2 ms·
It's the "too big to fail" scenario all over again. Just like with too big to fail companies and banks, only half of the equation is being applied: CPU indust
by Drakim 3y ago
It's the "too big to fail" scenario all over again.
Just like with too big to fail companies and banks, only half of the equation is being applied:
CPU industry wants to start using risky design techniques to get an edge on the competition? Who are we to tell them no? it's a free market.
CPU industry can't afford to replace the defects that came about from their risky choices? They are too big and important to be held accountable.
- nearbuy 3y agoYou're going to have to go back to the 1990s if you want to undo the "risky design technique" of speculative execution. You're ignoring half the equation too here. There are benefits to having faster and more efficient computers for the past couple decades that could execute more instructions per clock cycle. These vulnerabilities are subtle, hard to predict, and can take years or decades to discover. Anything that makes computers faster is a risky design technique. If you want safe chips, you'd better say goodbye to caches, multithreading, and hyperthreading as well. You'll be left with a slow, inefficient chip. Non-American companies that aren't forced to take zero risks would take over the market and be miles ahead.