Microsoft security essentials for the free one.

But you don't need a network exploit to worry. Secunia PSI routinely pointed out buffer overflows in how .mid files were handled in Winamp, prompting for an update. Now, I don't play .mid files, but if someone discovers an exploit like that in, say, how Winamp handles m4a or ogg files, with no dev to fix the security hole, how can you be sure your existing installation is not used to pwn your machine the next time you play a media file from teh internetz? Winamp hasn't changed much over the last few years, but at least they were patching security holes.

A factor of 1000 snuck out on you here. You went from milliohms to ohms in a second.

Only blue-footed boobies.

"Just enable" works when the 'make' you execute is the only one. In a maze of scripts and Makefiles and scripts generating Makefiles and scripts generating scripts that call make, euphemistically called "a custom buildsystem", hunting down the right 'make' to change into 'make -d' can take a while.

Seconded. I use a 65GB git repo across four machines (two of them virtual) to sync my entire work history on a daily basis. Occasional 200MB files in there, but mostly smaller stuff (source, binaries, PDFs, word documents, datafiles). Never had a problem so far. Admittedly, no branching, just keeping a few hundred thousand files in sync across four machines so that I don't have to worry where my files are when I go somewhere with my laptop only.

Hitler ate sugar!

However, they have no way of passing this knowledge to their offspring, because we do not select them for human scrutiny. Unless you agree with Lamarck.

Yes, "well-known algorithms" is my use case -- massive LAPACK generalized diagonalizations that take forever on a single CPU, almost forever when threaded with openMP-capable BLAS to, say 8 cores, and do not scale at all to distributed-memory clusters (ScaLAPACK with MPI) because the comms becomes a bottleneck.

Thus I'm hoping for a solution where the vendor themselves wrangles those intrinsics in their BLAS or LAPACK implementation in MKL with me oblivious to all that mess. Assuming the computation time scales O(N^3) and the memory transfer over the bus scales O(N^2), with the prefactors in my favour, I should be able to squeeze out a significant performance boost.

Excellent. Thank you.

Does Intel's MKL support the Phis out of the box? It would be very convenient if, instead of having to re-write code for them, we could just use phi-capable BLAS and LAPACK.

Through the Baldwin effect, presumably.