Linux Preparing Support for the RISC-V Framework Laptop 13 | Hacker News Reader

Linux Preparing Support for the RISC-V Framework Laptop 13

phoronix.com

43 pointsrbanffy24 comments

I really want to see more RISC and alternative processor devices, might have to buy one of these to push the drive towards that kind of future where we don't just have only ARM and x86.

Not just a third option, but one that's properly open and modular. I hate that the only ISAs that are feasible right now are proprietary, and in the x86 case, include backdoor big brother systems by default.

Not saying I don't believe it but do you have any links to back up that claim?

Intel: https://en.m.wikipedia.org/wiki/Intel_Management_Engine

Amd: https://en.m.wikipedia.org/wiki/AMD_Platform_Security_Proces...

ARM: https://en.wikipedia.org/wiki/Trusted_execution_environment#...

They are probably referring to IME in Intel and PSP in AMD.

They are essentially blackboxes that have control over everything in our PCs and nobody including the OS has any visibility into what they do under the hood.

What will the real-world performance of a RISC-V Framework Laptop look like?

My guess is not fantastic. Here's an older version of the board the article mentioned, along with some benchmarks comparing it an RPI:

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

It performs pretty poorly compared to even equivalently cheap hardware (~100 USD).

Performance depends on microarchiteture, not on price. It performs well compared to Arm boards with similar microarchitecture i.e. Pi 3.

Price is related to production volume. Raspberry Pi boards sell in large numbers and can thus be priced more keenly while retaining a profit margin.

We already have a bunch of SBCs with the same core, it's similar to the Raspberry Pi 3, probably a bit faster, but slower than the Pi 4.

See "VisionFive 2" review: https://youtu.be/ykKnc86UtXg

This is more up to date even though the board is much older.

At least as important to me, is what is battery life like? With a 61 Wh battery on what is essentially a mobile processor, can I get 20+ hours of use out of it?

A VisionFive 2 with the same chip uses 3.4 W idle, just with automatic down clocking (I think 375 MHz is the lowest). So that would be potentially 18 hours from a 61 Wh battery. But a laptop also has to drive a screen backlight. On the other hand, maybe the JH7110 has has some lower power sleep mode that was not being used by the software on the VisionFive 2.

Using the JH7110 chip they are starting with, somewhere around the last Pentium 3s and PowerPC G4s, but quad core instead of single core. And 64 bit. So maybe the closest comparison is the original Core 2 Duo MacBook Air, which ran at 1.6 MHz (for about 4 seconds before it throttled to 1.2 GHz, which the RISC-V shouldn't do).

Or, in the Arm world, a bit slower than a Pi 4, but a lot better than a Pi 3 (especially because 8 GB RAM and PCIe/M.2 not 1 GB and SD card).

I'd expect a 2nd generation motherboard fairly quickly. If it comes in the next six months (Eswin EIC7700) then probably about the speed of a mid-range Core 2 Quad. If mid next year (SG2380) then early i7.

Does RISC-V have ACPI and UEFI?

ACPI[0] and UEFI[1].

0. https://github.com/riscv-non-isa/riscv-acpi

1. https://github.com/riscv-non-isa/riscv-uefi

So in other words, the answer is: "no (but maybe someday)"

Refer to the ratified specifications[0]. Grep for ACPI or UEFI.

What this means is that RISC-V's work is done.

Compliant implementations exist as well.

0. https://lf-riscv.atlassian.net/wiki/spaces/HOME/pages/161547...

Not done until that github project has a non-boilerplate README, IMHO.

I don't know who Linux is, but I hope he succeeds

I think Linux is a pretty cool guy, codes operating system and doesn't afraid of anything.

Including showing a well deserved middle finger to unfriendly graphics hardware manufacturers. This surely sends a message that warms hearts in the FOSS community, but also closes some doors where money counts, and I wish more people in the professional IT world had the same balls.

He's this guy.

https://invidious.privacyredirect.com/watch?v=x7ozaFbqg00

I too don't know who Linux is or if it's a he but apparently they don't like it if you don't use the right terms[1].

[1]. https://docs.kernel.org/process/code-of-conduct.html

Reply on news.ycombinator.com