GlobalFoundries today announced a definitive agreement to acquire MIPS, a leading supplier of AI and processor IP. This strategic acquisition will expand GF’s portfolio of customizable IP offerings, allowing it to further differentiate its process technologies with IP and software capabilities.
↫ Press release about the acquisition
MIPS has a long and storied history, most recently as it abandoned its namesake instruction set architecture in favour of RISC-V. MIPS processors are still found in a ton of devices though, but usually not in high-profile devices like smartphones or whatever. Their new RISC-V cores haven’t yet seen a lot of uptake, but that’s a problem all across the RISC-V ecosystem.
I hope GF does something more with the MIPS ISA and IP. There’s a lot of developers and documentation out there for it (as there is for most established ISAs so it’s not hard to find developers.
RISCV is in exactly the same spot as many FOSS alternative OSes, like Haiku. No-one uses it, so there’s not so much incentive to develop and maintain software for it, so there’s not much incentive to develop hardware for it until the software matures. This is exactly the same reason it’s impossible to kill of x86. Too much software is available for it and works.
ARM is a good example as well. Whilst it has taken a good degree of time for it to become as successful as it has, the truth is ARM is old, and as such, there is, and was, a great deal of developers for it even 20 years ago. Niche, yes, but it was an established platform even before the smartphone/tablet boom.
RISCV just doesn’t have the market penetration at the moment. It might do one day, but i think it needs to organically grow as the other ISAs did. It needs to get established in embedded and IoT devices before we’ll ever see a successful RISCV based consumer device.
There are already quite a lot of consumer devices using RISCV. Most consumer devices are embedded, so the user is never exposed to the underlying architecture and usually isn’t even aware what it is. The same was always true of ARM. Architectures become successful from lowend to highend not the other way around – look at x86 and ARM.
MIPS was a wasted opportunity… When ARM was making the transition to 64bit a lot had to be done – compilers, kernels, documentation etc. On the other hand there has been a mature MIPS64 architecture since the early 90s already supported by multiple operating systems and compilers.
FWIW. RISC-V is pretty much the spiritual successor of MIPS. So in a sense the opportunity wasn’t as wasted.
MIPS is dead, long live LoongArch.