Re: "A diagram of C23 basic types"

On 16/04/2025 06:48, BGB wrote:

On 4/15/2025 5:56 PM, Lawrence D'Oliveiro wrote:

On Tue, 15 Apr 2025 00:40:48 -0500, BGB wrote:
>

Practically, picoseconds are likely the smallest unit of time that
people could practically measure or hope to make much use of.

>
“10¯¹² seconds ought to be enough for anybody.”
>
The lessons of software backward-compatibility baggage teach us that we
need to think a bit beyond present-day technological limitations.
>

 In all likelihood, computers will not get much faster (in terms of clock speeds) than they are already.
  If things were able to get much faster (without melting) then more fundamental rethinking would be needed about how things work, as clock pulses could no longer be used for global synchronization, and (going further) an inability to pass signals through metal wires.
 

Lots of communication goes several orders of magnitude faster than 3 GHz.  Fast processors, and especially switch chips, have clocks that are far too fast for global synchronisation.  At 3 GHz, a clock signal can travel about 6 cm per clock pulse - you can't have a 3 GHz global clock on a big chip.  This issue has been solved long ago, using transmission lines and appropriate buffered and registered communication between parts of chips.