i thihk it stands for Advance Placement, which are (correct me if I'm wrong) voluntary classes a student can take in advance to build enough credits to be able to take other classes they don't qualify for, or to get fewer hours required in a future course.

It doesn't seem to me to be for those with a strong interest or skills.

Tom's Hardware (which back then still was run by Thomas Pabst), certainly called out Intel on it, and correctly deduced that the much longer pipeline was to blame. It became even worse with Prescott, the Windows Me of CPUs, because Intel did not want to admit that increasing the pipeline was a problem, so they did it again. It worked great for unrolled benchmarks with minimum branching, but not so much for real life code.

Finally Intel took to reason, and scrapped the horrible P4/PD Netburst design, and went back to look at the PIII design when starting over with Core 2. And almost overnight, Intel took back the throne from AMD.

Also, the flip side of the same coin: excessive copying.
I've seen plenty of code where each thread works on its own copies of data and locks the global, possibly thinking that if the speed of a single thread is increased through register usage instead of memory, it will also hold true for the sum of all threads. This approach only works if it's highly unlikely that any other thread needs the same data - else it will be slower than accessing slow memory repeatedly.

And relying on other layers to handle mutexes, which usually means nothing will be freed to use by other threads until the next tick at best. If speed is important, make sure you handle your own freeing of shared resources, as soon as at all possible, even if it makes the source uglier or the best-case slower.

The pentium 4 was a lemon. A 1.4 GHz PIII-S outperformed the first P4s running at 2 GHz. The pipeline had grown so much that "normal" code spent a good amount of time refilling the pipeline.
With the Core CPUs, they went back to the shorter pipeline, and real world speeds per clock jumped.

But I think the real problem here is designed bottlenecks combined with 50,000 feet programming and abstractions where the squeezes are. I'm sure they use a profiler, but it can't unravel bad design or remove unnecessary data copying.

Which would be fine, if every part of that composite at the same time gave up their right to personhood.

If I'm a person, I can't claim that my left and right brain hemispheres also are persons. I don't get 3 votes.

And? We already tell them that we don't consider them full persons, by curtailing the rights that real persons have, like voting, running for president, and as member of a jury condemning someone. They're not our peers, but only persons in potentia.

From what I can tell from the above, it shows that it's impossible to even discuss this in an objective way - any discussion will deteriorate into appeals to feelings.

I think the judge needs to state that animals have a right to be judged by their own, and that any act of giving them personhood takes away that right.

While Gay Deceiver was cool, I think Asimov nailed it better with Solaria - a future where you don't travel, but conduct all business and relations online. Travel is for robots, not humans, who can be freed to spend their time and efforts on cerebral activities and arts.

My main server is Gentoo. I have a few others, with various OSes, but they all suffer from lack of long term support and upgrade options. My Gentoo system was installed in 2002, and now, 12 years later it still runs - with everything upgraded many times over, both hardware and software. Everything is up-to-date with patches too, and features I do not want for security or resource reasons are left out.
It's the only OS I know of where either of this is possible.

RHEL 6 which doesn't have systemd will have support until 2017, and extended support until 2020. So they may not notice the full impact of pandering to Poettering until later.

I suspect that they think they'll get a large number of their customers to convert to using the cloud and no longer admin systems, so that it alienates admins doesn't matter.
I believe they are betting on the wrong whores.

I think the right to personhood should be given to anyone who of their own volition can claim the right. And yes, that also means taking it away from many who have it today.
Including corporations.

I don't see any non-human lifeforms being able to claim that right. Future computers might, or genetically modified/engineered animals.

But I believe most animals should still have the protection of being sentient beings, much like we protect infants and retarded people.

Many countries do not allow dual citizenship, at least not for adults. If I were to accept cttizenship in the US, I would have to renounce my current citizenship.

I find that reasonable. It avoids conflicts of allegiance, and is egalitarian, in that one citizen won't have the protection of two countries while another doesn't.

For a house, I recommend Cat6 for all permanent lines, and 5e only for the leaf lines. Same speed, but Cat6 is more resilient in several ways, at the cost of cables being stiffer and more expensive. But over a few years, whether you pay $100 or $150 for the cabling doesn't matter much. That you won't have to re-cable or run diagnostics to find out where the problem is can be more of a concern.

You're using the systemd udev, patched to remove most (but not all) of the systemd integration. You also get goodies like UnpredictableNetworkInterfaceNames with that one.

eudev is the non-systemd alternative.
# qfile $(which udevd)
sys-fs/eudev (/sbin/udevd)

Unfortunately, even in Gentoo it is now hard to avoid systemd. It sneaks in through dependencies.
At the very least, you need to add sys-apps/systemd and sys-fs/udev to package.mask, and add -udev to the use flags of x11-base/xorg-server. And if memory serves me right, stay at an older version of nfs-utils that won't try to integrate with systemd.