3 ms·
Is RISC V the Future?
- avianes 5y agoThe first answer is very poor and weak. > It still concentrates on fixing those problems that we had in 1980s (making instruction set that is easy to pipeline with a simple pipeline), but we mostly don’t have anymore This criticism is highly uninformed. RISC-V is designed for modern superscalar processors not simple scalar pipelined processors: instructions are kept simple with little side-effects to allow simpler decoding and dynamic scheduling, instructions are designed for macro-op-fusion, C extension is designed to allow these two points above while providing a very good code density, etc. You can argue that RISC-V's choices are bad, but you cannot argue that they are not addressing modern problems. > ARMv8 is much more advanced and better instruction set which makes much more sense from a technical point of view. Should we recall that the RISC-V ISA is still under construction? Obviously, don't use RISC-V today if you want high performances. The recent ratification of several spec's is a step in the right direction but there is still a long way to go Discussions here: https://news.ycombinator.com/item?id=29417885 https://news.ycombinator.com/item?id=29417885 > Many common things require much more RISC-V instruction than ARMv8 instructions. There are many discussions in other HN posts that address this issue. Discussions here: https://news.ycombinator.com/item?id=29420622 https://news.ycombinator.com/item?id=29420622
- znwu 5y agoTo be fair, I dislike a lot of stuff in RISC-V. But there are just so many who won't even bother to read through the spec and start commenting. I personally considers the complexity of designing a microarchitecture-neutral ISA is beyond comprehension of any single human being. You could be a legendary* microarchitecture designer / compiler developer / application developer, but no one can be all at the same time. * By legendary I mean across multiple domains, from low-power to high-power to server, from OS to CRUD to HPC. So those kind of one-liner dismissal is really pale when RISCV spec is full of design rationales. IMO, it should at least include both: 1. Show that the rationale does not exist in spec / rationale is wrong 2. Show that your counter-point is significant enough (used widely and cause a bottleneck) Otherwise it becomes a joke for those: 1. who complain code size 2. who think add is better than addi for no-op 3. who is the author of an arbitray-precision numerical library and claims RISCV is shit.