Yup, imagine how easily another four cores would fit in there.
However that would compete with Xeons. Can't have that, it's where the money is.

... in severance packages. A hostile work environment will definitely reduce personnel.

Of course the smart people who have no problem finding another job will leave first.

Pfft. God put those bones in the ground 6,000 years ago to test our faith, duh.

It works perfectly well for me.

Obvious question.

It specifically forecasts a doubling of the number of transistors per generation. We've had 20-ish and recently 10-ish percent for many years.

The open drivers work just fine for everything that's not bleeding edge. We'll see how the amdgpu driver comes along.

I wish I had mod points. What can you expect if politicians are financed by the highest bidder?

Well voters rarely kick out the incumbent even if they didn't deliver what they promised. So yeah voters are to blame. They still believe today what Reagan promised just because he played a nice guy.

We'll see if they'll be fine with the negative publicity if the Linux drivers are crap.

That's because manufacturers run into limits, especially around cost, since Moore's law has reached the end of the line. A transistor on 20nm or 14nm is more expensive than a transistor on 28nm.

I put a SSD in front of my spinning rust RAID as a cache and use writethrough or writearound. Even if the SSD fails I won't lose data.

That could've been me 40 years ago. We had a whole group of bomb makers. They all ended up as chemists/chemistry teachers or MDs; I was the odd one out with CS.

Virtual desktops, scientific computing, big render farms. There's all kinds of things, they're not really for home users though. I mean, you could buy one but you're unlikely to be able to use it effectively. This is targeted at a market and computation scale that's much much larger than most people work with. People who buy these don't buy one at a time, they buy them by the dozens.

Not necessarily, in scientific computing cards like this are important. The biggest problem with GPU computing in general is the time it takes to copy from main memory to GPU memory and back. It really makes GPU difficult to work with and generally the gains in parallelization don't really pay off considering the amount of time it takes to make those memory copies. Being able to load more into memory and have it stay there is a big deal.