One major issue is dGPU performance (this is also relevant for WoA).
That’s true. But that’s more a usability issue than an architecture incompatibility issue.
My compute module is a RK3588, which in the ARM world is considered high end, and it only compares favorably with an entry-level laptop of 5 years ago performance-wise - CPU or GPU. So with respect to price/performance, it’s not there yet. But… well, it works as well as an entry-level laptop of 5 years ago, which is good enough for my meager needs.
If you need performance, you’re limited to x86 because there is nothing equivalent in the ARM world yet. But that’s not because the ecosystem isn’t supported.
Note that the keyword here is yet: as more people buy ARM stuff, more powerful and cheaper ARM stuff will appear on the market. Chicken and egg.
lot of closed source software that cannot run natively on WoA
That’s why I said “at least in Linux”. And you’re right, I probably should have added “for open-source software”.
I haven’t used Windows since 95, and I haven’t used proprietary software for quite a number of years either, because the open-source offering is now good enough that I can credibly do all my computing without anything else. And so for my type of use, there is no significant hurdle running on ARM compared to x86.
But don’t think I’m in an ultra-minority: more and more people only use open-source software nowadays, and they’re not even power users anymore. It’s that good now. So I didn’t feel like mentioning it, because while it’s not quite mainstream yet, it’s common enough.
I use as much open source as I can (quality is generally better), but there are many line of business applications that don’t work well on WoA, let alone Linux (x86 or ARM).
For open source only it works pretty well. I’ve been using ARM SBCs since 2018 and it’s been great.
On desktop Linux/ARM, I am guessing there will be a lot of extra issues; drivers, problems with emulation of Windows games (independent of dGPU issues), limited support opportunities etc.
That’s true. But that’s more a usability issue than an architecture incompatibility issue.
My compute module is a RK3588, which in the ARM world is considered high end, and it only compares favorably with an entry-level laptop of 5 years ago performance-wise - CPU or GPU. So with respect to price/performance, it’s not there yet. But… well, it works as well as an entry-level laptop of 5 years ago, which is good enough for my meager needs.
If you need performance, you’re limited to x86 because there is nothing equivalent in the ARM world yet. But that’s not because the ecosystem isn’t supported.
Note that the keyword here is yet: as more people buy ARM stuff, more powerful and cheaper ARM stuff will appear on the market. Chicken and egg.
That’s why I said “at least in Linux”. And you’re right, I probably should have added “for open-source software”.
I haven’t used Windows since 95, and I haven’t used proprietary software for quite a number of years either, because the open-source offering is now good enough that I can credibly do all my computing without anything else. And so for my type of use, there is no significant hurdle running on ARM compared to x86.
But don’t think I’m in an ultra-minority: more and more people only use open-source software nowadays, and they’re not even power users anymore. It’s that good now. So I didn’t feel like mentioning it, because while it’s not quite mainstream yet, it’s common enough.
I use as much open source as I can (quality is generally better), but there are many line of business applications that don’t work well on WoA, let alone Linux (x86 or ARM).
For open source only it works pretty well. I’ve been using ARM SBCs since 2018 and it’s been great.
On desktop Linux/ARM, I am guessing there will be a lot of extra issues; drivers, problems with emulation of Windows games (independent of dGPU issues), limited support opportunities etc.