Hi, it looks like you're trying to edit a document. Would you like to turn autocorrect on?

Yes | Of course!
------
(X_X) Sorry, the application Life Support has ended abnormally! Dicarbon Monoxide synthesis failed.

Ignore | Retry | Fail
------
(X_X) Sorry, the application Send Error Report has ended abnormally! Could not connect to report server.

Ignore | Retry | Fail
------
Feb 28 05:05:08 lifesupport.mars kernel: Fatal trap 12: clippy runtime error while in kernel mode
Feb 28 05:05:08 lifesupport.mars kernel: cpuid = 0; apic id = 00
Feb 28 05:05:08 lifesupport.mars kernel: panic: user error, replace user, then press any key.

It can't unless it is sold in volume. Quality software takes time (read: money) to write, support, and bugfix. If you want anything with a limited market segment, by its very nature it has to be expensive to be of reasonable (and useful) quality or you are relying on the generosity of a not-for-profit software development organization.

But such a constraint makes the comparison apples to apples. To further stretch the illusion of comparability consider an i7-4700MQ or similar. Same market segment (you find them in MS Surface Pros, that's a tablet, right?) but without the power constraint, at 45W that thing blows away pretty much any chip in either ARM or Atom lines. i7, atom, and arm are all different optimizations of price, performance, power. From what I gather, in the mobile segment, performance/power is pretty much king.

Require OEMs to have >80% of devices sold in the last 2 years capable of running the most recent version of Android within 3 months of release.

Racked at the house...
Dell C6100, 4x { 2x L5639, 32GB ram }
Dell C1100, 2x L5639, 48GB ram
Dell Powerconnect 5324 with a permanent fan error because I unplugged it, the noisy bastard.
Ubiquiti ERL-3

Maybe half full of SATAs, most of which are spun down. Further sliced into VMs because I like to play around with KVM/ESXi. Mostly used for ADC, media serving, local backup, test build environments, autodeployment script tests, etc. It'll draw 6-8A @110V easy with everything spinning max load, but it usually idles around 1.2-2A as I only wake 3 of the C6100 sleds on-demand. Config exactly mirrors my colo, which pays for both, but is not my job.

Workstation's notable components are the G530, GTX460, a cheap-as-it-comes asus MB, and 3x displays.
Half a dozen or so ARM SBCs (rpi, udoo, beaglexM, etc) doing menial things like RTP audio endpoints, ethernet-attached GPIO, openelec.

That sounds amazingly convoluted and backward, but I don't doubt that's how it works.

However, if Qualcomm sells Samsung a part without licenses and separately licenses/sells/supports the driver software without licenses (I am dangerously assuming Samsung didn't write their own), conditionally saying the driver cannot be used with the part because that would be in violation of the patent, then Samsung uses them together and distributes it, why would not Qualcomm go to town on Samsung in the spirit of cover-your-ass? They'd have to be in collusion to commit patent fraud. But if Qualcomm licensed the driver to Samsung for use with the part, the two together seem like patent exhaustion would have to apply, unless, like you said, Samsung assumed responsibility for paying Qualcomm's patent royalties and fees. Samsung wouldn't be using the patent directly, Samsung would be using a part that uses the patent.

I guess fundamentally I am misapprehending how you can somehow sell a product or combination of products as fit for purpose that are covered by patents and yet not assume patent license liability. Thanks for your insight on how this whole thing works.

Ok, I think I now see where you are coming from here, as well as AC below you. Since Qualcomm didn't have a license, Samsung couldn't have triggered patent exhaustion because Qualcomm never had a license to exhaust. It's then open to court interpretation whether or not Samsung should be liable for use of the patent.

The difference between this case and Quanta v. LG (2008) is that Qualcomm didn't have a patent license to sell their part. Even if Samsung is not practicing the patent itself, they might still be liable. This is then the part that confuses me: if Qualcomm acquires a license for historical sales, wouldn't Samsung again be protected from infringement?

Thank you all for straightening out my confusion.

I'm specifically referencing exhaustion/first sale doctrine. I believe that should apply, but as you pointed out, I could be wrong.

Just curious, has anyone done a gpu power/performance comparison of Imagination Technology's powervr and Nvidia's tegra platforms recently? I don't know that they would clean up by licensing their IP out; it seems to me Nvidia stuff has always been more performance focused than power conscious which is something that is extremely hard to sell in the mobile segment.

I am not a lawyer, but I find it hard to believe Samsung is violating any of Nvidia's patents directly by using Qualcomm's Snapdragon 801 and 805 in a product. They received the part and associated driver software from QCOM as a final product and all components and features therein are protected from patent violations. Just like you can't be sued for violating Nvidia's patents by using an AMD GPU which has Nvidia-patented features in your PC, Samsung is protected by purchasing the part from QCOM. Nvidia could block further sales of the Snapdragon CPU to Samsung, but not sales of derived products; even though to to the end consumer it amounts to the same thing. So unless Samsung is violating their agreement with QCOM by enabling features they didn't license from QCOM, NV can't touch them here.

Similar deal with Exynos (Samsung's SOC) since it licenses the IP involved directly from ARM and Imagination Technologies (Mali and PowerVR GPUs respectively). Unless Samsung's legal team is collectively idiots and/or assholes, they should be protected by their upstream licensing agreements.

Then again, NV is never going to sue ARM because they would be in a seriously shitty position to renew *their* ARM licenses (if ARM didn't just terminate them on the spot) and then ARM would laugh all the way to the bank about who isn't shipping products.

Based on that, it's my opinion that Samsung shouldn't be involved in this lawsuit and Nvidia just pulled them in because that's where the money is.

FMIQ (as implemented in the imgSeek server) is pretty good at detecting everything but crops, and that was 5-6 years ago when I last played with it. It likely has no problems with them now. This is a reasonably well solved problem.

The only part I care about is being able to take the interchangeable radio/baseband unit out of model A on carrier X and put it in model B and continue my service on X with them none the wiser, or even remove the radio entirely and operate without cellular features. Maybe even swap in a part97 radio instead. Ok, that's asking for the moon, but I can dream.

From the way the article reads, it's more like everyone else made their save vs spell... Or perhaps that he lost his save vs PPDM. Seems like after he made his initial critical mistake (allowing investment options to bypass his majority ownership), he couldn't recover without just divesting himself from TSR and starting over before the flagship D&D product was born, which, as a primary creator, he might have been able to pull off.
The behind closed doors shenanigans, manipulations and backstabbery are about right for any D&D game I've ever been in.

They probably have regression test cases to make sure bugs in the source are preserved. Wouldn't want to spoil the original experience...

Or more established/full featured, openvz, xen pv, lxc, cgroups/namespaces, and friends. I think linux (the kernel) already has the tools necessary to do task prioritization like the article requests.