Samantha Atkins wrote:
> A lot has happened in hardware in the last couple of decades.
In terms of raw speed, yes. In terms of return on transistor
investment things are not looking nearly as nice. A redesign
is in order.
> Unfortunately, not nearly as much has changed in software.
The problem is an ouroboros lockup, the software/hardware
architectures mutually lock themselves in bad local optimum.
Mainstream progress has essentially stalled. Fierce
competitiveness, short investment cycles have made experimentation
in new architectures (e.g. diving control flow into many
lightweight asynchonous threads, runtime reconfigurable
architectures) extremely risky, so players are not waging
much. Anything so dramatically new that existing programming
tools and skills infrastructure are not applicable to it is
essentially nonviable.
I guess this means we'll have to wait for the curve to nivellate,
until investments in new developments become again viable.
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:59:46 MDT