Batch Fir (Version 2) of Librem 5

Vega shader cores? Is it actually an AMD graphics component?

1 Like

Here is what might be easily confirmed:

Vega 1× (?):

• i.MX 8M Quad (28nm HPC) = GC7000Lite 3D GPU (4 shaders), OpenGL ES 1.1, 2.0, 3.0, 3.1, Vulkan, OpenCL 1.2 – 267 million triangles/sec – 1.6 Giga pixel/sec – 32 GFLOPs 32-bit or 64 GFLOPs 16-bit

Vega-Lite (?):

• i.MX 8M Plus (14nm LPC FinFET) = GC7000UltraLite 3D GPU (2 shaders), OpenGL ES 2.0, 3.0, 3.1, Vulkan, OpenCL 1.2; GC520 (2D) – similar to Marvell Armada PXA1908 (28nm) + GC520 (2D)

• i.MX 8M Mini (14nm LPC FinFET) = GC NanoUltra 3D GPU (1 shader), OpenGL ES 2.0; GC328 (2D) – modern smartphone usable?

1 Like

Increase the battery capacity to 6k mAh in Librem 5 Fir/EvergreenCompat to fulfill the demand on true gnu-mobile-device.

All I care about for Fir is a newer next-generation CPU for better power efficiency. I hardly care for the battery itself since a larger capacity also means a tradeoff in larger weight and bulk.

1 Like

Well, it could either be same size, more capacity or smaller, same capacity.
But yes. Don’t add weight probably.

2 Likes

somewhat related to https://forums.puri.sm/t/librem-6-look-ahead-wishlist/18694?u=captain_morgan

nobody knows what it will look like, but maybe you have some wishes!

Obvious items on the outside:

  1. maybe thinner / lighter
  2. more moisture proof
  3. etc etc

No matter what kind of more advanced CPU is used in the next revision of the Librem 5, I don’t think it will matter much for power savings. The Librem 5’s high thirst for an abnormal number of electrons has almost nothing to do with the efficiency of the CPU. All other phones in the market use systems on chip SoC. All of the high integration micro-circuitry exists on the same silicon substrate on a SoC and are extremely small and packed close together on a nano scale. The power loss is almost imperceptible in that case. On the Librem 5, most of the communications and driver silicon chips are packaged and sold separately. Purism mounts them all on circuit boards inside of the phone. But you have literally inches of board-level traces to connect everything up. Inches might still seem small. But it might just as well be miles when compared to the nano-meters used to measure the insides of a packaged SoC. And pumping electrons down long board traces burns current, a lot of it. So a slightly better CPU in the next revision of the Librem 5 is only marginally better for power savings. It’ll probably be too small a difference to measure. The phone actually gets measurably hot on the outside of the case. A small increase of heat loss efficiency on the nano-meter scale means relatively nothing for extending battery life. But increasing the battery from 5Ah to 6Ah might help noticably.

2 Likes

Support ROSRUS GLONASSv2 for Librem 5 Fir.
Спутник

1 Like

Please add: Dynamic-AMOLED monitor for Librem 5 Fir. TFT is obsolete and poor visibility ourdoor and poor power save in black color and etc. I not sure about the Heat Support in DAMOLED.
Make the Gnu-phone-Fir the best, of course under free software.

2 Likes

Why not: Librem 5 Beech a Gnu Risc-V mobile phone?

2 Likes

Wouldn’t that be a rather confusing name? i.e. by comparison with “Birch”

Regardless, if a RISC-V phone were even possible for Purism, it would be arguable that it would not be “Librem 5” and/or would not be “Librem 5 v2”. The point is that all indications are that Fir will be a drop-in replacement CPU with a smaller process from the same manufacturer, as far as this is possible. These are of course only my assumptions, and plans can change.

If Purism actually went RISC-V then they would need to make a decision as to the future of the ARM range. Do they abandon it? thereby causing difficulties for the stated support timeframes for the existing phones. Do they support both? thereby increasing support costs.

If they had both ranges then I would suggest keeping the “trees” for ARM and finding some other naming for RISC-V (but not “colours” since that is used for PureOS operating system code names).

While there is no shortage of topics in this forum that pontificate on RISC-V, you have to answer the fundamental question: What problem or limitation with the existing Librem 5 are you attempting to solve?

I understand the theoretical benefit of RISC-V but I think many customers would prioritise more storage and/or more RAM and/or quite a few other hardware considerations … ahead of changing the CPU architecture.

1 Like

Yeah a little.

First i was think to open new thread to prevent conflict to Fir batch as Freescale arm s.o.cs. based, but Fir v2 could be brave Risc-V as Risc-V it using same arm concepts like Low Energy CPUs but diferent arch.

Librem 5 a.t.m is an Autonomous Device, so Purism may easy support both. Risc-V is fine now and is time to start abandon Arm.

The main to resolve is the Closed ISA from ARM.

For me the highest priority for Librem 5 would be to find more free hardware powersave controllers and optimize the sofware for great stamina
ATM Risc-V SoCs does not supporting Suspend to RAM, but may some fancy advantages than arm socs.

1 Like

Economics: if there is not enough demand, the development work and endeavour will not be sustainable.

1 Like

More storage and RAM would be best. Although I am planning on migrating a 4GB chip for RAM and a 128GB EMMC to the motherboard at some point…

You’ll need to swap the motherboard PCB with that of a Liberty phone, as I hear that the PCB layout is different enough that an in-place chip upgrade is not possible.

1 Like

Is it even possible to buy the motherboard from Purism?

Not yet but maybe soon. Stay tuned.

2 Likes

I wish this were true, but the i.MX 8M Plus is a very different chip than the i.MX 8M Quad. It has added an NPU, ISP, DSP, 3D audio, video hardware encoding and secure keys, which means that it will take a lot of work to fully support it. Another problem is that an additional chip will need to be added in order to convert from HDMI to DisplayPort Alt-Mode. See: FAQ: 7.1 What will be the changes in the Fir batch and should I wait for it?

I guess Purism could use the Plus without supporting those new components, but using the ISP and video hardware encoding would substantially improve the camera and make it a lot more energy efficient.

The biggest advantage of the Plus is the fact that its FinFET 14nm node size should run cooler and consume less energy, compared to the planar 28nm in the Quad. In my opinion, the poor battery life and the excess heat are the biggest problems in the Librem 5. I would like to see Purism use the 800MHz secondary processor in the Plus to implement a new low power mode for the phone. That would dramatically lower the power draw for many tasks, and would allow the phone to receives messages over the WiFi that it currently can’t do when suspended, and avoid the problem with the cellular modem not always waking up and taking too long to wake up.

However, it will be difficult for Purism to raise the funds to undertake a redesign of the Librem 5. Purism can’t use crowdfunding to finance its new projects any more, due to the reputational hit that it took with the Librem 5. If Purism had the finances, it probably would have designed the Librem 11 based on the i.MX 8M Plus, rather than using an ODM design with an x86 processor. I’m hoping to see the Fir based on the Plus, but I’m skeptical that Purism can get the finances for it, and the market will be small for it.

3 Likes

I doubt that: Purism did fully reach shipping parity with the Librem 5 and can use it as proof of their crowdfunding campaign success, in spite of all of the obstacles encountered throughout its journey.

Also, Purism has other means to fund their operations as stated in their most recent investment email.

To be clear … I was imagining an i.MX 9 series CPU, one that is more directly compatible with the existing (Evergreen) CPU.

3 Likes