This is the right ruling. These guys are the biggest IP bullies out there in tech. Their entire business model depends on it.
But seriously, Qualcomm is the worst company ever to work with (except maybe Oracle). They are like a gorilla - they won't budge no matter how hard you push. Every little thing costs licensing money - from documents to tooling. We have a Qualcomm SoC for my current project, and support is abysmal.
They put in a minimal effort to provide support and only after you've completed the onerous requirements to submit a support ticket. It involves retrieving tons of dumps and logs and such using their various Q-tools (which are very buggy and half-baked; many of them cost $25k+/year to license). Most of the time they'll just shrug and say "works on our setup" and refuse to help. We'll spend hundreds of man-hours debugging their crap until we find the issue in their kernel driver and then they won't even acknowledge it was a problem and will say "qmi_wwan is an open source driver. We do not support it." Right, so you expect everyone else to maintain and fix the bugs in qmi_wwan even though virtually ALL of your customers use it and you yourselves rely on it as a critical piece of your development platform. Good one, Qualcomm.
I thought Oracle was the premier example of that :)
http://bonkersworld.net/organizational-charts
> But seriously, Qualcomm is the worst company ever to work with (except maybe Oracle).
OK, we're on the same page ;)
But they are really good at testing. They will make darn sure their software works under their extremely narrow and fragile definition of a test platform.
If you deviate from their test platform at all, you can kiss support goodbye.
"Oh, you are using Linux 4.X to talk to our chip? Nope, no support, we specified Linux 3.19 in the test setup, you are on your own."
What sucks is that our platform involves a lot of different vendors - a Marvell SoC, a Qualcomm SoC, etc. All the other vendors will work with you.
But with Qualcomm it's their way or the highway. They know they can get away with that behavior.
Predictably, months after the acquisition, all atheros documentation initiatives were killed, and their new generation of chipset requires a gigantic closed source blob that barely functions, with an absolute minimum of functionality (you cant even change the MAC address!)
Do you ever point out some terrible design decision and they reply “working as designed” and then do nothing?
"This is a terrible design decision, we should revisit it."
Like ordering a parachute, and the vendor ships you a concrete contraption. And when you call in, they seem incapable of understanding how there could be a problem because it's within their specifications.
Tech jobs in San Diego pay terrible, but Qualcomm is something else. There's a Chili's a block away from Qualcomm, on Mira Mesa Blvd, and I think you could make more money waiting tables there.
No joke, I've had recruiters hit me up for jobs at Qualcomm that paid about 50% more than minimum wage and required a college degree and five years of experience.
I was only an intern there, but I'm 100% sure you are blabbering.
I think you are talking about Rambus, not Qualcomm.
>They put in a minimal effort to provide support and only after you've completed the onerous requirements to ...
Don't you have to sign a BCPA or something along that line to get full support? Apple had full access to source code -- which then Apple turned it over to Intel, QCOM alleged -- and also demanded that QCOM update their hardware and drivers at least once a year.
Probably. I work for a much smaller company than Apple so no doubt we are doing everything we can to prevent Qualcomm from sucking up the entire project budget. I don't know the exact details but I'd bet we have the lowest tier support plan available.
There is, IMO, some validity to the claim that Qualcomm's SEP portfolio enables a significantly different device. Their analogy refers to the ipod and the iphone, whose BOM in some cases differs by only the modem and yet it allows them to charge their customers an enormous markup. Arguably the value-add of that part and the patents it leverages must be a large portion of that markup, so perhaps it's fair to consider the device's cost with respect to the royalty.
What you're measuring there is the value of the standard and the cellular network as a whole. That includes not only Qualcomm's patents but also everybody else's wireless patents and the entire network infrastructure built by Verizon et al. Which is exactly why standards-essential patents are treated specially.
Moreover, the other obvious problem with it is that there are other devices with exactly the same modem technology in them which vary in price from $50 to $1000 based on processor speed, screen size, operating system, etc. and yet Qualcomm is claiming a percentage of the total device including the value of all the components they had nothing to do with.
In other words, could QCOM effect price discrimination organically without knowing the cost of the full device?