Video download link | md5sum 0d23ef8bd6577dacbd01f9526968ecc2 Going Back to Basics and Low Power Usage Creative Commons Attribution-No Derivative Works 4.0



Summary: There's a lot to be said about what the "end of an era" for x86 would mean not only to GNU/Linux but also the hardware scene; to accomplish and complete key tasks we've long had sufficient computational power

Now that Microsoft is circling down the drain along with the x86 patent pool (major collapse in sales of hardware an thus a collapse in Windows licensing) it's worth discussing a more desirable state of affairs. We've already peaked in terms of performance and we have far more computing power than tends to be needed, especially on the client side. Dr. Andy Farnell wrote about it last week.

At the moment, hardware is very highly complex, super-proprietary, very energy- or power-hungry, heavily patented, and barely reliable due to workarounds and 'cheats' (like Dieselgate for benchmarks). It's time to consider hardware freedom and go back to basics, or a level of simplicity that makes auditing hardware actually feasible. Someone has noted in IRC that "it's also fairly straightforward to port (assembly) code from one risc architecture to another" and potability is another thing we stand to benefit from.

The race towards higher speeds by making more and more transistors (and then cores) led to a certain unreliability and a lack of trust in hardware, much of which gets manufactured abroad. For real security and for the feasibility of education we need to at least consider another paradigm. ⬆