News for hackers. Stuff that matters.
Re:Librem 5 v2: A phone that frees us all by fsflover on Friday November 13, 2026 @10:47AM (#50485238)
> has every order been fulfilled yet?

Yes, and it's mentioned in TFA.

> the slowest phone you could spend money

https://puri.sm/posts/the-danger-of-focusing-on-specs/

> ANCIENT Snapdragon 4s Gen 2 you can get in a $20 flip phone from Walmart Right Now has performance cores

And you will not control your main computing device, https://news.ycombinator.com/item?id=47367568

Re:Librem 5 v2: A phone that frees us all by bigfatkitten on Friday November 13, 2026 @03:31PM (#50489224)
> And you will not control your main computing device

If the device isn’t useful, what’s the point?

Re:Librem 5 v2: A phone that frees us all by fsflover on Friday November 13, 2026 @05:21PM (#50490854)
It is useful. I use it as a daily driver.
Re:Librem 5 v2: A phone that frees us all by nonesuchthing on Friday November 13, 2026 @08:44PM (#50492706)
Dawg I am literally an FSF associate member, I own a Talos II, I port phones to mainline Linux, I had their sticker on my car!

On your first point: Reading around (you can find a lot of these experiences on HN itself), it seems pretty obvious that order loads reduced later in the Librem 5's lifetime due to chargebacks, allowing them to fulfill all remaining backer orders. Even if they did, why the fuck would I want to play that roulette again! Receiving a phone with 2015 specs in 2023 because you expected it several _YEARS_ earlier isn't any sort of consolation. I don't even necessarily like Pine64's approach to hardware development and the Pinephone has similar problems but at least those ended up in people's hands way, way sooner.

On the rest: The specs are worthy of complaining about, you freetard. What point is any phone, even a libre one if I can't reasonably expect to do anything useful with it as a computing device! 8 slow ass in-order cores aren't going to handle even basic tasks like web browsing well anymore (judging by my MSM89x7/MSM8953 mainlining experiences). So you get your freedom phone in 2028, the bloat isn't ever getting better, and the result is a blob-less dumbphone that can't accomplish any of the tasks a phone is expected to be capable of, plus you spent flagship money on it!

We're not even on step one of freeing phones, which is most phones being capable of running user-modifiable operating systems that have reasonable user freedom. This is "cool" but really a waste of money unless you're the very specific kind of person that values "zero blobs" over "literally anything else". What good does freedom do me if it can't spend?

And that $20 flip phone? That's a TCL Flip 4 5G. It's root-of-trust is the AOSP test keys, so you can run whatever the fuck you want on it provided you're willing to port it.

(insert gahnoo/car meme here)

Re:Librem 5 v2: A phone that frees us all by Hackbraten on Friday November 13, 2026 @11:45PM (#50494025)
I also feel completely puzzled about their choice.

I've been daily driving the L5 v1 since 2023, and while it's true that performance has improved since then, it's still a huge regression when you come from any flagship or even mid-tier phone.

Examples:

1. Can't scroll on some websites that use CSS effects because they render at ~ 1 fps so you can no longer tell where your scroll position is.

2. The browser keeps alerting me that my password manager extension is slowing it down. (It's never done so on other phones)

3. The Signal app (Flatpak, based on Electron) has a startup time of several minutes.

I'm now used to it but come on, Purism is going to pick a SoC for its next gen with less memory bandwidth (and possibly weaker single-core performance) than the i.MX M8 Quad?

The only plausible explanation I've been able to Stockholm together from all this is that some of the engineers at Purism must have vastly different phone usage patterns than I do.

Re:Librem 5 v2: A phone that frees us all by seba_dos1 on Saturday November 14, 2026 @07:32AM (#50496523)
Ad 1 & 2: Are you using Firefox? WebKitGTK is much snappier. Firefox doesn't even use the GPU at all unless you run Mesa 26.3 which was released last week.

Ad 3: That doesn't sound like a performance issue, there must be something weird going on as Electron apps usually launch within seconds for me (never used the Signal app though). Are you perhaps launching it from the SD card?

Re:Librem 5 v2: A phone that frees us all by Hackbraten on Sunday November 15, 2026 @06:26PM (#50511994)
Yes, I use Firefox. I’m already having a hard time due to the exotic device (Akamai’s WAF flags it as anomalous and suspicious, which means no more public transport for me). I’m genuinely afraid using an even more niche browser would make this even worse. Even if I changed browsers, I’d lose both my password manager and uBlock Origin.

Not sure if you actually browse without a password manager and without ad block (which would be fair but I simply can’t do that), or I’m still missing something.

Yes, I offload Flatpaks because my /usr, /var and /home (only the parts that live on the eMMC) are 19 GB, 2 GB, and 3 GB (83 % full). I’d love to bring it back to try it out but there’s literally no space, and I think I already have offloaded as much as possible. Can’t offload installed system packages :(

Are microSD cards slower on the L5 v1 than elsewhere? If so, then TIL; they’ve been snappy enough on all my computers/Raspis. What bottlenecks does the microSD reader have other than USB 3.0 bandwidth?

Re:Librem 5 v2: A phone that frees us all by seba_dos1 on Sunday November 15, 2026 @07:36PM (#50512564)
What takes so much space in /usr? I have a few flatpaks installed and plenty of debs that I rarely use and /usr takes 7GB, /var 5GB and the rest is pretty much all in /home. I only keep "cold" stuff on SD card, such as music and photos.

Epiphany does ad blocking on its own. At least you're going to get GPU rendering in Firefox on PureOS Dawn soon.

Yes, SD cards are unfortunately pretty slow there, as SoC's native SD interfaces are already taken by eMMC and Wi-Fi. The SD reader is integrated into the USB 2.0 hub and doesn't support UHS. Interestingly, older cards have a chance of being faster, as UHS cards often don't implement non-UHS "High Speed" mode that reaches 25MB/s and are therefore limited to 12.5MB/s "Default Speed".

Re:Librem 5 v2: A phone that frees us all by seba_dos1 on Saturday November 14, 2026 @06:22AM (#50496110)
> it seems pretty obvious that order loads reduced later in the Librem 5's lifetime due to chargebacks, allowing them to fulfill all remaining backer orders

So obvious it's completely wrong. Once the final production run finished there were enough phones produced to fulfill all the orders and have plenty to spare. It was actually refund requests that were the problem to handle due to limited cashflow. If you look at all the "I've got no phone yet" complains over these years, most of the time they turn out to mean "I requested a refund at some point and got no money back yet".

> Receiving a phone with 2015 specs

2018. Many key components, including the SoC itself, did not even exist back in 2015.

Re:Librem 5 v2: A phone that frees us all by nonesuchthing on Saturday November 14, 2026 @09:34PM (#50503425)
I don't buy the story but I'll relent, maybe this thing will ship. But come on with the Librem 5 v1 apologetics. Sure, the I.MX 8 came out in 2017, but quad core A53s were available as early as 2014 in MSM8916, running at very similar clockspeeds. MSM8937, which dates to 2016 has approximately similar fillrate and compute output (depending on SOC, due to clockrate). If we want to go higher than even the absolute lowest end junk that Qualcomm put out, MSM8956 eats the I.MX 8's lunch. Two A72's, 4 A53s, 3x the GPU compute. The GC7000 also has worse standards support, to the point where GTK 4 applications now run unaccelerated (see: megapixels). You should know this, given you seem to be a booster for the damn thing.

I'm not actually sure how v2 is much of an improvement here. 8GBs of ram is certainly plenty, but the BXE-2-32 is barely supported PowerVR sludge. The CPU actually provides _MORE_ FLOPs than the GPU! How? HOW? It's behind the MSM8956 mentioned earlier, and nearly equivalent to the GC7000lite in the first gen. The only consolation is that the hardware is actually _capable_ of VK 1.2, and should get full GL via Zink. As for the CPU, it's more cores, higher clocks, and the vector performance is pretty good for a mobile part, but actual scalar single thread for simple productivity applications is still stuck in the 2010s. We're still at "behind A72" levels of performance per core, and my experience with all A55 phones is that they are still kind of unbearable. We're fighting out here for sticks and rocks. If I wasn't giving you the most favorable comparisons with mid-late 2010s mobile SOCs anything I even mention would steamroll the K1 into oblivion.

You are wasting your money.

Re:Librem 5 v2: A phone that frees us all by seba_dos1 on Sunday November 15, 2026 @09:03AM (#50507010)
Sure, the phone was never even close to approaching high-end when it comes to performance, but you simply couldn't produce a device with its specs in 2014. i.MX 8M Quad came out in 2018 (and its actually usable mask revision in 2019) and in practice runs circles around 2015's A64 despite having similar cores. LPDDR4 was getting defined late 2014 and started appearing in flagships in 2015, but it was still somewhat common for middle-range devices to use LPDDR3 in 2018. Librem 5's camera was used in 2019 Android phones (and is still being used for ultrawide sensors today). eMMC 5.1 did not even exist in 2014. Four-constellation GNSS only started appearing in high end phones around 2017; Wi-Fi 6 in 2019. This all contributes to L5v1 still being useful in 2026, even if it does show its age at times.

> The GC7000 also has worse standards support, to the point where GTK 4 applications now run unaccelerated (see: megapixels).

Your information is outdated ;)