Rust will never be used in real-time robotics

Guess what, branching every time you read or write memory is slow as shit. The people who are not afraid of pointers WILL make the fastest robots.
The future of civilization is C-style C++ running on a GPU.

Mike Stoklasa's Worst Fan Shirt $21.68

CRIME Shirt $21.68

Mike Stoklasa's Worst Fan Shirt $21.68

  1. 4 weeks ago
    Anonymous

    robots will never be relevant outside factories

  2. 4 weeks ago
    Anonymous
  3. 4 weeks ago
    Anonymous

    >another shitty flame war thread.
    Frick off.

  4. 4 weeks ago
    Anonymous

    Would robots needs x64 processing?
    I'm wondering if pointer size is considered when it comes to space.

    • 4 weeks ago
      Anonymous

      one with an AGI would require x128 processing minimum

      • 4 weeks ago
        Anonymous

        we already have 128bit or bigger registers if needed, 64bit is already past the point of diminishing return for many use cases for example zero-ing/pushing/poping them take time, on average no more than ~23bits are set in practice (don't ask me where I get these numbers), take more physical space and energy, we need bigger internal and l1$ cache and many more issues, we definitely do not need 128bit wide general purpose registers and a fricking robot will never ever need what general purpose 128bits have to offer.

    • 4 weeks ago
      Anonymous

      >Would robots needs x64 processing?
      Most robots might not even need as much power as a 80386. The ones with sophisticated control systems though...

  5. 4 weeks ago
    Anonymous

    Thank you for your highly informed and well-researched take on this very very important topic.
    We need more people like you.

  6. 4 weeks ago
    Anonymous

    Ah yes! Another thread made by a college / NEET moron who thinks he knows the world and is cool by liking C and C++

    • 4 weeks ago
      Anonymous

      And he's already better than you.

    • 4 weeks ago
      Anonymous

      you will NEVER be a programmer

  7. 4 weeks ago
    Anonymous

    the 20% overhead rust adds will never be a bottleneck in robotics

    even so, you can make low-level optimizations through mental gymnastics

    • 4 weeks ago
      Anonymous

      >you can make low-level optimizations through mental gymnastics
      Like you?

  8. 4 weeks ago
    Anonymous

    "IT'S OVER!" Rust daily

  9. 4 weeks ago
    Anonymous

    >branching every time you read or write memory is slow as shit
    Rust doesn't do this except for when doing arbitrary array access. And you still have get_unchecked if need be.

    • 4 weeks ago
      Anonymous

      bounds checks

  10. 4 weeks ago
    Anonymous

    >Guess what, branching every time you read or write memory is slow as shit
    why make a thread about shit you don't know?

  11. 4 weeks ago
    Anonymous

    >Guess what, branching every time you read or write memory is slow as shit.
    That's not what Rust does. Rust can guarantee that pointers are valid and aren't null so you don't have to branch. C and C++ are where you always have to check for null pointers and uninitialized pointers.

    • 4 weeks ago
      Anonymous

      Only shit programmers have to check for null it means they can't hold the program state in their heads

      • 4 weeks ago
        Anonymous

        >t. has never programmed anything larger than 400 LOC

        • 4 weeks ago
          Anonymous

          Why would a control program for a robot need to be more than 400 lines of well-written C?

          • 4 weeks ago
            Anonymous

            well-written C does not exist

            • 4 weeks ago
              Anonymous

              kek, unironically this, all C libraries I ever used had atrocious codebases
              >libcurl, libsqlite, glfw, sdl, blah blah
              yes they're incredibly useful but at the same time they're incredibly shit

  12. 4 weeks ago
    Anonymous

    OP, second place in the DARPA Robotics Challenge was won by a robot running java. That's right java for control and balance algorithms. Their only concerns for making it run in realtime were restricting it to run on a single core. The choice of programming language used really doesn't make much of a difference. Choice of language is so fricking far from the actual problems robots have and what makes them fast it's not funny.

    • 4 weeks ago
      Anonymous

      OP is obviously a clueless no-coding BOTtard.
      But as is tradition, with every field BOTtards think Rust is not useful, it's actually used there already.
      Chinks in particular have been using Rust in realtime environments for years.
      Let's see how BOTtards can "ywnbaw" their way out of this one. lol.

    • 4 weeks ago
      Anonymous

      that remind me of this https://en.wikipedia.org/wiki/Lego_Mindstorms_NXT
      I've learned real-time programing on these things by reimplementing real-time schedulers in java for https://en.wikipedia.org/wiki/LeJOS which was the firmware we used.
      people are wasting to much time caring about ideological purity or some shit meanwhile real engineers are using java on simcards https://en.wikipedia.org/wiki/Java_Card
      there is really no point in discussing anything with these people, we should focus on getting shit done instead, if something requires a rewrite in C later on then so be it, who gives a frick.

      OP is obviously a clueless no-coding BOTtard.
      But as is tradition, with every field BOTtards think Rust is not useful, it's actually used there already.
      Chinks in particular have been using Rust in realtime environments for years.
      Let's see how BOTtards can "ywnbaw" their way out of this one. lol.

      don't waste your time, rust is perfectly fine for firmware, I've written a rust firmware myself for some old advantech io card at work, it just fricking work, I didn't micro-benchmark it but it has been running on 100k+ hardware without any reported issue so...

      • 4 weeks ago
        Anonymous

        >rust
        nocode shitter

    • 4 weeks ago
      Anonymous

      Java has always been very performant though, I'm not sure why it has this image of being some sort of purely functional third order framework. I feel like people don't even remember what Java's use case was originally intended to be.

  13. 4 weeks ago
    Anonymous

    You have no idea what you are talking about.

  14. 4 weeks ago
    Anonymous

    Assembly line robotics do not require conditionals, with the only exception being another system feeding it input for timing purposes.

    Everything else can be purely procedural.

  15. 4 weeks ago
    Anonymous

    >The future of civilization is C-style C++ running on a GPU
    Bend is the future.
    https://higherorderco.com/

    • 4 weeks ago
      Anonymous

      chapel seems superior
      https://chapel-lang.org/tutorials/Oct2023/03-Chapel-Parallelism.pdf
      also funny how a "chud" had to bring it up, troony

    • 4 weeks ago
      Anonymous

      this is just reinvented futhark/taichi without all the good tooling/platform support. Same shit with mojo.

  16. 4 weeks ago
    Anonymous

    https://www.theroboticsspace.com/blog/Rust-For-Robotics/
    > branching every time you read or write memory is slow as shit
    Yeah and? What does Rust have to do with this?

  17. 4 weeks ago
    Anonymous

    the next robotics breakthrough will be written by some ai script kiddie copy-pasting Python code from so and accidentally solving the 3-axis problem

Your email address will not be published. Required fields are marked *