Announcement

Collapse
No announcement yet.

TUXEDO Developing A Snapdragon X Elite Linux Laptop

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

  • #31
    Originally posted by NateHubbard View Post
    I'd really like to see a mini-itx board with ram slots using one of these new chips.
    I'll vouch for on-package RAM if they can guaranteed the reliability, have better pricing vs slotted RAM, and adequate amount of RAM (Min. 16GB. I'll go 32GB or 48GB).

    Edit:
    Just make the SoC/CPU socketable.
    Last edited by t.s.; 11 June 2024, 10:28 PM.

    Comment


    • #32
      Originally posted by Alexmitter View Post
      And the device tree is provided by the firmware, so it acts as ACPI just without being shitty.
      I don't understand how it helps: if there is no standard way (e.g. UEFI) to communicate that device tree to the OS, then nothing is solved.

      And another issue comes to mind: how do you enter a "BIOS password" to prevent malvare installed in the boot stage (whether in U-boot or in the subsequent kernel loading sequence)?
      Also, related issue: How do you sign a bootloader/kernel without UEFI?
      Last edited by drastic; 12 June 2024, 01:55 AM.

      Comment


      • #33
        Originally posted by drastic View Post

        I don't understand how it helps: if there is no standard way (e.g. UEFI) to communicate that device tree to the OS, then nothing is solved.
        Every uboot already contains a device tree, the same as the kernel tho either can have a newer one which then you can decide which one to favor.
        And there is a standard way of passing it to the kernel in the boot chain, and it does not matter if you boot via EFI directly, EFI grub via uboot, extlinux or by directly loading the kernel in uboot via fatload etc.

        Comment


        • #34
          Originally posted by Alexmitter View Post

          Every uboot already contains a device tree, the same as the kernel tho either can have a newer one which then you can decide which one to favor.
          And there is a standard way of passing it to the kernel in the boot chain, and it does not matter if you boot via EFI directly, EFI grub via uboot, extlinux or by directly loading the kernel in uboot via fatload etc.
          The kernel does not usually contain a device tree, except when it is a kernel built specifically for a single device (instead of a generic kernel). And, such a special kernel is quite a hassle, and that is exactly what the UEFI solution avoids.

          Therefore, the device tree must be in the firmware or in the u-boot.

          I can't quite figure out what is the overall architecture of the system you are describing: is the u-boot a part of the SoC firmware, or is it loaded from external flash-able storage as a second-stage after the firmware? Is the SoC firmware stored on the SoC, or is it also loaded from external storage on every boot?

          Anyway, the problem with security = BIOS password = bootloader&kernel signing remains in any solution lacking such a functionality in the (probably closed-source) firmware.

          Comment


          • #35
            Originally posted by Doomer View Post

            that's not happening. the extreme ewaste production disease is spreading to desktop, say goodbye to modularity and say hello to heavy manufacturer lockdown and hardcoded lifespan, it will be a miracle if their 50 gorrillions transistors chips last more than 3 years...
            tech illiteracy and people not realizing what is taken from them will be the cause of our doom, sacrifying freedom and user/environment-friendliness for a bit of conveniency and a few worthless bigger numbers will be the cause of our doom...
            yeah you are right it's pathetic how they limit modularity and kill sustainability smh

            Comment


            • #36
              Wish Valve sponsored Box64 or the like to run full x86_64 Steam and games on these!

              Comment


              • #37
                Originally posted by rmfx View Post

                Yes, finally this bloated over complicated obsolete closed crap can get the retirement it deserves.

                Finally a world where innovation is not locked because of an evil duopole!
                The dynamics of AMD and Intel has been the reason x86 has taken down competitors like PowerPC and ARM. Who is exactly competing with Apple? Who exactly is competing with Qualcomm on laptops? It's not like Apple hasn't already made stupid claims of how much faster their new ARM based M chips are compared to previous generation. Meanwhile nobody bought Apple's M2's when it was barely an upgrade over M1's. Wonder why Apple is sticking M2's in iPad's in 2024 and a $3,500 Vision Pro? Nobody bought them, that's why. Why is nobody buying them? So far in 2024 50% of software is based on Apple's ARM.

                I can guarantee you that x86 ain't going anywhere. ARM might, as it was bankrupt recently and almost got bought by Nvidia. AMD is already as efficient as Apple, and Intel's new Lunar Lake looks to even outclass Apple and Qualcomm at power efficiency. Meanwhile Apple was losing 34% Mac sales year over year, probably because people realized that an ARM based Mac is stupid. You think Intel is worried about Apple or Qualcomm? Their main competitor is AMD. Apple has to compete with themselves and the same goes for Qualcomm.

                Last edited by Dukenukemx; 13 June 2024, 01:51 AM.

                Comment

                Working...
                X