2 ms·
Much the same promise is made by VLIW today - it should let you "jack up" core counts and clock rates compared to even RISC, because you don't need to do energy
by zozbot234 5y ago
Much the same promise is made by VLIW today - it should let you "jack up" core counts and clock rates compared to even RISC, because you don't need to do energy-intensive OOO reordering and speculation so it becomes feasible to avoid thermal and power constraints even with otherwise heavy CPU use. (The clearest issue with it is that it leaks way too much of your μarch choices in the programming model, so every ISA is inherently bespoke.)