Linux-libre on POWER

5 risposte [Ultimo contenuto]
jxself
Offline
Iscritto: 09/13/2010

I've updated my APT repository for Linux-libre; https://jxself.org/linux-libre to include support for IBM Power CPUs (POWER8 and above);
like those machines from Raptor Engineering: https://www.raptorcs.com/TALOSII/

They have the Talo II Lite motherboard for pre-order, which is "only" $1,100. That's almost affordable. Of course you also need to factor in the cost of a CPU (the cheapest is the 4-core version for $375) as well as a case (it's a big board so you need something to support EATX), as well as a power supply, RAM, and etc.)

The Parabola people have support for this CPU architecture in progress:
http://lists.nongnu.org/archive/html/gnu-linux-libre/2018-06/msg00017.html

Now if Trisquel can get support for something (anything) that's not x86 happening we'll be on our way.

It's time to start evacuating the Titanic: https://jxself.org/titanic.shtml

GrevenGull
Offline
Iscritto: 12/18/2017

Did you single-handedly code the linux-libre kernel to work with these CPUs?

That's amazing.

jxself
Offline
Iscritto: 09/13/2010

No. The kernel named Linux has support for lots of machines:
https://en.wikipedia.org/wiki/List_of_Linux-supported_computer_architectures#Releases

GrevenGull
Offline
Iscritto: 12/18/2017

Right, are you working together with fsfla?

marioxcc
Offline
Iscritto: 08/13/2014

>Now if Trisquel can get support for something (anything) that's not x86 happening we'll be on our way.

But supporting architectures for which machines are not easily available (like POWER) does not add much value for users. OTOH, ARM is more widespread. RISC-V only has FPGA implementations and extremely expensive AND underperforming (for their cost) custom chips.

jxself
Offline
Iscritto: 09/13/2010

You're right that it's not necessarily the only solution.

Your description of RISC-V sounds like as of today. I find it helpful to not look at how things are today but how we want them to be a decade from now. Because that's probably how long it will take to evacuate the Titanic. And in that time RISC-V will have matured. From that perspective I find it to be an interesting candidate and am keeping an eye on it.

ARM is too, especially if the EOMA68 ever starts shipping. I continue to keep an eye on things like that as well.

While I'm trying to evaluate things from the perspective of software freedom I do acknowledge that POWER is expensive. Today. But keep in mind that I'm trying to look at things on a decade-long timespan and history has shown that the price of a given technology decreases as time moves on.

But, even if you want to look at how things are today, I feel I also need to point out that POWER is not necessarily more expensive or less available than the other high-end computers of today.

My primary reason for doing this though is to call attention to the iceberg that we've hit, not to say what road to take. Indeed; it's probably better if we don't all go down one road because there's no way to know what is going to work out well for software freedom in the long term. Maybe it will be ARM. Maybe POWER. Maybe RISC-V. Maybe some combination. Maybe something else entirely that doesn't even *exist* right now. We've already hit the iceberg though. Indeed, as we go down by the head some might say that the water level on the boat deck at the front of the ship is already up to our knees. Many people standardizing on Intel has helped create this problem. Splitting up helps improves our odds of successfully getting out of it in this decade-long timespan.