Announcement

Collapse
No announcement yet.

Rust-Written Apple DRM Linux Kernel Driver Renders First Cube

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by karolherbst View Post

    I am wondering why people keep bringing up such non issues. The linux kernel also fails to compile due to newer compilers and other funny things. If we'd go by your logic, we wouldn't write software at all, but still discuss standards.
    He is Probably someone who just screams at the clouds.

    Comment


    • #12
      Originally posted by muncrief View Post
      I'm very curious to see how an unfinished and ever changing programming language like Rust will work in an operating system. Heck, I couldn't even get four out of five demos to run a few years ago because Rust is such a mess that too many things had changed since the demos were written a year before. It was really odd, as it seemed that their was no concept of backward compatibility in Rust development at all.
      Either you tested Rust before 1.0 came out (which is 7 years ago...), or your "examples" made heavy use of unstable / nightly language features, or some other weird thing is going on. Rust's backwards compatibility story is quite good. It's more the other direction that is likely to be a problem: trying to use an old compiler on recent versions of various libraries. But that's not a fundamental problem with the language or the compiler, it's more that there is not yet a critical mass of companies who for one reason or another have decided they want to use an ancient compiler and are willing to put in the work to deal with the resulting pain.


      Comment


      • #13
        Originally posted by muncrief View Post
        Instead I would wait until development had finished, standards had been fully established, and extensive tests had been completed to assure functionality and safety.
        Would be endless fun to watch Thompson and Ritchie waiting for ANSI C standard before they write a single line of Unix code.

        Comment


        • #14
          I was following the streams and what's interesting is that the driver was remarkably stable even throughout development. When writing drivers a lot of time is often spent looking for the reason for the next crash, here there were just a few crashes and she spent most time looking for logic errors. That's a very big win for driver development, or development anywhere.

          Also Rust features are used for versioning of FW structures so it'll be easy to support changes without copy/paste of a lot of code.
          Not just secure, but more reliable! This particular GPU has a limitation that means that if the firmware crashes, the user needs to reboot the...

          Comment


          • #15
            Originally posted by hyperchaotic View Post
            I was following the streams and what's interesting is that the driver was remarkably stable even throughout development. When writing drivers a lot of time is often spent looking for the reason for the next crash, here there were just a few crashes and she spent most time looking for logic errors. That's a very big win for driver development, or development anywhere.

            Also Rust features are used for versioning of FW structures so it'll be easy to support changes without copy/paste of a lot of code.
            That's also what I noticed writing rusticl. Logic errors are what you track down most of the time. If it's a crash, then it's either because the C code was wrapped and used incorrectly, or it's something inside C code.

            Comment


            • #16
              Incredible progress by what is mostly a one-(wo-)man show.
              Without Hector we might have never gotten Linux on Apple silicon but it's great we have it and I *might* consider buying a laptop from Apple in the future because the battery life combined with performance is just unmatched.
              Last edited by tunnelblick; 25 September 2022, 02:51 AM.

              Comment


              • #17
                Originally posted by tunnelblick View Post
                Incredible progress by that is mostly a one-(wo-)man show.
                Without Hector we might have never gotten Linux on Apple silicon but it's great we have it and I *might* consider buying a laptop from Apple in the future because the battery life combined with performance is just unmatched.
                Though I admire the herculean effort to get linux running here, I'd probably not try it out on the new M2, M1 macs. These are expensive piece of equipment, writing something accidental into some registers or getting it into a weird state could permanently brick it.

                Comment


                • #18
                  Originally posted by swoorup View Post

                  Though I admire the herculean effort to get linux running here, I'd probably not try it out on the new M2, M1 macs. These are expensive piece of equipment, writing something accidental into some registers or getting it into a weird state could permanently brick it.
                  Why is that only valid for Apple laptops? The same could happen on an x86 laptop, Apple is not special in that regard.
                  On the other hand there are safety measures in hardware nowadays, be it actual hardware and/or firmware. Also you can get an M1 or M2 Macbook Air below $2000 which is not way more than you have to pay for a quality x86 laptop. Yes, they are still pricey but not double the price.

                  Comment


                  • #19
                    Originally posted by muncrief View Post
                    I'm very curious to see how an unfinished and ever changing programming language like Rust will work in an operating system. Heck, I couldn't even get four out of five demos to run a few years ago because Rust is such a mess that too many things had changed since the demos were written a year before. It was really odd, as it seemed that there was no concept of backward compatibility in Rust development at all.

                    Don't get me wrong, I like the ideas Rust is attempting to implement. However I also like the ideas nuclear fusion is attempting to implement, but still wouldn't make an experimental Tokamak part of an energy infrastructure.

                    Instead I would wait until development had finished, standards had been fully established, and extensive tests had been completed to assure functionality and safety.

                    And then introduce Rust, or Tokamaks, as functional products.
                    Must have been before 1.0. it's clear you have no clue of how it works today.

                    Oh and how did you miss the Koreans running a tokamak for 30 seconds the other day?

                    Seems to me you need to update your viewpoints.
                    Last edited by dlq84; 25 September 2022, 03:10 AM.

                    Comment


                    • #20
                      Originally posted by swoorup View Post

                      Though I admire the herculean effort to get linux running here, I'd probably not try it out on the new M2, M1 macs. These are expensive piece of equipment, writing something accidental into some registers or getting it into a weird state could permanently brick it.
                      for the performance you get they are not expensive. They are rather cheap actually. Also.. hardware isn't that easy to brick anymore.

                      Comment

                      Working...
                      X