D-Link → Re: [Rant] DI-634M Users: Happy or Not?

I thought I read it in two articles. But now I can only find one, and I see now that I didn't read this one very carefully. Here is the link, please go to the tab called "Analysis and Caveats".

»www.tomsnetworking.com/S ··· age1.php

This MIMO Face-Off artical seems pretty good. Please also see Figure 23, where it ranks the D-Link dead last in WQS Uplink. Is this very important? (I don't really know what WQS means to me, but I assume this represents the uplink throughput)

Also, in this same review, in the 'Product Recommendation' pages, it seems to say that if you want both good throughput and less variability, then the chipset that D-Link does NOT use is the better choice. But if you want the best throughput under ideal conditions, then the chipset D-Link does use is better (at least that's how I read it, you may interpret it differently).

One more think, I found the following link on the DI-624M. Please see the bar-chart figure on mixed mode. It puts the DI-624M in last place. I wonder how the 634M compares to the 624M? If nothing else, this evaluation suggests to me that some routers handle mixed mode better than others.

»reviews-zdnet.com.com/D_ ··· 033.html

The ZDNet reviewer was perplexed by the technology, as was the D-Link tech support person that he mentioned who told him to put it in Static Turbo mode. That's not going to work at all with a non-SuperG testbed. (Un-)fortunately, it also crashed the router on him.

As for the mixed-mode speed, it is a little strange that the test results were so different than the other products using exactly the same chipset.

On possible explanation...

In a SuperG with non-Super 802.11g AND 802.11b in the mix, the AP can spend precious little time in a Channel Bonded (Turbo) state before it has to fall out in order to talk to the non-Super B and G clients.

So the trick is to give the Turbo connection enough service to get a high throughput without being in Turbo-mode so long that the non-Turbo clients sense that the AP has gone away entirely, or has a weaker or intermittant signal. (There apparently is some timing tuning that the OEMs, like D-Link, can do to make a difference. The end-user has no control over this timing, it's all done by the Atheros chipset and perhaps some patches or parameters that the OEMs can use. I only know that because there recently was a D-Link firmware update that adjusted this tuning, and I read about it in the release notes.)

Ironically, all of this switching speeds and frequency widths (bonding) for performance enhancements often results in slower throughput than if the Turbo feature is turned off entirely. This has been my own experience, too.