Three Stream N Performance: A First Look

[RogerSC]

This is an interesting article. I'd really like to have had the Cisco E4200 included as well as the Netgear WNDR4000, though, just to get a feel for the comparison between the two with three-stream.

Thanks very much for doing this kind of review.

 

[thiggins]

Didn't have an E4200 on hand, but Cisco is sending another.
Look for a follow up article with E4200 and D-Link DIR-665 results.

 

[john4200]

3-stream bridge?

Are there any 3-stream bridges? I would like to see the throughput of a 3-stream bridge paired with a 3-stream AP/router.

Maybe there are no 3 stream bridges yet? When the TRENDnet TEW-680MB comes out, it will be interesting to see the throughput when paired with the TEW-692GR.

One thing I am suprised that I haven't seen yet is....ummm, cannot find a good word for it. It should be simultaneous dual-band (2.4GHz and 5GHz) transmission, but that phrase is already used to mean something different. I mean that the wireless link should split the data between the two bands, effectively adding the throughput of the two bands together in one data link.

 

[thiggins]

Yes, TRENDnet makes both single and dual-band versions.

Already did this for one of the TRENDnet bridges.

Dual-radio clients don't exist.

 

[john4200]

Yes, TRENDnet makes both single and dual-band versions.

Already did this for one of the TRENDnet bridges.

Dual-radio clients don't exist.

I guess I should have said "3-stream N 5GHz bridge". That is what I meant.

I figured "dual-radio" clients do not exist. But I am surprised that no one has tried to make a bridge - router combo that does it. You could almost double the throughput of a conventional bridge - router link.

 

[stevech]

I guess I should have said "3-stream N 5GHz bridge". That is what I meant.

I figured "dual-radio" clients do not exist. But I am surprised that no one has tried to make a bridge - router combo that does it. You could almost double the throughput of a conventional bridge - router link.

Combining IP streams requires use of an IP (IETF) standard that's rather complex and wouldn't have much of an audience for WiFi client devices.

Also, it has been said that Wireless is for high mobility devices, or conversely, without mobility, there's no need for wireless (with the exception of too-hard-to-run-cat5-to-living-room). Then, the more mobility one has, the less speed is needed, in broad/general terms.

 

[john4200]

stevech:

Was I not clear? Do you not know what I mean by a bridge - router combo?

The reason I am asking is because your response is irrelevant to what I was talking about.

 

[thiggins]

The reason I am asking is because your response is irrelevant to what I was talking about.

Not at all. It reinforces the point of why it is unlikely you'll see simultaneous dual-band solutions (of any format) developed.

 

[thiggins]

I guess I should have said "3-stream N 5GHz bridge". That is what I meant.

You were clear. I didn't read carefully.

The TRENDnet 2.4 GHz bridge ("gaming adapter") is still the only three-stream device available.

 

[john4200]

Not at all. It reinforces the point of why it is unlikely you'll see simultaneous dual-band solutions (of any format) developed.

No, his response was irrelevant, because he was talking about mobile clients, not bridges to non-mobile computers.

It is obvious that there is a demand for non-twisted-pair high bandwidth links. Otherwise there would be no powerline networking adapters. But powerline networking throughput is highly dependent on the wiring in a home. I have seen some truly horrible throughput with powerline networking.

There are also a number of wireless bridges being sold today, some of them named such that you can see they are going after movie streaming and/or gaming markets, where a non-mobile computer is far from a twisted-pair network port. For these types of adapters, it seems you are doing very well if you can get a reliable connection that never goes under 60 Mbps.

But if you could utilize both 2.4GHz and 5GHz bands simultaneously, it should be possible to get a link that never goes under 100 Mbps, even in moderate signal conditions. Probably could get a minimum of 50 Mbps even in fairly weak signal conditions. Such a link would be a lot better for streaming high-bandwidth movies (1080p with HD audio) than a slower one that sometimes drops below 40 Mbps because it only utilizes 2.4GHz or 5GHz.

If only a company would create the hardware to do this. It seems like a company like TRENDnet should be quite close. They sell simultaneous dual-band bridges and routers. If only they would work on a link protocol and algorithm for dividing the traffic among the two bands, they would be able to do it. Obviously it would be non-standard and only work between the same companies hardware, but that would seem to be a big advantage for whichever company does it first.

 

[Morgan]

Disappointing results, I wonder why I still haven't seen a router based on the promising Quantenna chipset.

I do would love to also see a three stream test added for the new Trendnet TEW-692GR

 

[thiggins]

Disappointing results, I wonder why I still haven't seen a router based on the promising Quantenna chipset.

I do would love to also see a three stream test added for the new Trendnet TEW-692GR

Quantenna has been focused on application specific markets, such as media bridging.
NETGEAR WNHDB3004 3DHD Wireless Home Theater Networking Kit Review

TEW-692GR review is coming. Don't expect that it will be any / much better since it is Ralink based.

 

[voodoobs]

chart update?

whilst looking up information on buying a new router, i stumbled upon this awesome
website.

but one small thing.

on this page,
http://www.smallnetbuilder.com/wire...ree-stream-n-performance-a-first-look?start=1

you note that netgear asked to retest the throughput with a (then) new firmware
on the wndr3800, showing drastically improved results.

the new numbers you tested are not shown in the actual review,
http://www.smallnetbuilder.com/wire...gabit-router-premium-edition?showall=&start=1

OR in the charts,
http://www.smallnetbuilder.com/lanwan/router-charts/bar/74-wan-to-lan

any reason why, or just forgot to update?

 

[thiggins]

The retest was for the WNDR4000, not the WNDR3800.

 

[sfx2000]

Combining IP streams requires use of an IP (IETF) standard that's rather complex and wouldn't have much of an audience for WiFi client devices.

Also, it has been said that Wireless is for high mobility devices, or conversely, without mobility, there's no need for wireless (with the exception of too-hard-to-run-cat5-to-living-room). Then, the more mobility one has, the less speed is needed, in broad/general terms.

Stevech - how does 3*3 have anything to do with the IP layer? The streams are combined at the PHY... IETF has nothing to do with the PHY, that's IEEE's job...

And, I'm not certain about your second comment about mobility and speed... MIMO is all about multiple paths and combining, and mobility is an area where MIMO is a benefit with STA's that are truely mobile (e.g. not nomadic - fixed in different locations).

 

[mgiammarco]

I do not understand test results

I ask for a simple thing (I do not understand tests results):

if I bridge two 3x3 mimo accesspoints (eventually using dd-wrt or openwrt) can I reach greater speeds than 100mbits (10mbytes) per second?

I hoped to reach 200mbits or greater!

Thanks,
Mario

 

[thiggins]

What do you mean by "bridge"? You cannot connect a client to more than one access point at a time.

 

[mgiammarco]

What do you mean by "bridge"? You cannot connect a client to more than one access point at a time.

Thanks for reply. I mean no clients: two access points that connect each other. So I can bridge two differents wired networks.

 

[thiggins]

You might be able to reach > 100 Mbps if the distance between the two APs is short and there is little traffic from neighbor networks on the channel you are using for the bridge. You would also need to use 40 MHz wide channels, which you should use only in 5 GHz.

Note that you have a better chance of reaching high total throughput for multiple connections. Single connections will probably not reach > 100 Mbps.

 

[jayjayk]

3 stream clients?

does anyone know how many or how ubiquitous 3 stream clients are in the market place right now?

also, does 3 stream benefit 2 stream client performance?