Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
-
I don't see anybody moving over to ARM right now for desktop usage, besides Apple users. There is one simple fact that's gonna keep x86 alive: it works, proofen by decades.
Im not gonna start crosscompiling for ARM, my company has no intention of doing that, we don't even know if our software would ever work on ARM, and nobody else gives a flying fuck about ARM besides some Linux enthusiasts and people/companies with Apple overlords.
And why should we what is the big benefit of ARM? Faster Processor? Find me an ARM chip that can hold a candle to an Epyc 7003. Less power consumption? The Graphicscard is 90% of it, the CPU is the lesser problem. -
ARM is a revolution because it lets anyone build and customize market-relevant chips instead of just Intel/amd. And because it's a simpler system than x86 overall. That's about it. I'm no ARM fanboy but there does seem to be a lot of confusion about it.
- not all ARM chips are the same: correct. But ARM does have a lot of standard components because of the way ARM architectures are licensed (eg. if your processor is ARMv8 it largely works with any ARMv8 code) and compiler toolchains are fairly mature. Extensions exist but because of the modular nature of modern compilers, adding extensions doesn't make other things or vendors worse. Plus ARM is somewhat easier for compilers, being a largely simpler system. Most large chip vendors go for fairly compatible designs. But yes, fragmentation is a large issue with ARM that x86 doesn't have. But as in the case of Android vs iOS, fragmentation doesn't have to make something unusable.
- lot of interesting features that haven't been checked for security because they aren't as common as x86: true for new ARM designs and new features, but then every new microarchitecture has this problem (yes, even x86 when they change microarchitecture every couple of years). But otherwise there are an enormous number of ARM chips out there because phones, tablets, supercomputers, embedded systems, etc. all run ARM. I hardly think it's "less used". There's a ton of security work on ARM chips because of phones alone.
- ARM vs Thumb: afaik AArch64 mode can't use Thumb, so I'm not sure why this is a problem. Thumb isn't relevant to the side of ARM that competes with x86, it's usually found in embedded systems.
- backwards compatibility - agreed and this is a big problem. Apple has largely solved it with Rosetta 2, as a M1 MacBook Air user I can tell you the x86 emulation is very, very good. I expect other companies to follow eventually. For server software etc. recompiling isn't a big deal, the part that needs this is user facing stuff.
(Warning: This rant includes nonsense, nightposting, unstructured thoughts, a dissenting opinion, and a purposeless, stupid joke in the beginning. Reader discretion is advised.)
honestly the whole "ARM solves every x86 problem!" thing doesn't seem to work out in my head:
- Not all ARM chips are the same, nor are they perfectly compatible with each other. This could lead to issues for consumers, for developers or both. There are toolchains that work with almost all of them... though endianness is still an issue, and you KNOW there's not gonna be an enforced standard. (These toolchains also don't do the best job on optimization.)
- ARM has a lot of interesting features. Not a lot of them have been rigorously checked for security, as they aren't as common as x86 CPUs. That's a nightmare on its own.
- ARM or Thumb? I can already see some large company is going to INSIST AND ENFORCE everything used internally to 100% be a specific mode for some bullshit reason. That's already not fun on a higher level, i.e. what software can be used for dev work, etc.
- Backwards compatibility. Most companies either over-embrace change and nothing is guaranteed to work at any given time, or become so set in their ways they're still pulling Amigas and 386 machines out of their teeth to this day. The latter seems to be a larger portion of companies from what I see when people have issues working with said company, so x86 carryover is going to be required that is both relatively flawless AND fairly fast, which isn't really doable.
- The awkward adjustment period. Dear fuck, if you thought early UEFI and GPT implementations were rough, how do you think changing the hardware model will go? We don't even have a standard for the new model yet! What will we keep? What will we replace? What ARM version will we use? All the hardware we use is so dependent on knowing exactly what other hardware will do that changing out the processor has a high likelihood of not being enough.
I'm just waiting for another clusterfuck of multiple non-standard branching sets of PCs to happen over this. I know it has a decent chance of happening, we can't follow standards very well even now, and it's been 30+ years since they were widely accepted.
rant