Librem 5 Will Begin Shipping In The Weeks Ahead, But Varying Quality Over Months Ahead

[oiaohm]

Purism is a software and product company. If you backed Purism, you really backed the software effort as much as the physical product. Based on their website, they seem to have dozens of professional engineers and a distributed operation. Their goal is to build a cohesive open source product.

This sounds good in theory. There is a downside to this lack of experience in hardware development hurts.

Pine is a manufacturer management company. They manage suppliers and focus on delivering the cheapest hardware possible without considering basic design or legal items like licenses. If you take a look at the board "designs", they're not really designed so much as they slap together the ports on board. PinePhone is really a hack (A64 glued to a USB 2.0 modem) when compared to the Librem 5 with i.MX 8.

http://bkhome.org/news/201902/pineph...pment-kit.html you have just mentioned one of the biggest problems with librem 5 its been the i.MX 8.

Yes Pine is a manufacturer management company meaning they are experienced getting boards out the door that in fact work by using chips that are already mass produced in volume and are dependable supply.

Throwing hardware at developers is the easiest thing in the world and cost next to nothing. An hour of professional software development is worth 10 dev boards. Please don't try to compare the two companies or products or open source support because there is nothing in common.

This is something you have wrong. Both pine and librem 5 shipped out development boards to developers. Pine was successful at shipping out working boards to developers first. Shipping out development boards that don't work end up with professional software developers working on something else that works instead.

In some ways the pinephone will partly save Purism phones ass. From a userspace application side the 4 core A53 processor design in the i.MX.8 and A64 are the same. The GPU is different Vivante GC7000 in the i.MX.8 and Mali 400 MP2 in the A64.

Does not matter how many software developers you have if your hardware does not work.

Hardware also means needing experience dealing with FCC and equal.

Really both Pine and Purism bring two different sides to the problem. The application side for the Purism device would be a lot worst place if pine had not been putting out working development boards for developer to work with.


Pine does not do the cheapest possible. Pine does the cheapest possible hardware that is reliable. Pine does not have the strictest rules on software. We do need a company in the mix who prime focus is getting dependable hardware into developers hands to work with.

[LoveRPi]

PLEASE don't try to try and disguise your fanboy-ism behind a faux cloak of objectivity, "LOVE RPi".

Have no fanboy-ism so don't know what you are talking about. The last few generations of RPi hardware has been mediocre at best and full of proprietary non-sense. Good job making claims without basis and contrary to fact. Anyone who thinks any thought went into "design" after taking a look at any of Pine's SBCs is delusional.

https://cdn11.bigcommerce.com/s-a1a2072ow5/images/stencil/1280x1280/products/486426/2692491/32904589024_Final1565431576482.82__36717.1565431601.jpg?c=2?imbypass=on

https://cdn.shopify.com/s/files/1/1884/2123/products/Rock64_top_x700.jpg?v=1524512481

http://files.pine64.org/wiki/PINE%20A64.png

Someone chewed up a bunch of ports and pin and spit it out randomly on a PCB. Anyone who knows how to design proper products will not use an SoC with a broken timer (A64).

https://lore.kernel.org/patchwork/cover/904606/

Facts are too hard for fanboys.

[LoveRPi]

Hardware design is the ultimate bitch. It's like software development, but each time you compile you pay thousands of $$.

Send a subsystem patch to upstream Linux and let me know how long it takes you to get it merged. Nevermind designing a cohesive product with proper software architecture across multiple layers of stack. It's hard to screw up hardware repeatedly but there are some talented companies namely Raspberry Pi and Pine.

[LoveRPi]

http://bkhome.org/news/201902/pineph...pment-kit.html you have just mentioned one of the biggest problems with librem 5 its been the i.MX 8.

Yes Pine is a manufacturer management company meaning they are experienced getting boards out the door that in fact work by using chips that are already mass produced in volume and are dependable supply.

iMX8M is a proper chip designed for the use-case (phone). A64 is not working: https://lore.kernel.org/patchwork/cover/904606/

In some ways the pinephone will partly save Purism phones ass. From a userspace application side the 4 core A53 processor design in the i.MX.8 and A64 are the same. The GPU is different Vivante GC7000 in the i.MX.8 and Mali 400 MP2 in the A64.

Mali 400 is far inferior to Vivante both in software and hardware. Mali 450 is a huge improvement on Mali 400 and can run a lot more shaders. Vivante is a class above that with OpenGL ES 3.x and compute shaders. It is a "modern" unified shader GPU where as Mali 4xx is obsolete in terms of design.

Does not matter how many software developers you have if your hardware does not work.

PinePhone's choice of A64 has an unfixable timer bug that prevents basic things like tracking time. Kernel and userspace will go haywire and corrupt things because of this. This is a non-starter any rational product company. That's like having a car without circular wheels.

Hardware also means needing experience dealing with FCC and equal.

PinePhone cannot be FCC certified because of its hardware design. Legally you can't use it on any provider's network.

Pine does not do the cheapest possible. Pine does the cheapest possible hardware that is reliable. Pine does not have the strictest rules on software. We do need a company in the mix who prime focus is getting dependable hardware into developers hands to work with.

https://forum.armbian.com/topic/9775...4-revision-30/

https://forum.pine64.org/showthread.php?tid=6176

I mean basic stuff here. It's cool to have options but it's not cool to half-ass everything and pretend to be a coherent product. Purism was in over its head but they're trying their damnest and at least following a proper process. It's crowdfunding for a reason with a meaningful goal.

[tildearrow]

Have no fanboy-ism

He thinks you do due to your username.

[starshipeleven]

Send a subsystem patch to upstream Linux and let me know how long it takes you to get it merged. Nevermind designing a cohesive product with proper software architecture across multiple layers of stack. It's hard to screw up hardware repeatedly but there are some talented companies namely Raspberry Pi and Pine.

Lol no you are comparing "trying to merge something into a project with high standards and lots of bikeshedding" with "going on your own".

Raspi hardware-design-wise is a piece of shit (see the project of reverse-engineering the board boot firmware for example), same goes for the PoE hat. But they are on their own. There is no "upstream", no gatekeepers. As long as it works at all it's good.
Same approach that they follow with their own Linux distro, they are using a hacked downstream kernel fork, where they have full control and can do as they please.

I can't speak for Pine as I don't follow them that much, but again there is none judging them or vetoing their designs.

And we are still talking of basic and simple stuff, low power SoCs with high integration and limited connectivity.

Even just moving into more powerful stuff like Marvell Armada and you see that everyone and their dog buys SoMs because it's so much cheaper than having to design themselves.

[LoveRPi]

Lol no you are comparing "trying to merge something into a project with high standards and lots of bikeshedding" with "going on your own".

Raspi hardware-design-wise is a piece of shit (see the project of reverse-engineering the board boot firmware for example), same goes for the PoE hat. But they are on their own. There is no "upstream", no gatekeepers. As long as it works at all it's good.
Same approach that they follow with their own Linux distro, they are using a hacked downstream kernel fork, where they have full control and can do as they please.

I totally agree with you with the exception on the cost of hardware. You can design low volume hardware for the price of a plane ticket to Shenzhen and a few thousand dollars. It's software that cost real money. It would take Purism 3 month at the most to do schematic, layout, sample, verification, and bringup if the software compatibility and driver support was there. Additionally modems and certification are incredibly complex and time consuming processes mostly rooted in firmware/software.

I was being facetious when I said "It's hard to screw up hardware repeatedly but there are some talented companies namely Raspberry Pi and Pine." How they manage to mess up every release is just amazing.

He thinks you do due to your username.

Hopefully he can focus on facts of the matter and not jump to (wrong) conclusions.

[oiaohm]

iMX8M is a proper chip designed for the use-case (phone). A64 is not working: https://lore.kernel.org/patchwork/cover/904606/

A64 has known faults and workarounds because it old enough to have work around.

PinePhone's choice of A64 has an unfixable timer bug that prevents basic things like tracking time. Kernel and userspace will go haywire and corrupt things because of this. This is a non-starter any rational product company. That's like having a car without circular wheels.

This is not silicon. No matter what chip you choose you are going to have square wheels in it somewhere.

https://community.nxp.com/thread/500975

Yes there are timer bugs in the IMX8M as well in different places. Problem is the IMX8M silicon had not been widely deployed enough yet for all those opps the silicon have done something that either sends Linux kernel nuts or kernel does something and the the silicon just freezes.

Mali 400 is far inferior to Vivante both in software and hardware. Mali 450 is a huge improvement on Mali 400 and can run a lot more shaders. Vivante is a class above that with OpenGL ES 3.x and compute shaders. It is a "modern" unified shader GPU where as Mali 4xx is obsolete in terms of design.

This is because you are really not aware how long this takes. Yes I am not saying the mali 400 is superior. But problem with using a older known chip is that you have older tech. Price of the choice.

PinePhone cannot be FCC certified because of its hardware design. Legally you can't use it on any provider's network.

Stop speaking out your ass.

http://files.pine64.org/doc/cert/SOPine%20FCC%20certification%20VOC20170428.pdf

Every step along the PinePhone process has had a FCC certificate done by a approved FCC party. This is the first pinephone development board. Do note the date "April 2017". So every pinephone device has been legal to use on any provider network as they have been careful todo the require FCC and other certifications.

So the base design for PinePhone is 2017. So the pinephone releasing this year has been in development for over 2 years. You do look at the first pinephone prototype it did have the cellular network board by M.2 port they removed that because they were told they would not get approval for general sale if they kept it.

I will give librem 5 developers respect for having the balls to argue with the FCC and other regulators over this point with risk their project could have been completely killed by arguing. I have not seen the librem 5 final certificates. Pinephone does have all its final FCC certificates so all they have todo is produce them and sell them at this point because they have all the i dotted and t cross for legal usage even if the software is not.

Final certificate need final casing. Up until the point of final certificate from FCC a phone product may be completely illegal to import/use.

https://forum.pine64.org/showthread.php?tid=6176

This one here is something that happens the raspberry pi 4 has 1 to few resisters and the Rockpro64 first version has 1 too many resistors. Turns out it was not that the Rockpro64 had to been tested with SD card port with emulator card just happens the sd emulator card they used works with extra resister in mix. This is case where QA can go wrong. They tested the SD port just test said it worked but when you put real sd card it does not.

I mean basic stuff here. It's cool to have options but it's not cool to half-ass everything and pretend to be a coherent product. Purism was in over its head but they're trying their damnest and at least following a proper process. It's crowdfunding for a reason with a meaningful goal.

Raspberry pi and other board vendors have had the same problems as pine group. Hardware development is quite hard and quite simple to screw up.

[esbeeb]

Yawn. If the phone isn't actually shipping in a stable state yet, then it's just a bunch of speculation, from my perspective. I've become jaded over the years from seeing efforts like this fail again and again.

Having said this, I wish Purism the best of luck, in realizing their sky-high ideals in a phone market where only slapdash, break-neck-speed quick-to-market devices can seem to survive whatsoever, let alone win the day.

I'm content with my LineageOS phone for now, and I'm pretty sure my next phone will probably also be LineageOS, considering the rate of development and progress of alternative OSS phone efforts like Purism's, and Pine64's.

[LoveRPi]

A64 has known faults and workarounds because it old enough to have work around.

You can't fix a fundamentally "broken" timer. The upstream merged workaround still doesn't fix the problem completely. There are many dfiferent workarounds for the A64 timer randomness but it's a tradeoff between merge-ability, excessive CPU time, userspace hackery, and the relative "brokeness" of the clock.

This is because you are really not aware how long this takes. Yes I am not saying the mali 400 is superior. But problem with using a older known chip is that you have older tech. Price of the choice.

Stop speaking out your ass.

We are yuq's sponsor for the lima project among many other open source projects in the ARM space. We only work with hard facts. Vivante is far superior to Utgard. Vivante has a solid Mesa Gallium3D GL 2.0 and GLES 2.0 driver. Utgard has a demo driver that can only run kmscube. There's still at least two man years of work per yuq.

http://files.pine64.org/doc/cert/SOP...OC20170428.pdf
Every step along the PinePhone process has had a FCC certificate done by a approved FCC party. This is the first pinephone development board. Do note the date "April 2017". So every pinephone device has been legal to use on any provider network as they have been careful todo the require FCC and other certifications.

Nope. Wrong. TL Lim's own words: https://ameridroid.com/blogs/ameribl...ls-from-tl-lim
"Unfortunately, smartphone FCC and CE certifications are a different ball game compared to standard electronic devices"
Where's the certifications for Pine64's WiFi modules? Reckless disregard is a moral hazard.

Pinephone does have all its final FCC certificates so all they have todo is produce them and sell them at this point because they have all the i dotted and t cross for legal usage even if the software is not.

This does not exist and will not exist. To quote you proverbially: "Stop speaking out your ass." Fanboy-ism helps nobody.