Simple answer is that there’s still nothing comparable to M architecture. It has the benefit of unified memory and SoC which eliminates the bus. And being RISC, it has small fixed size instructions which can be read in batches, then any instructions that don’t have a dependency within the batch can be executed in parallel. This makes it possible to just add more cores and parallelize against them. You can’t do this with CISC because instructions are all different length, and figuring out the dependencies and what needs to be read into a batch becomes expensive really fast negating the gains of running stuff in parallel.
I know how and why RISC is a better design (I wrote an x86 C compiler before, ugh), but how come there is still no one else making great RISC PCs/Workstations that are relatively affordable and don’t suck? I personally would kill to get a decent RISC PC from a non-BigTech Company that comes with a Manual for Users and Programmers and has all of its chips documented and I can just install/compile any OS I want, or even write my own.
Some chinese companies could easily kill the proprietary tight-lipped bullshit companies that way. And I really hope someone does at some point. I really don’t want to enter a future where we have technically excellent RISC Computers, but they’re all locked down and everything is so proprietary that you can only run Verified™ software on them, and the only online documentation you can find just tells you to not eat the batteries and don’t shove the power plug up your ass, etc.
Simple answer is that there’s still nothing comparable to M architecture. It has the benefit of unified memory and SoC which eliminates the bus. And being RISC, it has small fixed size instructions which can be read in batches, then any instructions that don’t have a dependency within the batch can be executed in parallel. This makes it possible to just add more cores and parallelize against them. You can’t do this with CISC because instructions are all different length, and figuring out the dependencies and what needs to be read into a batch becomes expensive really fast negating the gains of running stuff in parallel.
I know how and why RISC is a better design (I wrote an x86 C compiler before, ugh), but how come there is still no one else making great RISC PCs/Workstations that are relatively affordable and don’t suck? I personally would kill to get a decent RISC PC from a non-BigTech Company that comes with a Manual for Users and Programmers and has all of its chips documented and I can just install/compile any OS I want, or even write my own.
Some chinese companies could easily kill the proprietary tight-lipped bullshit companies that way. And I really hope someone does at some point. I really don’t want to enter a future where we have technically excellent RISC Computers, but they’re all locked down and everything is so proprietary that you can only run Verified™ software on them, and the only online documentation you can find just tells you to not eat the batteries and don’t shove the power plug up your ass, etc.