The 802.11ac press releases are coming thick and fast lately, this time it is Broadcom breaking new ground in marketing rather than tech. Unfortunately SemiAccurate can’t bring you a tale of new silicon or capabilities, instead Broadcom is pushing the boundaries of legitimacy in spinning.
You might remember Qualcomm’s MU-MIMO announcement from a few days ago which was followed by Quantenna’s 8×8 802.11ac release. Not to be outdone by these two 802.11ac Release/Wave 2 reveals, Broadcom came right out with a “six stream” press release. That appears to slot between the four of Qualcomm and the eight of Quantenna, right?
If you notice the quotes around six stream above you will be on the right track. The full Broadcom press release is entitled, “Broadcom Doubles Wi-Fi Speed of Devices with First Six Stream 802.11ac MIMO Platform“. Go read it before you continue, it looks pretty interesting on the surface. Doubling speeds, leadership, and unmatched performance are quite heady claims.
If you parse the claims instead of just believing the hype, the first of those three is just wrong, there are plenty of 4×4 802.11ac devices on the market now, a 6×6 would not double that by any means. The other two are similarly vague and could be technically accurate but again seem to be intentionally deceptive. Why? Lets take a deeper look at the claims.
There are three claims at the top of the release, the first being, “Offers up to 50 percent speed advantage over Multi-User, Multiple-Input, Multiple-Output (MU-MIMO) routers“. Qualcomm and others have a 4×4 router on the market so that would be four 80MHz streams, to the best of our knowledge Broadcom does not have any 4×4 solutions, only 3×3, but now they have six plus MU-MIMO, right?
A six stream chip would indeed double the speeds of the 3×3 devices Broadcom devices currently on the market but not that of the 8×8 Quantenna announced. We take this to mean the Broadcom part either has shipped or will be shipping really soon now, and we will assume it has. So this would be leadership and unmatched speed right?
If so Broadcom is still wrong because Quantenna has 4×4 devices on the market so this is only 50% faster. If it isn’t shipping however, Qualcomm’s MU-MIMO devices will smoke it by offering 50% more than that, 3×3 with 3 streams in MU-MIMO mode for a total of nine concurrent streams. Hold your questions for a bit, there is a point to this.
The second bullet says, “Delivers data rates up to 3.2 Gbps for the highest performance in a home router”. The math for 802.11ac says that each stream at 80MHz is capable of 433Mbps and 160MHz channels support 867Mbps. 3200/433 is 7.39, 3200/867 is 3.69. The Broadcom chip in this case is a claimed six channel part, not a 7.39 or 3.69 channel device. If you do a little math though, 6×6 gets you to 2.6Gbps and another 3×3 802.11n on the 2.4GHz channel gets you another 600Mbps for a total of 3.2Gbps.
This may seem like a leap of logic because adding up all the potential frequencies and modes of a device is about as sleazy as marketing gets, but we can’t get to 3.2Gbps any other way. Then again later in the release Broadcom says, “5G WiFi XStream is a platform for home routers that includes six 802.11ac streams and three 802.11n streams that run simultaneously.” The 3.2Gbps numbers are bandied about many times like, “By offering Wi-Fi data rates up to 3.2 Gbps“. Broadcom really wants to make it sound like they can do a 3.2Gbps stream with their wording, and it just seems unethical even if it is technically correct.
Then they go off the rails with marketingspeak, try this “key feature” on for size. “Five core “Penta Core” central processing unit (CPU) made up of two 5GHz radios and one 2.4GHz radio as well as 2.96 GHz processing power to maximize network capacity.” Wow this has 2.96 GHz processing power, whatever that means, across five cores? Gee whiz that is impressive, right? Each radio has one CPU and the 4707/8/9 Communications Processors likely used here have two more for a grand total of five.
Since the 470x line has dual core ARM A9s on a 40MHz process and typically run at about 1GHz, that would be 2GHz of processing power right there. The last GHz being spread across the three radios would put them right about where similar devices sit for clocks too. So Broadcom is taking five different CPUs across four chips, adding up the clocks of all the cores, and claiming it as a single clock speed? Again technically correct wording but flat-out intentionally deceptive.
Remember we said hold the thought about adding up total bandwidths across three discrete radios and calling it a number? Remember the part about not wanting to think that Broadcom was that sleazy? I still don’t want to think that but, well, what can you come up with for a better explanation?
The next “key feature” is that this platform has, “Support for recently released spectrum from the (FCC) through simultaneous multichannel capabilities in 5 GHz band to enable six 802.11ac streams over 160 MHz of bandwidth.” Wow they do 160MHz channels, no one else does. Actually if you read the text again, Broadcom doesn’t do 160MHz channels, they threw technobabble in with big terms like FCC to make it seem like they do but they specifically do not claim 160MHz channels because they don’t support it. More slime for the bucket.
Then there is this gem for a feature, “Automatic separation of coexisting 802.11n and 802.11ac Wi-Fi devices into distinct frequencies“. Once again, wow! Err, no wait, 802.11ac is 5GHz only, 802.11n is typically run on 2.4GHz but 5GHz is an optional part of the spec. Then again there are three distinct radios in the Broadcom platform, you would be insane not to run them at different frequencies, most consumer 802.11n devices don’t have 5GHz radios in them. This one is just so dumb I pity the idiot who wrote it and thought it would sound impressive.
The spectrum allocation for 802.11ac at 5GHz technically has enough bandwidth to support about six 160MHz channels even if things like crosstalk make it very expensive to do so in a single device. That said there are some high-end devices that do on the market now but they are not the norm. If you use the same Broadcom BendyMath (C)(R)(TM)(P) as seen above, Qualcomm could announce a 48-stream device capable of 20.8Gbps and Quantenna could have 48-streams with 41.6Gbps. Both companies for some reason don’t play such games, and we are better off for it.
In the end Qualcomm and Quantenna announced new silicon with 4×4 and 8×8 streams respectively. Both support 802.11ac Release 2 MU-MIMO and Quantenna also supports the other two optional features. Broadcom doesn’t have a 4×4 device yet but is claiming 6 streams by bundling two discrete 802.11ac 3×3 radios in a box, then counting the aggregate bandwidth. Not to be outdone they throw in the 802.11n radios too, after all a bigger number is just that. Don’t ask the question of how dumb they think we are, they just answered it. None of the radios support 802.11ac Release/Wave 2 features like MU-MIMO or 160MHz channels even though Broadcom desperately tries to make it look that way.
If there were just one of these technical mis-wordings you could forgive Broadcom but there are many, some repeated multiple different ways. They are way behind the 802.11ac device curve and didn’t want to be left out so they tried to one up the competition. Without the silicon they had to resort to the time-honored tactic of throwing a lot of mediocrity in a box and trying to obfuscate with the most tenuous of technically correct wordings. In total it is flat-out unethical and Broadcom has none of the tech they so desperately try to intone ownership of. At least this explains why they keep canceling briefings with SemiAccurate at trade shows, if this is all you have to stand on, I would avoid real questions too.S|A
Have you signed up for our newsletter yet?
Did you know that you can access all our past subscription-only articles with a simple Student Membership for 100 USD per year? If you want in-depth analysis and exclusive exclusives, we don’t make the news, we just report it so there is no guarantee when exclusives are added to the Professional level but that’s where you’ll find the deep dive analysis.
Latest posts by Charlie Demerjian (see all)
- HyperX ships it’s 60 millionth enthusiast memory module - Oct 15, 2018
- Bittware/Nallatech water cools 300W of Xilinx FPGA - Oct 12, 2018
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018