I'm pretty sure there are ways to constrain the range of possibilities. One obvious thing is that no life forms will most likely be based on xenon or gold because these elements don't really form the same kind of a wide range of interesting compounds that carbon does. The laws of physics (and chemistry) are the same pretty much everywhere, and just because our brains (and computers) are incapable of reaching more significant conclusions on this issue at this very moment doesn't mean that it's going to stay like that forever.

But is that even important? Caches themselves introduce execution trace non-determinism (certainly in the presence of interrupts and multitasking) because you don't know in advance if the memory reference is going to hit or miss. The program can't rely on timings in those cases (it's my understanding that this is why Cortex-M for real-time control doesn't have any caches at all - if you really have to care about worst-case behavior, trying to improve the average case may be pointless for many applications).

I like how you wrote all these things with a straight face. Perfect deadpan!

That's all fine and dandy, but the technological limitations of Flash memories put this in the "not quite there yet" territory when it comes to non-volatile RAMs. You wouldn't want to put your plain old in-memory data structures into that thing, so we're not quite there yet when it comes to unified memory architectures.

Actually, it's been the convention in Indo-European languages for a millennium or so; in fact, the structure of many of these allows for no other choice (Slavic languages, German, French etc.), and many English speakers - if not most - are L2 speakers for which this is the most compatible and the only natural alternative (the "uncanny valley" subproblem of interference in interlanguage fossilization). As a bonus, with the rapidly increasing number of perceived and recognized psychological genders (not the linguistic ones) on the very short timescale of the few recent decades, there's no need to rewrite texts and textbooks (considering that the purpose of written texts is to span not only vast amounts of space but of time as well); one could easily argue that "improvements" of limited scope such as your "he or she", while attempting to sound inclusive (for whatever strange reason some people might perceive it that way) are, for example, distinctly interphobic - which, again, would be a social construct with limited longevity compared to the potential timelessness of any written text.

And how do you propose that a machine like this should work? By poking into everything, or by giving it a brain we don't have?

So humans won't be able to repair them, because you're not going to send them, and you're arguing the case for machines repairing themselves is equally bad? So what *is* your proposed solution? The "disposable camera" model? That really doesn't scale very well.

(It is also interesting to observe how the "notoriously unreliable" human body of yours actually deals with some space-related conditions such as moderately intense radiation actually better than the majority of our technology, which is prone to hard errors, and if scaled to the cognitive capacity of the human brain would fare even worse.)

A Cardassian detainee, I suppose?

I'm a speaker of an Indo-European language; do the math. Regarding the rest, well, then replace the geologists with mission planners of any other kind, capable of deciding into which rock the machine should poke next time. Or devise a way to make the machines maintain themselves far away from Earth, because that's another thing humans would be able to provide on site and current technology doesn't.

Don't ask in the headline! You've ruined it for the scientists; now it can't be tested scientifically anymore.

That's the former, not the latter, but OK. (I also said "in many places", one would have thought it obvious that these things sort of trickle down from the top over time, especially given the initial limitations on the technology.)

I don't dispute the possible savings and their value on large scale, but in general, it seemed to me that these proposals (what TFA describes) covered inter-application interactions, and not intra-application performance management. That's what I had in mind. With application-dedicated nodes (in cloud systems), improving performance is still of paramount importance but you do that with better data structures, careful application design, basically using internal domain knowledge etc., not with some some sort of app/OS generic resource allocation protocols. Or did I miss something?

That reminds me of the (rejected) compromise that suggested that we index arrays starting with 0.5. :)

Honestly, in MVS (z/OS), it probably makes perfect sense to have this in an OS, especially if you're paying through the nose for the hardware already. But solving it on the VM level surely makes it a huge win for everyone.

When you go cloudy, you can do the same things on a somewhat higher level. As in, when you go Google-sized, the allocation and management of resources with a granularity of a computing node doesn't probably bother you much, because you have tens or hundreds of thousands of them. Trying to solve these problems on the single system level might be a waste of time for many applications. This is more of a problem for on-site big iron. It's an interesting problem, and if solved, could be of use to many people, but it would be much less useful for cloud providers.

Uh, no. PCIe SSDs are just coming into regular use in many places, and I haven't even heard of non-volatile memories being on the market (GB-sized, mind you - not tiny FRAMs for embedded applications).