• 0 posts
  • 12 comments
Joined 2 years ago
Cake day: June 13th, 2024
  • Not watching a 15 minute video, just responding to your comment. For AMD hardware, Linux is great. Most games run with the same performance, many run even faster, than native Windows, despite needing a compatibility layer.

    Windows is the biggest bottleneck on PCs. The Windows scheduler is so bad that they have to bypass it and automatically pin fullscreen and game processes to CPU cores in order to get acceptable performance. They also recently updated Windows 11 to max the CPU frequency multiplier while mouse input is detected, in order to solve performance problems with responsiveness.

    Windows fans will point to benchmarks of UE5 games, which still run with marginally worse performance on Linux compared to Windows. I play several UE5 games on Linux and the performance meets my expectations. The difference is not noticeable unless you are watching a performance graph. Time will tell if further optimizations put Linux at the same or better performance as Windows with UE5 games, too, like it already is for other games.

    Pretty much the only games that cannot be played on Linux are games that deliberately disallow server connections from Linux clients or games that use Windows-only anti-cheat software. Everything else runs with good or great performance. This is the current state of gaming on Linux.

  • Maybe the statistical approximate just coincidentally lined up with the actual solution?

    Yes, right, and this can happen. I didn’t say they are a bad approximation. LLMs may be the most advanced and sophisticated statistical models ever created (if there are other examples of statistical models that are more sophisticated, I would love to learn about them). But given what an LLM actually objectively is, a statistical model of the next token to follow from a sample of language, what other explanation could there be?

    We need to keep in mind what the tools that we are working with actually are.

    As a code generator, they can produce great results, especially simple stuff like generating a script or some function implementation. Once you get to software engineering tasks like designing system architecture and designing maintainable code, it starts to fall apart really fast. You end up doing all of the work for it in natural language and just using the LLM for a usecase that it is actually great for: translation, from detailed spec to code.

  • 8GB vram is perfectly acceptable for the performance level of this box.

    I think the argument being made is that the performance level of this box is not really acceptable for a game-console-replacement product in 2026.

    The performance is pretty bad when compared to the PS5 Pro, which is much cheaper. And comes with a controller.

    You can’t bring your Steam library to a PS5, which might be a significant factor, but also if you’ve heavily invested in a library of Steam games, you probably already have a gaming PC, this probably isn’t in any way an upgrade from it, which makes it a pure convenience product for $1k+.