6 ms·

The rpi foundation are in trouble. The current SoC is old, limited at 1GB maximum RAM, and the GPU core is dated from 2011, 8 years ago. In my opinion they sho

by danaos 8y ago

The rpi foundation are in trouble. The current SoC is old, limited at 1GB maximum RAM, and the GPU core is dated from 2011, 8 years ago. In my opinion they should severe ties with Broadloom and open up to global competition.

dcbadacd 8y ago
There's also no crypto core and 64bit userland support (that could allow actually secure ASLR).
fsloth 8y ago
There is inherent value in having a stable, battle tested platform. The fact that the same chip has been sold for general public for the last 8 years is to me rather appealing than a turn off. Given the complexity of the modern software stack, there have been a zillion bugs that have ruined someone elses day and have been hopefully fixed. Although, I have no idea of the support quality of the Pi foundation since I have not used their products. The platform is sufficiently beefy to run lot of non-trivial applications. It's not like those applications have magically increased in complexity. We've just been brainwashed in the consumer space to always demand bigger numbers so that the vendors have something new to sell. There is really no need for a certain category of desktop applications to get any slower with time, or, to be able to provide any more value regardless how fast they are run if the interactivity level is already so fast that improvements would be undetectable by humans. Note: this is not an argument that the current Pi "is fast enough for everything". I'm saying it's fast enough for some things, and the mature software stack is considerable added value.
- snazz 8y ago
  I agree with this sentiment, but I do think that the RPi Foundation is going to have a hard time improving from their current state without breaking backwards compatibility, because they made some decisions early on that no longer make much sense today. The big ones, in my opinion, are the write-limited, slow SD card interface, the microUSB power connector, the 1GB maximum RAM, Ethernet that shares bandwidth with USB, and ever-increasing power requirements. I think the Pi Foundation wants to churn out more new units by increasing the CPU speed, but with the super slow SD interface and the capped-out RAM, performance hasn’t made significant improvements in a long time. I also think it was a bad decision to not focus on open-source software with the Pi. They have been including proprietary user space applications forever (Mathematica, for instance), the GPU boots a proprietary ThreadX operating system that has DMA and complete CPU control, and the thing requires lots of binary blobs to boot. The education aspect (particularly as a low-level programming platform) would have been made more accessible without such an opaque boot process.
  - fsloth 8y ago
    "The education aspect (particularly as a low-level programming platform) would have been made more accessible without such an opaque boot process." Not that I disagree, but as a counterpoint learning programming with Scratch has very little to do with getting to know the implementation details of the underlying computational substrate. I'm a software professional and I can't bother with that stuff! I'm far more interested in fiddling with my programs than tweaking the underlying substrate. I don't think this is a stance everyone should take but I don't feel a worse person because of it. I can administer my own machine but I don't enjoy it.
    - benj111 8y ago
      I cant speak for anyone else, but I got into programming from using Linux. The fact that you have all these languages just there, and its all an open system etc really encourages you that yes this is something you can do, and not just toy examples. I've seen the same point made about boot to Basic, 8 bit computers also. I don't by contrast program on my phone because it just feels like a gimmicky toy. I'm not saying the raspberry pi is anything like a phone in this regard, but in does have a limit to how far you can dig, and that's a real shame. Is it limiting to the scratch programmers? Probably not, but the Scratch programmers read about and move on Python, who then move on C. But if the boot process isn't open, there isn't anyone to write about it, so people don't discover it. Which for an educational endeavour is problematic.
      - fsloth 8y ago
        By the time person has moved on to C from scratch I find it likely they would have acquired a less restrained computer. Likewise, the fact you can't easily reverse engineer a rasberry does not stop one from getting a less restrained computer.
        snazz 8y ago
        Maybe this is true, but I would much rather work on my own operating system as an experiment on a small, simpler, inexpensive machine that I wouldn’t feel bad messing up. Although I applaud what the RPi people have done for computing education, this still feels like a bit of a missed opportunity.
pjc50 8y ago
Objectively, they don't seem to be in trouble, they seem to be expanding. The stable ecosystem with documentation aimed at beginners is the product, and it's a popular one. It's not as performance-dependent as you might think.
megous 8y ago
I suppose they quite like the fact that if you build something on Rpi you have to buy from Rpi and can't go on to design your own board and buy the SoC from Broadcom.
- SmellyGeekBoy 8y ago
  Is this true? It seems there are plenty of SBCs out there that share an identical form factor and are pin-compatible with the Pi's GPIO.
  - megous 8y ago
    None has the broadcom chip. So if you want to make a custom hardware project based on the broadcom chip that's in Rpi you can't. So if that's on your radar as an eventuality, you can just as well ignore the whole rpi thing, and start with more easily available chinese SoCs. That's why there's much more hacker friendly innovation around Allwinner SoCs for example, than the SoCs that are used in rpi.

by danaos 8y ago

dcbadacd 8y ago

There's also no crypto core and 64bit userland support (that could allow actually secure ASLR).

fsloth 8y ago

There is inherent value in having a stable, battle tested platform. The fact that the same chip has been sold for general public for the last 8 years is to me rather appealing than a turn off. Given the complexity of the modern software stack, there have been a zillion bugs that have ruined someone elses day and have been hopefully fixed. Although, I have no idea of the support quality of the Pi foundation since I have not used their products. The platform is sufficiently beefy to run lot of non-trivial applications. It's not like those applications have magically increased in complexity. We've just been brainwashed in the consumer space to always demand bigger numbers so that the vendors have something new to sell. There is really no need for a certain category of desktop applications to get any slower with time, or, to be able to provide any more value regardless how fast they are run if the interactivity level is already so fast that improvements would be undetectable by humans. Note: this is not an argument that the current Pi "is fast enough for everything". I'm saying it's fast enough for some things, and the mature software stack is considerable added value.

snazz 8y ago
I agree with this sentiment, but I do think that the RPi Foundation is going to have a hard time improving from their current state without breaking backwards compatibility, because they made some decisions early on that no longer make much sense today. The big ones, in my opinion, are the write-limited, slow SD card interface, the microUSB power connector, the 1GB maximum RAM, Ethernet that shares bandwidth with USB, and ever-increasing power requirements. I think the Pi Foundation wants to churn out more new units by increasing the CPU speed, but with the super slow SD interface and the capped-out RAM, performance hasn’t made significant improvements in a long time. I also think it was a bad decision to not focus on open-source software with the Pi. They have been including proprietary user space applications forever (Mathematica, for instance), the GPU boots a proprietary ThreadX operating system that has DMA and complete CPU control, and the thing requires lots of binary blobs to boot. The education aspect (particularly as a low-level programming platform) would have been made more accessible without such an opaque boot process.
- fsloth 8y ago
  "The education aspect (particularly as a low-level programming platform) would have been made more accessible without such an opaque boot process." Not that I disagree, but as a counterpoint learning programming with Scratch has very little to do with getting to know the implementation details of the underlying computational substrate. I'm a software professional and I can't bother with that stuff! I'm far more interested in fiddling with my programs than tweaking the underlying substrate. I don't think this is a stance everyone should take but I don't feel a worse person because of it. I can administer my own machine but I don't enjoy it.
  - benj111 8y ago
    I cant speak for anyone else, but I got into programming from using Linux. The fact that you have all these languages just there, and its all an open system etc really encourages you that yes this is something you can do, and not just toy examples. I've seen the same point made about boot to Basic, 8 bit computers also. I don't by contrast program on my phone because it just feels like a gimmicky toy. I'm not saying the raspberry pi is anything like a phone in this regard, but in does have a limit to how far you can dig, and that's a real shame. Is it limiting to the scratch programmers? Probably not, but the Scratch programmers read about and move on Python, who then move on C. But if the boot process isn't open, there isn't anyone to write about it, so people don't discover it. Which for an educational endeavour is problematic.
    - fsloth 8y ago
      By the time person has moved on to C from scratch I find it likely they would have acquired a less restrained computer. Likewise, the fact you can't easily reverse engineer a rasberry does not stop one from getting a less restrained computer.
      - snazz 8y ago
        Maybe this is true, but I would much rather work on my own operating system as an experiment on a small, simpler, inexpensive machine that I wouldn’t feel bad messing up. Although I applaud what the RPi people have done for computing education, this still feels like a bit of a missed opportunity.

pjc50 8y ago

Objectively, they don't seem to be in trouble, they seem to be expanding. The stable ecosystem with documentation aimed at beginners is the product, and it's a popular one. It's not as performance-dependent as you might think.

megous 8y ago

I suppose they quite like the fact that if you build something on Rpi you have to buy from Rpi and can't go on to design your own board and buy the SoC from Broadcom.

SmellyGeekBoy 8y ago
Is this true? It seems there are plenty of SBCs out there that share an identical form factor and are pin-compatible with the Pi's GPIO.
- megous 8y ago
  None has the broadcom chip. So if you want to make a custom hardware project based on the broadcom chip that's in Rpi you can't. So if that's on your radar as an eventuality, you can just as well ignore the whole rpi thing, and start with more easily available chinese SoCs. That's why there's much more hacker friendly innovation around Allwinner SoCs for example, than the SoCs that are used in rpi.