I heard that the alzheimers brains were preserved in a solution containing aluminum. The healthy brains were not. But this is a bit different. They claim that their model has been used to reverse some of the symptoms of Alzheimers. That's a stronger claim. It's still a small study. There will be at least a few cases that don't fit the model. But it seems well designed and worth taking note of.

I agree about the coffee quality. However, Starbucks isn't really in the business of selling coffee. They sell space. You really buy a break in a pretty comfortable, relaxed environment; a living room for rent. The overextracted cardboard mug of coffee is just a bonus.

The Shinkansen typically runs every 15-20 minutes or so. On the busiest lines (Tokyo-Osaka) at peak times there's a new train every five minutes. You just show up, get a ticket atthe vending machine and step on to the next train.

Flying may be cheaper, but the trains are just so much faster and more convenient. I love them.

Even if you could create a version of Windows that would run well in a cluster environment, the applications aren't there. One reason Linux is so dominant is that the kind of code you want to run on a big cluster is all written for Linux.

With that said, some HPC apps are hybrid; you run a client on your local machine (Windows app, or a web app) that dispatches the work on to the cluster in the background. With those you can treat the cluster as a big accelerator for your local app.

Todays supercomputers and clusters are all descendants of the Beowulf clusters. No need to imagine anything :)

I agree low birthrates is a problem (although the reasons are principally economical rather than social). I live in Japan and see this first hand.

Low birth rates is not the cause of the rural depopulation, though. That has been an ongoing trend since long before the population stopped growing; and it's a trend in countries whose populations are stable or still growing at present.

This is really about urbanization, not population changes, and it's not limited to Japan.

The issue is that young people leave rural areas - for school, higher education, jobs - and don't return. Cities grow while rural areas shrink, and eventually the population becomes too small and too sparse to support a good range of public services. Similar things are happening in Europe and in north America as well.

Apart from having to rewrite existing code, one issue is that in some fields (HPC and supercomputing) the facilities can be very conservative about what they install. In many places, they install whatever is the conservative, safe default when the system is built, then they never update it. If you do offer newer versions of something, you add it alongside the existing software, not replacing it. When the system us upgraded or replaced, you make very sure to add (or backport) the old versions for the new system.

A major reason is that research projects - that can go on for 5-10 years - don't want to switch software versions mid-stream. If you are comparing an analysis of your current data with data from five years ago, you want to be sure any differences is due to the data, not because you changed software versions somewhere along the line.

This means that many systems may not offer Python 3 at all; or if they do, they still point to Python 2.7 as "the" python, and they will until the system is decommissioned years and years from now. And that means a lot of scientific software still primarily (and sometimes only, though that's becoming rare) target Python 2, since that's where a lot of their users are.

Python 2 will in practice live on for at least another decade, and quite likely for longer than that. I do agree that new projects probably should seriously consider using Python 3, but Python 2 disappearing will not happen.

No, more than half of the machines this year were using Infiniband. I get the impression (note: not hard data) that IB is pushing out 10G Ethernet on the lower end of the HPC field. The latency wins are worth it for a lot of applications.

Infiniband is the most common interconnect in the HPC space today though. Doesn't seem right to say it's a failure when it dominates the segment it was designed to handle. Or do you mean specifically the Intel implementation of it?

That's not what "brick" means.

Words mean whatever its users decide they mean. You are aware, I hope, that "brick" didn't mean anything related to computing, right? Somewhere there's a bunch of annoyed bricklayers fuming about how the nerds are using the word all wrong.