3 ms·
If you accept the premise of the article, you also need to accept that assembly is not a low level language, and that it is impossible to program any CPU curren
by ori_b 3y ago
If you accept the premise of the article, you also need to accept that assembly is not a low level language, and that it is impossible to program any CPU currently for sale in a low level language.
The abstraction CPUs give you is more or less a fast pdp11 with some vector registers bolted on.
The implementation internally is not.
- variadix 3y agoThis is pretty much my issue with this article. By its criteria assembly isn’t low level, which makes the claim pretty uninteresting. You could argue there are/were processors that had low level ISAs (VLIW, no OoO exec, no memory reordering, no caches, no branch prediction, etc.) but they are all niche (usually low power embedded DSPs) or failed to capture the market (Itanium) because they failed to deliver performance comparable to the highly abstracted CPU supposedly “designed” to run C (also a questionable claim, I think the reality is it has more to do with sequential execution being the way humans think, a point Jim Keller has made in several interviews).