Since there have been many reports of problems with the Linksys WRT610N that seem to point to poor thermal design, I decided to do a little stress testing on the 825. I set up two notebooks with the DWA-160 A1 in one and the B1 in the other. I then ran a few multi-hour tests with IxChariot running both links full blast simultaneously.
I ran the tests for three hours, then five hours and both ran without a hitch (Figure 14). The speeds that I got when running both radios simultaneously matched up nicely with what I measured when running only one radio at a time.
I also swapped the A1 and B1 versions of the DWA-160 so that each operated in both bands. I didn't see a significant speed difference between them, but remember that I was testing under same-room conditions.
Figure 14: 5 hour simultaneous dual band stress test
When I tested the WRT610N, I found some interaction between wireless and routing throughput. So I decided to run the same test on the 825. I kept the two notebooks running simultaneous links on both radios and added a third LAN to WAN wired stream to exercise the routing portion of the 825. The wireless traffic ran to a LAN side client, so didn't add to the single wired LAN to WAN stream.
Figure 15 shows the three stream test. The router kicks in at the 10 second mark and you can see a slight decrease in the 2.4 GHz radio throughput, which is running higher than the 5 GHz radio. But the really interesting part happens when the wireless streams end. When the 2.4 GHz radio goes idle, routing throughput jumps by around 60 Mbps! Then when the 5 GHz stream stops, it jumps up another 80 Mbps!
Figure 15: Simultaneous 2.4 and 5 GHz plus LAN > WAN
I was surprised that the Ubicom processor is doing so much heavy lifting for the wireless radios. My guess is that this is due to WISH (Wireless Intelligent Stream Handling) processing, which is on by default and that I left on for the test.
I ran a few other tests with routing traffic speed dialed down. Figure 16 shows routing traffic speed limited to 100 Mbps, which still causes wireless throughput reduction.
Figure 16: Simultaneous 2.4 and 5 GHz plus LAN > WAN - 100 Mbps routing limit
But when I limited routing traffic to 50 Mbps, there was no effect on the wireless streams.
Figure 16: Simultaneous 2.4 and 5 GHz plus LAN > WAN - 50 Mbps routing limit
100 Mbps + Internet connections are pretty rare and continuous traffic like that used for these tests is also unlikely. So I doubt if the average buyer will come anywhere near stressing the 825's CPU and looks like the thermal design is adequate.
As I said at the top, the good news about the DIR-825 is that it is competitively priced and, more importantly, available. My guess is that D-Link is going to focus on the 825 as its mainstream simultaneous dual-band router and just let the 855 whither away.
There is really nothing in the 855 that I can see that justifies a 2X price difference. The real question is why D-Link didn't bring out all three antennas on the 825, given that the radios are 3T3R designs and the cost of the combiner components is peanuts. This holdback would indicate that D-Link intends to keep on selling the DIR-855 for anyone foolish enough to pay the ridiculous premium.
As the "best throughput" tables back in the Performance section showed, there is no current product that delivers excellent performance both uplink and downlink and across both bands. No matter what you choose, you'll be making a compromise.
The bottom line is that if you have been thinking about buying the 855, then save some money and get the 825. But if you want to get the "best" simultaneous dual-band router available then, frankly, keep your money in your pocket. There really isn't a "best" in this category right now and I don't see anything on the horizon. If you're buying D-Link, you would be better off with the single-radio DIR-628 or DAP-1522 (if you don't need a router) and just buy two, if simultaneous band support means that much to you.