X * f(x) semantics
-
only really the supercomputer people will really use
IIRC, the chip was designed to support up to 8 CPUs — not really supercomputer, more like high-end (for its day) servers and workstations. The testing was, of course, stressing the cache controller far more than any real system would in normal use.
-
IIRC, the chip was designed to support up to 8 CPUs — not really supercomputer, more like high-end (for its day) servers and workstations.
Hah. I was thinking more about the 1024 core monsters we had. All those fancy backplanes and interconnects were not cheap…
-
1024 core monsters
Nope, never worked on anything like that, just mainstream server chips. Although I did apply for a job recently at a company that does computational molecular biology and builds their own supercomputers to run the computations. (Nothing came of it. The recruiter told me — after the fact, of course — that he'd never had one of his referrals even get an interview.)
-
computational molecular biology
Computational chemistry gets really complicated. They like to do things like predicting just how molecules interact and react so they can tweak stuff like drugs to make them work better. Which basically means simulating these things at the level of the quantum mechanical fields (so you can get fairly accurate predictions of what all the electrons are doing). Since you're working in solution inside the body, you'll also be simulating quite a few (thousands of) water molecules.
Solving the Schrödinger equation for that sort of thing (well, numerical approximation to) at that scale is pure evil for sheer quantity of compute power needed. I don't know if they can do protein-protein interactions yet; I know we're not doing it in our current project, and are instead sticking to simple approximations that only need a conventional compute cluster. ;)
-
You mean to say that the computer simulation of me failing to write good code is still a decade away? I was hoping to retire soon!
That reminds me of this ridiculous calculation of what professions will be replaced by computers. Computer programmers have a 48.1% chance of being replaced; with just 22.5% for hardware engineers It's scientific because there are three significant digits! If you're still doing hardware you might want to switch so you can retire sooner!
you'll also be simulating quite a few (thousands of) water molecules.
Maybe if I eliminate water from my diet I will be easier to replace?
-
Maybe if I eliminate water from my diet I will be easier to replace?
Dead people are all pretty interchangeable as far as on-the-job performance goes, I suppose.
-
So the question reduces to are computers simulating dead people a good replacement for dead people?
-
It's probably easier and cheaper to just use dead computers.
-
Computer programmers have a 48.1% chance of being replaced
We replace entry level programmers all the time. But we do that by giving them better tools so they can do more, not by eliminating them.
-
That's not high-level, that's spaghetti-level.