The Current Hardware Specifications For Purism's Librem 5 Phone

[Girolamo_Cavazzoni]

Not a fan of iOS or Apple but thee speaketh true.

[ldesnogu]

$600 for a phone running 4 Cortex-A53? No thanks.

[satai]

$600 for a phone running 4 Cortex-A53? No thanks.

Exactly!

[Girolamo_Cavazzoni]

As if performance would matter on a phone - especially on such a niche-nerdy one. For managing your schedule, writing mails briwsing the web you don't need out-of-order execution. And the price tag is as it is because if the low production volume.

You spoiled brats. :-P

[starshipeleven]

It should be trivial to ssh in from a family member's laptop (or RDP, VNC..) and run Libreoffice, browse the pictures gallery, run your browser (with your bookmarks and features), etc.

You should be able to use USB networking with the USB cable between desktop/laptop and phone, and fix it when it doesn't work, e.g. ssh in from the wireless network. When you fucked up all your wireless networking you should be able to use a USB keyboard instead ($1 USB-C to A dongle. can't find micro-USB to A in physical stores by the way..).
If you have fucked that up as well you'll still have a real terminal with the GNU commands rather than some crappy busybox with nothing in there except some files for a whole non GNU, non Unix OS that you have no idea how it works.

Rather than scourging web research results about how to back up your user data from Android (all leading to suggestions of Google Play apps) you'll back up your /home with cp -a, tar, scp, rsync or anything. Or you'll use a file system that supports snap shots if you want to.

This is the kind of things we might hope for I think.
So, it'd do all sorts of things and very reliably, because you can fix it when you mess it up, or backtrack, or wipe it and get your system back within minutes including your configuration dot files in the ~ directory, in the same ways you deal with x86 desktops and servers..

I will not continue further so as to not make a wall of text of soliloquy.
The other probable non user thing that will fill the RAM would be developers building their application on the platform.

I'm not sure how this is in any way related to using or needing more than 3GB of RAM.

[nomadewolf]

$600 for a phone running 4 Cortex-A53? No thanks.

You are paying for privacy, not (hardware) specs.

[starshipeleven]

Did you know the latest $529 iPad 2018 128GB cellular+wifi has 2 GB of RAM?

They say the reason it's sufficient is because iPad runs native high performance ObjC / C++ / Swift apps whereas Androids run interpreted garbage collected (twice the memory reqs) JS and Java apps.

I'm still not sure Android needs more than 2GB either. Has anyone actually checked the RAM usage of their apps?

[Guest]

Chrome user spotted.
In all seriousness, I don't see how it would fill so much RAM. It's a phone, not a laptop.

This would be a valid question from an iOS user but Android would beg to differ. Java is truly a monster runtime.

[ldesnogu]

You are paying for privacy, not (hardware) specs.

Oh I see privacy only is for people with enough money.

I hope at that price they also checked the modem firmware doesn't contain any backdoor. That is if they have access to the modem firmware source code, which is unlikely. And even if they have, it likely is under NDA which means *they* would have to check the whole stack by themselves.

Reading their blog they consider the modem as a black box so I guess they don't know what's inside. So much for privacy and security.

And what about what will be on their store? Do they ask devs to provide source? Who will look at the source? Who will guarantee privacy?

[L_A_G]

They still haven't been able to decide which exact SoC they're going to be using? You'd have thought that it would have been decided by the time they got the dev board out the door, but apparently not. Because while the name wouldn't suggest much of a difference between the i.MX 8M Quad and i.MX 8M Quad Mini, the latter is a 14nm die shrink with some significantly bumped up clocks (the main CPU cores running at 2 GHz as opposed to 1.3 GHz) and I'm not sure they're pin-compatible or even have the same package size.

I could easily find the datasheet for the older "regular" 8Ms, but not the newer Minis that were announced only last year; so I can't say if it's just a drop-in replacement. Apparently the older and newer versions of the i.MX8 are pin-compatible at least within their series, but couldn't find anything that explicitly stated pin-compatibility was also cross-series. The fact that it's listed as a "preproduction product" on NXP's website is probably the actual reason why they haven't decided if they're going to be using it. Maybe they're waiting on the necessary data (because there really isn't a publicly available datasheet for it yet) or production that isn't just engineering samples to actually start.

It's not like you can't start shipping out products, or at least dev devices, using engineering sample silicon. I've personally worked on a product that used a FPGA dev board using an SoC that really was listed by the manufacturer as an engineering sample. That was an "early" version of a dev board with issues like clear omissions in the documentation, but from what I could tell the final board didn't seem to have any other changes other than a version of the SoC whose model number didn't end in "ES" (as Xilinx engineering samples do) and the PCB was colored black rather than red.

However I could also be completely off-base here and this could just be a mix-up between the i.MX 8M Quad Mini and i.MX 8M Quad Lite, that latter of which the manufacturer explicitly states is a completely pin-compatible drop-in replacement for the regular i.MX 8M Quad (along with the cut down i.MX 8M Dual).