2 ms·
Huh? RISC-V with the compressed extension is incredibly efficient in its encoding. Better than x86 or ARM in both static and dynamic bytes per program. Also,
by _chris_ 10y ago
Huh?
RISC-V with the compressed extension is incredibly efficient in its encoding. Better than x86 or ARM in both static and dynamic bytes per program.
Also, Icache pressure is a huge problem in modern warehouse-scale computers.
Any processor that cares about performance will almost certainly be implementing the C extension to RISC-V. It also enables more efficient macro-op fusion, turning common two instruction 4-byte idioms into a single, more powerful instruction.
- microcolonel 10y agoThanks for going into more detail. I was basing my assumption that it wasn't a huge problem on the fact that the only people who complain about it first seem to the folks designing the Mill. They have a ridiculous/insane/cool solution to it. Everyone else seems to first mention their cool branch predictor, or vector processor.