I’m just wondering - the L14 is getting pretty long in the tooth (the CPU was released in 2019…) and, while I’m not planning to get any more Purism products in the near future, I’m curious if there’s a roadmap for a hardware refresh.
1 Like
The Librem 16 is officially acknowledged (there is even a category for it in this forum) but almost no details have been released officially.
I think Purism would be reluctant to release a roadmap, with or without dates on it, because then people would hold them to it, rightly or wrongly.
Whether “long in the tooth” is really a problem depends on what you need or want to do on the given computer.
The computer that I am sitting on at the moment has a much older, and more importantly less powerful, CPU still (i.e. way less potent than what’s in the Librem 14) but … it is sufficiently responsive doing the things that I need it to do … so I don’t feel the need to ewaste it just because it is old.
On the other hand, if I were working with multi-gigabyte LLMs, I might need something gruntier and something more modern e.g. with built-in AI model acceleration.
So, assuming that you currently have a Librem 14, only you can say whether it is in need of a refresh for your needs.
2 Likes
There are two minor versions of the Librem 14:
1 Like
All you need is a potent graphics card with as much as possible GDDR. Everything else is just a bonus (except you’re making money where “time is money” is true).
1 Like
I struggle with performance on my L14. I could increase RAM from 32GB to 64GB, but I’m pretty sure my bottleneck is the CPU.
I run Qubes and typically have four or five app vms running for different security domains (three different tiers of work access, a personal environment, and tor browser). It’s fairly usable right up until I hit a site with some required js that pegs the CPU on a given VM and locks it up entirely.
While I like minimalist computing and use a lot of older computers for single-purpose tasks, there are just some things you can’t do in a performant fashion with an older system. A machine you’re using as a hypervisor is always going to need more resources than bare metal.
2 Likes
Maybe this Evil Opensource?
1 Like
Can you be more specific?
Not specifically, but a lot of Microsoft stuff (O365, Azure Admin, etc), YouTube, Facebook, Odysee, Thingiverse. Any modern-ish site, the more content it has, and the more tabs I open, the more likely it is to cause a hang. If I had enough foresight to open top in another window, it usually is the Isolate+ process that starts grinding it to a halt. It isn’t browser-specific issue, either; it’ll do it in firefox-esr, Chromium, or Edge.
1 Like
For YouTube, you can use an (unlisted) Invidious instance instead:
There are no (maintained) privacy front-ends for any of the other websites you listed, but an alternative to Thingiverse is Instructables, of which I have a Structables instance:
See also:
1 Like
Some Qubes users say that they need more than 32GB RAM.
Run the xl info command in dom0 and read the total_memory field to see the total memory available to Qubes. Lack of RAM may be your bottleneck.
1 Like
Same link on my Redlib instance:
1 Like