4 ms·

The road towards ARM is very rough for those who do not control the entire ecosystem around the hardware. Microsoft is on their second attempt, and they have m

by Alupis 2y ago

The road towards ARM is very rough for those who do not control the entire ecosystem around the hardware. Microsoft is on their second attempt, and they have much more sway over their ecosystem than Framework ever will. The magic Apple was able to achieve is almost entirely due to their iron-fisted control over their ecosystem. It's one thing to have a killer chip, but if nobody supports it, it's DOA...

candiddevmike 2y ago
> The road towards ARM is very rough for those who do not control the entire ecosystem around the hardware. Isn't that what Framework does? Framework creating an ARM variant with in-tree drivers and bundled with something like tow-boot out of the box would be a huge win for the community. The pinebook pro sucks and is the only real competition in this space unless you count chromebooks. > It's one thing to have a killer chip, but if nobody supports it, it's DOA The beautiful thing about Linux/OSS these days is you can typically assume if you have amd64 support, there's probably a working arm64 port, too.
- talldayo 2y ago
  > Isn't that what Framework does? Framework kinda serves as a glorified systems integrator. They need partners like Intel and AMD to work with them to make sure their chips are suited for the sort of chassis they're making. The recent RISC-V board was almost entirely a third-party offering as far as I can tell. So, I'd argue it's not really what Framework does. Someone like Nvidia, Broadcom or Rockchip would have to provide a particularly compelling chip that fits the form factor and power/thermal envelope, with potentially another middleman designing/integrating the chip into a compatible board. Making, upstreaming and maintaining those in-tree drivers is a commitment of it's own, and you'd have to validate the plethora of expansion cards/peripherals with a device that doesn't have UEFI or probably even PCI. Like the parent said; you'd need a sizable ARM license-holder to commit to the entire vertical process for it to release in a quality state. It's not infeasible that one of them might step up to the plate for this someday, but the incentives would have to line up for them before they'd consider making it a product.
  - snvzz 2y ago
    To put it in another way: ARM is late, as RISC-V is already available. There's no point in switching from a proprietary ISA (x86) to another proprietary ISA (ARM).

by Alupis 2y ago

candiddevmike 2y ago

> The road towards ARM is very rough for those who do not control the entire ecosystem around the hardware. Isn't that what Framework does? Framework creating an ARM variant with in-tree drivers and bundled with something like tow-boot out of the box would be a huge win for the community. The pinebook pro sucks and is the only real competition in this space unless you count chromebooks. > It's one thing to have a killer chip, but if nobody supports it, it's DOA The beautiful thing about Linux/OSS these days is you can typically assume if you have amd64 support, there's probably a working arm64 port, too.

talldayo 2y ago
> Isn't that what Framework does? Framework kinda serves as a glorified systems integrator. They need partners like Intel and AMD to work with them to make sure their chips are suited for the sort of chassis they're making. The recent RISC-V board was almost entirely a third-party offering as far as I can tell. So, I'd argue it's not really what Framework does. Someone like Nvidia, Broadcom or Rockchip would have to provide a particularly compelling chip that fits the form factor and power/thermal envelope, with potentially another middleman designing/integrating the chip into a compatible board. Making, upstreaming and maintaining those in-tree drivers is a commitment of it's own, and you'd have to validate the plethora of expansion cards/peripherals with a device that doesn't have UEFI or probably even PCI. Like the parent said; you'd need a sizable ARM license-holder to commit to the entire vertical process for it to release in a quality state. It's not infeasible that one of them might step up to the plate for this someday, but the incentives would have to line up for them before they'd consider making it a product.
- snvzz 2y ago
  To put it in another way: ARM is late, as RISC-V is already available. There's no point in switching from a proprietary ISA (x86) to another proprietary ISA (ARM).