Librem 5 Hardware Uncertainty and Backing

I love the idea of the Librem 5 phone. I would love to back the project but for me, the hardware is a very important aspect. Right now, there is but uncertainty about the most relevant parts which is holding me back.

  • What will the resolution of the 5" display be?
  • 3GB of RAM seem to be pretty much default for higher-end devices. Why not go for 4GB? Is it not feasible for a 2019 goal?
  • I would very much prefer the i.MX8 CPU. You should commit to it or at least make a clear decision for either i.MX6 or i.MX8 before the deadline.
  • What kind of battery will be included? I would prefer much power like the Nexus 6P (3450 mAh). What are your goals here?

Is there any chance you can update the tech specs accordingly? It would make me feel much more comfortable about buying a phone that ships as late as 2019. If you, by any chance, add hardware which is crappy today, think about how crappy it will be in 1.5 years…

Thank you!

2 Likes

Thank you for your feedback!

Concerning your questions:

  • display resolution: We strive for the best we can get but sourcing displays is a really nasty process, you will only definitively know what you get once you sign the contract, which we actually can not yet do. What we can say for sure is that the display will at least be HD and that we would like to have fullHD if we can source them

  • 3GB RAM is IIRC about all the i.MX6 dual can deal with, though I currently can not give you a reference for it

  • we would love to use the i.MX8 right from the start but… first of all the i.MX8 is not yet available and we do not want to promise stuff we probably can not deliver. Second it is unlikely that the i.MX8 will have proper and full mainline kernel support right from day one - whenever this day will be. So currently the only viable choice we can hold in our hands is the i.MX6. Should the i.MX8 appear to be available soon after ending the campaign, at a reasonable price and have sufficient free software support, we will us it. Judging from the datasheets the iMX8 is the much better CPU for our phone use case… fully agreed.

Hope this helps a little.

Cheers
nicole

2 Likes

Thanks for the prompt reply, much appreciated. I understand that it’s not easy to finalize the hardware design right now but I cannot shake the feeling that this phone will be outdated (hardware-wise) the moment it is released. I hope I can get one more comment from you on the following thoughts:

  • I don’t have any insight or clue when it comes to sourcing displays, so bare with me. In my option having only HD is really bad (like really really). My five-year-old Nexus 4 has that. in 2019 the Librem 5 will feature 7 years old display technology. Why not commit to at least full HD? You can postpone the release a few months (which will likely happen anyway due to unforeseeable reasons). For example, the Nexus 6P has a 5.7" screen with a 2560x1440 resolution which is a much higher display density than what you are aiming for and it’s from 2015 selling for $499.

  • I could not find the specs regarding maximum RAM for the i.MX6 but I could live with 3GB I guess. On the other hand I strongly disagree with the i.MX6 choice. Again using the Nexus 6P as an example, it features “an octa-core Snapdragon 810 v2.1 system-on-chip (SoC) with four ARM Cortex-A57 cores” (Wikipedia) which are the same cores featured on the i.MX8 except it is from 2014 and supports even 4K displays. Is there any reason to limit the choice to the i.MX6 and i.MX8?

I understand that it’s hard to compare your company and this project with giants like Google and Huawei. But compared to the two-year-old Nexus 6P, the hardware specs look old-ish and are more comparable to the 5-year-old Nexus 4. If the project size limits the hardware specs (due to higher prices for purchasing hardware), why not up the price a little? I would rather pay $799 for a Nexus 6P-like phone than $599 for a Nexus 4-like phone.

I hope you can see my comments as constructive and understand that I ask these questions because I eagerly await a mobile phone that runs pure Linux. Thank you for the hard work you are putting into this!

Can you at least consider the option of have an upgradable chip, so for example could be possible to have the i.MX8 after the launch of the phone?

This is my perspective on things:

1: While a high-resolution display will give you more numbers on a spec sheet, there’s a tradeoff to be made in terms of power consumption. More pixels takes more computing power to render and therefore more battery drain. While not the biggest factor, it still matters and in a mobile device, battery is king.

Also, think about the physical pixel sizes. A Samsung Galaxy S5 has a 1920x1080 OLED, 113x64 mm display - this is 17 pixels per millimetre. Only with the aid of a magnifying glass can I even begin to make out individual pixels. Even on a larger screen with lower resolution (Asus Zenfone Max, 5.5" diagonal, 1280x720 LCD), pixels never annoyed me. Perhaps your eyes are better than mine - have you noticed and gotten annoyed with pixels on your Nexus 4?

My point here is that there’s a limit to what your eyes can resolve, and that increasing the resolution past that point is spending both money and power for a gain which you will never be able to notice and only serves to puff up numbers on a marketing sheet. Again - your vision may well be better than mine and that limit will vary between people, but the point of diminishing (or no) returns will always be there.

2: The problem with Qualcomm parts is that a) there’s a lot of stuff in there which you may not have control over (eg. signed bootloaders at the very least, also not sure what TrustZone stuff they use but I doubt they’d let each user set their own master key for their own device) and b) I know that their graphics chip (possibly other sub-components, don’t know about anything else) does not have Free drivers and if you can even get hardware acceleration in the first place, you’d be stuck with an un-auditable blob from who knows how long ago which will never be updated.

If I understood you correctly, the majority of mobile phone manufacturers are installing higher pixel density screens just to show off and possibly to maximize the battery drain? I assure you they have better reasons. I can indeed see the pixels on my Nexus 4, from a close range. But what is worse is that I can clearly and without much effort see aliasing on fonts and images and I bet if you are being honest you can, too. The effect is magnified for smaller fonts up to the point where they become barely distinguishable.

Saying the screen resolution doesn’t matter because you cannot see a difference is not an argument in my opinion. Extending your logic there should not even be a market for 4K monitors and beyond. Power consumption on handheld devices is hugely influenced by the screen backlight anyway and that is basically the same for both variants.

I agree that not having the possibility for free drivers is a no-go for this project. But there must be other manufacturers (the Nexus 6P hardware was just an example, and IMHO a good one due to the price/hardware ratio) that create SoCs similart to the Qualcomm one. Take the $369.99 Meizu PRO 5 Ubuntu Edition (which has much higher hardware specs than what we are to expect for the Librem 5 phone) as an example - it features a Samsung Exynos 7420 Octa SoC and runs Linux. I would assume that it has free drivers support? At least there seems to be strong effort and progress towards Mainline Linux support (http://linux-exynos.org).

BTW, Qualcomm is acquiring NXP which acquired Freescale Semiconductor, the manufacturer of the i.MX SoCs. So in the end it does come down to the openness regarding individual chips instead of the openness of the manufacturer, i guess.

I genuinely cannot see the pixels on my Samsung S5 without a magnifying glass. They were noticeable on my Zenfone Max if I brought it close to my face, but it wasn’t visible at longer ranges for me. Your eyes are obviously much better than mine. As for putting higher and higher resolution screens on, of course they won’t do it to deliberately increase the battery drain, and of course, people tend to pick the thing with the nicer-looking number after it (sacrificing things like battery life or being able to repair it in favour of having a thinner device).

Desktop monitors have the advantage of being extremely large compared to a phone screen, so the extra pixels would be more readily useful. Regardless of that, however, you notice pixellation at 1280x720 on a 5" screen, so it is indeed a potential issue.

I personally wouldn’t complain about having a 1920x1080 screen, but my absolute #1 priority is a phone that will have a very hard time betraying me (read: all communication components are separate from one another, may only talk to each other with the explicit consent of the software running on the main CPU, and have absolutely zero influence on anything else) and as such I don’t want to strangle it in the crib with feature creep. I’ve already seen the GTA04A5 die, I’d rather not see another similar project fail for whatever reason (whether it’s delays in sourcing more panels, or difficulties in getting an as-yet untested CPU running, or anything else).

And yes, backlight level does have a much larger effect on power drain than the processing power to render more pixels, but that tiny little overhead is still there.

The contents of that link do look promising so yes, Exynos chips might be a possibility. I admittedly do not know that much about the state of free drivers for mobile graphics acceleration other than PowerVR is complete crap and that Adreno has only the beginnings of a functional driver. I don’t know enough about the other things (how easy is it to get them, what other components do they contain which need drivers, what possible security issues exist from the chip’s design) to say any more. Again, I just want it to work, and more importantly, I want it to actually exist.

First of all, thank you for taking the time and discussing this topic. To be honest I was surprised to get an official response at all, let alone that fast. I have seen many projects fail, starting with OpenMoko. Back then I was enthusiastic enough to even buy a t-shirt :wink: (in addition to the phone) and truly believed in the idea. Then the Nokia N900, Maemo/Meego, and the Ubuntu Phone. I even may have forgotten a project or two but IMHO most of them had several issues in common that lead to their downfall: Bad hardware and/or bad communication and organisation. The OpenMoko project had the crappiest hardware design I have ever seen and the Nokia N900 was a big and heavy phone (which had a great hardware keyboard though, I used it a long time). The Nokia N900 had the worst organisation while having a great community and the last project to fail - Ubuntu Phone/Touch - had awesome hardware but failed to deliver even half-decent software. The Ubuntu community/mailing list almost went crazy due to the lack off communication by Canonical…

What I am seeing now is a mixture of the old issues and a brand new one:

  1. Judging from this forum, communication with the Purism company isn’t that great.
  2. The hardware is decent (not great) for a 2017 phone, not so much for a 2019 phone. But - and this is new:
  3. It is not even clear which hardware will be used in the phone when you are being asked to buy one of them.

I hope I am wrong but I feel like this project is bound to fail for these reasons. Most people who want such a phone are techy. I think - and that is probably one of the main reasons for me to hope for more powerful hardware - that such a device should last for many years and should still be usable even with increasingly demanding software. Sure, the privacy feature is a must-have, but not at the cost of cheap hardware. So, the first thing I did when I heard of this project was scroll down the page looking for the hardware specs. I almost decided to move on and ignore the project but then decided to ask for clarification here.

So here I am thinking I will have to wait this one out as well. I will probably buy a Librem 5 phone after it is released (if the fund raiser succeeds) given they managed to put in a 1080p screen and the more advanced processor. I hope that they can make it. Otherwise this is my advice for the next try:

  • Choose hardware that is at least on a par with existing high-end devices
  • Only start a fund raiser after the hardware design has been finalized (it does not mean changes are impossible)
  • Keep the community happy during the fund raiser and development phase (community management, weekly updates, etc., etc.)

Sorry for the lengthy post, I think that is just the frustration about all these failed projects so far.

1 Like

i understand your concerns but i think different (can i say that? :joy: )

it seems to me too just reading the forum, i think the main reason is the team is small and they have to manage alot of stuff do not depend on them, i remember when someone from purism wrote about it, and i understand, so what i hope is they just calculate this when they write about shipping and release, just add some months, and if no problem the customer will be happy to see his product before the deadline

it depends, if you are talking about camera/display specs for you should be better have a 4k display, imho a hd is better, because of power required and this means more autonomy, a fullhd laptop is bigger than a 5" smartphone and i still no see pixels on it, and i consider i just have to call message and browse with a smartphone not to use photoshop/gimp for professional use

that’s why they are smart, the imx8 is not avalaible yet and they (and me too) hope will be before they ship the smartphone, so they can use it, otherwise will be the imx6
same will be for the screen and other stuff, i think it will also depend on raised money, more money they have more investment they can do

as customers i know we usually have high specs with a lower price, but this is because they make money with your data, so they can do it, santa claus is not real, people work for earn money and sometime, like purism also for a social purpose
i don’t really care about higher specs just for numbers, when i can do everything i can do it with other smartphone like photos, offline maps, calling, texting, internet etc without lags like a saw on ubuntu touch even on high spec phone, i’m fine with that

no one can say what you should do, it’s your money, i just think if this project will fail no one else will try again to make something different from android/ios, that’s why you should consider to help the campaign

don’t worry about the lenght, we are there for talk and help each others, i understand your frustration about all failed projects, that’s why everyone who believe privacy is important and everyone who love linux and open source, should do something now, for avoid to have another failure

1 Like

Thank you all for this very good discussion and the input you give!
I want to assure to you that we definitely read all of it and keep it in mind while going forward.

I just wanted to add to the CPU discussion a bit of information why we still stick with the i.MX6. There have been quite a number of proposals for alternative CPUs here in the thread and elsewhere - thank you for this!

The main issue is that pretty much any other CPU we found (and which have been proposed) do not have free GPU drivers :frowning: This is also true for the very nice Samsung Exynos, they use MALI GPU cores for which there is a reverse engineering project (LIMA) but this has not reached a usable state yet - and seems to be pretty stale for some time.

The GPU is a pretty vital part. We can not go without it since we definitely need the hardware acceleration for the GUI or user interaction will become awfully slow (drawing app windows, scrolling etc.). We also definitely do not want to make any compromise here, i.e. we do not want to use any form of binary blob to support the GPU. The GPU is a complex CPU on its own and it has access, to some extent, to the main memory. So the GPU software (the blob) carries the risk to be exploited which is a high security risk. Apart from this it limits users’ freedoms which we also do not want.

So we want and can only use CPUs/SOCs with free GPU support. And this in the end still boils down to the i.MX6 with the free Etnaviv driver which is also not yet perfect but it can support Wayland good enough to run our GUI.

If anyone can give us any hint for a better mobile CPU/SOC with free GPU drivers and good mainline Linux kernel support, we are all ears! We did quite some research on this issue and also evaluated choices like Intel Atom X5 or X7 (Baytrail/Cherrytrail) but all of them had sooner or later revealed severe issues with free software.

Finally I can assure to you that we will definitely switch/upgrade to the i.MX8 if it becomes available during the course of the project and if it then meets our standards of freedom.

Cheers
nicole

3 Likes

Thanks for the explanation. Is there any way to have an upgrade of the CPU after the shipment, maybe as a payment option? I understand that this could be hard to achieve, but could be fantastic for the longevity of this phone: at this point my only concern is about that part, unfortunately the I.MX6 is a bit ancient, but I prefer it instead some other blob-dependant-but-more-powerful solution, if it’s the only way

This would require a modular design where the CPU resides on some kind of exchangeable module. While this is not 100% impossible to do it requires a much more complex hardware design which we currently do not aim for. A modular approach also requires more space since you need extra connectors to connect the modules. We will not have this space since we use up any possible space by separating things (baseband, GPS, CPU etc.). Integrated designs with mobile+WiFi+GNSS inside the same chip package as the CPU save a lot of space/room, which we can not :frowning: So we will end up with bigger device anyway and we do not want to make it even bigger by adding the complexity of modules.

Though I fully understand your concern and wish!

Once we have a case design (I mean a design and fit of internal parts) we can discuss with the manufacturer a possible solution so that users might be enabled to replace the mainboard by a later version, e.g. with i.MX8 and so at least reuse case, display, battery etc. I can not promise anything right now but we will surely keep this in mind.

Cheers
nicole

2 Likes

Yeah, even a homemade “soldering” solution or a socketed CPU (not sure if that take too much space btw) could be good, but also the replaceable motherboard is perfect. My main concern is that the hardware would be too weak, and even if this seems not to be a concern for many, is certainly one for a lot of people. And one of the main problems of openmoko was the underpowered CPU.
Thanks for the reply!!

Do we have an idea which GPS chip will be used?
The BCM47755 is said to locate with all location systems out there now and layer 5 support up to 30cm in accuracy.

@nicole.faerber maybe you could contact this guy who tries to create also a total free computer: https://www.crowdsupply.com/eoma68

The EOMA68 is a very neat idea! And we would love to embrace it if it would feature a CPU with free GPU drivers. The Allwinner A20 is a pretty badly mainline supported CPU and especially it features a MALI GPU for which no realistically usable free GPU drivers exist. The Allwinner CPUs traditionally are pretty badly Linux mainline supported, their SDKs even come with a bunch of GPL and other free/libre software license infringements. Yes, it mostly works with mainline kernels but with too many “yes, but…” to be usable for us. In order to meet user expectations especially concerning user interface interaction we must support hardware GUI support by a GPU - and we need a free/libre solution for it.

Cheers
nicole

1 Like

Thank you for the continued discussion and (rather unexpected) official statements. I changed my mind and will back this project as @eagle might me right: it could be the last chance for a long time to get a true Linux phone. I don’t particularly like the decisions regarding the hardware (although I do understand the limitations @nicole.faerber stated regarding driver availability) but I will see the backing as an investment into the philosophy and hopefully similar future projects.

7 Likes

Thank you very much @fkhhf!

We do our very best to create the best open / free device we can. But technology / industry imposes quite a lot of limitations on our project :frowning:

Cheers
nicole

This is a problematic drawback but of course you would like to use a more modern and powerful processor and most of us understand that. The limitations are there.

Maybe we could make balance and ask the community (the bakers) how do they feel about waiting more time if that could mean more possibilities of having the i.MX8.
I’d like to finally have a free smartphone as soon as possible as I find frustrating how unethical the Android environment is sometimes. In the other hand this project will potentially have a lot of contributions from big organizations like Gnome and KDE and a vibrant community of free software and privacy enthusiasts. I don’t care about a fancy 1080 screen or super fast DDR4 memory but the CPU is the most important component here and it can be the difference between a sometimes laggy UX and a perfectly fine experience.
I prefer a longer wait if that helps to alleviate the problems of imminent deadlines when you can start working with the MX8.
I don’t know what others think about this.

2 Likes

Luckily we are still in a position where we do not have to make a final decision now. There is still quite a mile of road ahead of us. If all works well and the internal information we got from other sources close to NXP is correct we will be able to transition from iMX6 to iMX8M. You might probably phrase it this way: i.MX6 would be the bad outcome of this development. Nobody really wants it but currently it is the bird we can get hold of.

During the next 2-3 months we will get more information and this will hopefully make it clear if a project delay is feasible in order to target iMX8. As you have seen from the original NXP announcement, they wanted to have first silicon out by beginning of 2017. If this continues we can not count on it since we can not delay the whole project by yet another year. If the delay is a month or two, or maybe three, be assured that we will not only ask the backers to wait a little longer, we will rather beg for their patience :slight_smile:

Let us all hope for the best and that NXP gets their stuff done in time :slight_smile:
Cheers
nicole

5 Likes