going to attract onlookers like flies to refuse, lets try this.  Why don't we all get our machines and run a network throughput test and post the results.  As noted above a Windows file transfer is a joke so lets use a reputable network benchmarking suite for this test.

Please include as much information as you can with your results, we need all of the variables so we can replicate and validate your results.

This will allow people to see not only what the equipment is and is not capable of but will spread the wealth of knowledge when it comes to network benchmarking.

<edited by AWDL, for tone and misunderstanding problems.>

[Lycan]

This post is becoming flamish. If useful information is not posted in the next post I will lock the thread.
Also watch the tone of the posts please, we're all friends here. If you feel that your router is not performing up to par, try asking questions before blasting the product. That is not productive for anyone.
Thank you.
 

[kpiersci]

Hello All --- I am the OP and I appreciate your feedback.  I'm not sure why this is such a heated issue:  I am simply trying to get some decent performance out of it and I am pointing out that the quantifiable performance I have observed and measured is sub-par.

I do acknowledge that file-copy is hardly an optimal benchmark, but I'm not a HW reviewer.  Using the right sort of performance-measuring utility, it's probably trivial to verify 300Mbps on these products.  One of the replies is about 'Chariot':  that's great, but will my data transfer faster if I use it?

That's not my issue.  I care about a real-world scenario:  I want to transfer files between my home media server and my laptop.  Arguably, laptops are neither RAIDed nor optimized for network performance, but why else would I invest in wireless products?  Why, for that matter, does dLink sell the DWA-643, which is a product designed for laptops and advertised to deliver 300Mbps, if that number is inherently unrealistic?  I think my laptop can do better than 5MB/sec -- that's all.  And, yes, with a wired connection it does. I might never be able to do a sustained data transfer at 300Mbps, but I should get better performance than what I'm seeing today.  Based on what I've seen, dLink's DIR-655/DWA-643 combo is throttling my network throughput and I wish to understand why.

Windows file transfer is NOT a joke.  Actually, windows file transfer is the bottom line.  I can run this benchmark and that benchmark, get all kinds of numbers, but at the end of the day I just want to use my network and my dLink network products to copy data from one HDD to another.  I'm intersted in how fast other people are able to do this.  So, yes, please, post your numbers.  If you have RAID5s or 5400RPM laptops... all numbers are valid.  Please let us know what kind of end-to-end throughput you're seeing.  I'd be reassured if I was having issues elsewhere and my network HW was not to be blamed.

Moderator:  kindly do NOT lock this thread.   The question I'm bringing up is a valid one, I think.  Let's discuss it.  If dLink has an official response to this, I'd love to read it.  And if there is somebody from your support department who thinks they can help, so much the better.

[AWDL]

I hope this thread isn't ignored. It is a great opportunity for the members of this community to show what they can get out of our routers and a cards with certain set-ups.

May someone can choose a file we all download (non-virus) to send around our network to get sharing speeds (drag drop, top level) or ftp send/receive speeds.

I know YMMV, but no one is asking for anything but your mileage in your environment.

This will not be "connection only" speeds. They will be real world, spinning harddrive, through controller over Motherboard, out Wireless card, through house, to Router, out wired or wireless port, to computer with the components  and OS that it has.

Note to kpiersci: Don't describe results as benchmark without details. The responses to you will seem heated.

Note to everyone: Have fun good luck.

[kpiersci]

AWDL -- What details, exactly, are missing?  I'd like to be as helpful as possible.  I've compiled some interesting PERFMON graphs showing transfer ratest over time (GIFs) and I'd like to attach them to my posts, but the file-attach option does not seem to work.  I'm seeing great variance in draftN transfer-rate, as much as +/- 50% of average.  In comparison, my WirelessG is rock-stable, fluctuating within a very narrow band.  Is this the expected behavior of draftN?

[Rara Avis]

It sounds as if I have been a bit rash, the reason this was a heated issue was that your claims seemed to be a claim that it is now clear you were not trying to make.  My apologies.  As for settings to significantly boost performance as you requested in the top post and hinted at in your second post there is no real magic bullet that said there are some hard facts that I'm just gonna throw out there as a good starting point to understanding.

Wireless interference directly detracts from data rate.
Due to the MIMO nature of N it is more susceptible to wireless interference than G.
Encryption directly detracts from data rate.
Advertised data rate is never seen in a network benchmark let alone a file transfer.

Notice in the above I have been ignoring your wired problem, which is because I have a test I would be pickled pink if you would run.  If you connect your 2 gigabit computers with a crossover cable what transfer speed do you get?  While we’re at it lets try something a bit weirder, let’s turn off your router and connect Ad-Hoc and what speed do you get?  What do these prove?  Not a whole lot, however it guarantees that you don’t have a problem with a router slowing your network (which I suspect).

As for suggestions, have you run the gamut with changing channels and toggling encryption?  I believe your N fluctuations may be due to the interference (hence the big spiel on it).

And call me stubborn but I still speakernet files, the biggest file I transfer over my router are the front page to Slashdot (not really but it’s close) and I get that fast enough at G.

P.S.  Thanks for the edit AWDL

[CrunchyDoodle]

I too am quite disappointed, though not surprised, that the real-world performance of D-Link N wireless is only a fraction of the marketing claims. This morning I spent an hour on the phone with D-Link technical support on this issue. I have a DIR-655 with the most recent firmware and using WPA2 security as outlined in the first post of this forum. I have two DAP-1555 bridges; one in the next room and the other in the livingroom one level down. I cannot stream HD. I can stream SD. I appear to be getting about 20mbps in the wireless link. This is in spite of the DIR-655 wireless status showing 90% signal and up to 264mbps on this link. Yes, here in the real world I have at least a dozen wireless networks that show up in a scan. With the DIR-655 set for auto, it always finds an unused channel.

Bye.       

PS. I've tried Buffalo and ASUS with similar results.

[Rad Spencer]

With the DAP-1555, are you running encryption? Does disabling the encryption effect the overall performance?

[CrunchyDoodle]

As I mentioned in my original posting, I have at least a dozen wireless nodes that show up in a scan. That means they can see me too. Running WPA2 on the router is recommended for 300mbps performance in the first posting of this forum. That means that a DAP-1555 that might be linked to it must be using the same security.

Good security is mandatory. Except if you live deep in the woods and can't see any other wireless nodes, there is no other real-world choice. Yes, someone might try using MAC address access control, but this is relatively weak.

I'm not even willing to try any testing without security enabled, and I should not need to since WPA2 is recommended for 300mbps, which would be most satisfactory.

Bye.     

With the DAP-1555, are you running encryption? Does disabling the encryption effect the overall performance?