On 04/05/2025 19:36, Steve Lewis wrote:
But aren't modern ASICs essentially that idea of software-implemented-
in-hardware, mainly for dedicated performance?
I would think of it more as "hardware assisted software" - figure out
what parts are especially critical for performance and implement as
much of that in hardware.
If you had "perfected"
your software and accepted no code change was necessary, the coding did
exactly whatever it is you needed to do, you can then commit that to
some combination of hardware-logic-gate-stuff (for the sake of executing
it wicked-fast, just as modern specialized crypto-ASICs do)?
Sure. But while that works well for specialized algorithms typical
of crypto, it is hard to do for a more general case.
I just
had the impression Wang was doing some early form of this, as they
referred to their BASIC as "hard-wired." Or in other words, you can't
point to a single chip and say "there is the Wang BASIC ROM" (but I'm
speculating, hence the question to try to clarify on their pre-1974 systems)
If seem to remember that they were "hard-wired" in the sense that the
keywords were tokenized early, and transmitted to the CPU as tokens.
Julf
