Linux Preparing Support For The RISC-V Framework Laptop 13

[bachchain]

https://frame.work/au/en/blog/introducing-a-new-risc-v-mainboard-from-deepcomputing

"This Mainboard is extremely compelling, but we want to be clear that in this generation, it is focused primarily on enabling developers, tinkerers, and hobbyists to start testing and creating on RISC-V."

They made it pretty clear who the target market was. Purely for tinkering with RISC-V, not for practical purposes.

The problem with that reasoning is the CPU already exists in its own dev board, for cheaper, without carrying the expectations of an actual generally usable product. To be clear, this isn't an asinine waste of resources because it's an underpowered test platform, it's an asinine waste of resources because it's an underpowered test platform enclosed in a thousand dollars worth of laptop components it has no hope of driving properly

[mechkbfan]

The problem with that reasoning is the CPU already exists in its own dev board, for cheaper, without carrying the expectations of an actual generally usable product. To be clear, this isn't an asinine waste of resources because it's an underpowered test platform, it's an asinine waste of resources because it's an underpowered test platform enclosed in a thousand dollars worth of laptop components it has no hope of driving properly

But that's it's own dev board. That doesn't help writing drivers for RISC-V in a Framework 13 does it?

No expectations were set of a usable product

I would assume people that get this already own a Framework 13, and are intending to swap motherboard over to RISC-V, play around, write drivers, etc. then swap back

"Driving properly" clear isn't in the scope. If the product isn't for you it's fine. Let others do the work of writing drivers, getting basic functionality going, then wait for next generation to come out that probably has the performance of a Raspberry Pi then decide if it's for you.

[grigi]

This is not a regular SBC board replacement.

Note that RISC-V is suffering from many fundamental issues. Some examples:
1) The software isn't optimized for it yet
2) The hardware hasn't been around long enough to develop fixes for areas that have bad performance (e.g. instruction X is fundamentally flawed ito performance, let's introduce replacement instruction Y)
3) Dev-boards tend to have limited hardware connected, so you don't know if e.g. the various subsystems's validation processes are adequate without something that can have a direct comparison.

This device is really about exploring issues 3 and then 1 and 2 to a lesser extent. It provides a fully configured platform that can share identical components with an existing mature system.
So you can know that if e.g. the panel is flickering, if it's a fault on the display or the motherboard. You can literally connect the identical display to an existing, working, intel/amd motherboard and validate it's performance and quirks.

Is it something that interests me?
More that it exists and will allow ~100 people worldwide to look at this hardware from a usability PoV.

Does it mean that it allows faster iteration?
Yes, only the motherboard needs to be revised.

Will it be useful at some time?
Possibly, in a few iterations when the fixes have been "upstream" for a while, and performance is much closer to parity.

[nintendo1889]

I've been following haiku development. One thing that I've learned on their forum is that riscv does not use flat device treesnas arm does. So a riscv port of an operating system is much easier versus arm. The device tree doesn't need to be tweaked for every individual board and chipset. Hopefully this will change as arm eyes datacenters and AI.

[Subjunctive]

It could be compelling if it's priced right but it's Framework so it won't be.

they offer the best pricing on modular, upgradeable and repair-friendly laptops I know of. who do you prefer in terms of value?

(I've upgraded my original Intel-11th FW13 enough that I just ordered the remaining parts needed to build a complete one for my daughter, which is going to be a very fun project for us I think)

[gnwiii]

https://www.phoronix.com/review/visionfive2-riscv-benchmarks

In short, have zero expectations and you'll still be disappointed. I'm still baffled they ever thought this product was worth the time and money.

There are use cases were power consumption is a primary concern such as work in remote locations checking monitoring devices, capturing data, etc. where performance is not an issue. Some of these currently rely on dedicated hardware, so you end up having to haul and charge multiple devices and lots of D or C-cells. A low-power device with flexible interfaces could replace multiple aging dedicated battery hogs.