HN user
joseph dot kogut at gmail
Device trees are certainly less portable across hardware, and more difficult to create than just booting a generic kernel on a machine that supports ACPI. However, the device tree is shipped as part of the kernel, and as long as the bootloader correctly loads the DTB and kernel, it's no more difficult to swap out the userspace portion/rootfs than any other Linux system, generally.
Interesting to see the adjective "small" used to describe a single binary in the same amount of space used by an entire embedded Linux-based operating system.
Probably to interface with existing cars by replacing the engine, retaining everything after, including the existing transmission, differential(s), and suspension.
Mistiming the market is losing out on potential gains, which is losing money compared to sitting on cash. Cash doesn't grow.
Most people are best off investing in index funds and forgetting about it for 10+ years.
You in fact can now! In the past week, a transformer framework called FastFlowLM [0] supporting XDNA 2 NPUs officially started supporting Linux.
I posted it here the same day I found and started using it, to almost no reaction.
[0] https://github.com/FastFlowLM https://fastflowlm.com/ https://huggingface.co/FastFlowLM
On average according to Geekbench, the M5 compared to the 9950X is ~17% faster in single thread performance and ~30% slower in multithread performance.
Individual benchmarks tell the bigger picture. These two are optimized for different use cases, with Apple heavily leaning towards low latency single thread throughput with low sustained power usage.
https://browser.geekbench.com/v6/cpu/compare/16833358?baseli...
EDIT: The M4 Max compares much more closely https://browser.geekbench.com/v6/cpu/compare/16834801?baseli...
There's also a guide on the Framework community forums for getting this built and running on Arch Linux.
https://community.frame.work/t/guide-use-npu-xdna2-with-arch...
I think the point of the line of questioning is to illustrate that "tools" like a code interpreter act as scratch space for models to do work in, because the reasoning/thinking process has limitations much like our own.
There's a "newer" (almost ten years old) Unigine benchmark available called Superposition you should check out.
For anyone who doesn't know yet, there are a wide variety of ONVIF supporting cameras that you can setup with a local NVR running Frigate. You can block internet access to the cameras, so they can't create outbound connections, and only inbound connections to the video streams are allowed.
Tailscale has a free tier that's a good option to remotely access your network and cameras.
As a former user of Zulip at a previous company, thank you for this software, I enjoyed using it. Maybe I'll setup a private instance for friends and family so I can enjoy it once again.
There's a difference between carrying ten pounds small distances for short durations, and carrying an extra two pounds over twenty hours of travel, across multiple connecting international flights in a single day. It's also not just an extra two pounds, it's an additional proprietary power cord, bulk, more mass moving in and out under an airliner seat, it all adds up. Especially when you're sleep deprived and physically exhausted.
There's also ZLUDA, which can run llama.cpp and some other CUDA workloads already without any modification, but it's still maturing.
What kinds of applications do you use that require CUDA?
The Radeon RX 9070 XT performs at a similar level to the RTX 5070, and is retailing around $600 right now.
Note that I have retained the original title of the post, but I am not the author.
It may not, but that doesn't mean you need to install Windows. Any distro that packages Steam and the Nvidia driver will work.
EDIT: It appears to be supported for RTX 20xx and newer GPUs.
Unless you have an Intel Arc iGPU, Intel Arc B50/B60, or fancy server GPU, you won't have SR-IOV on your system, and that means you have to pass the entire GPU into the VM.
This is changing, specifically on QEMU with virtio-gpu, virgl, and Venus.
Virgl exposes a virtualized GPU in the guest that serializes OpenGL commands and sends them to the host for rendering. Venus is similar, but exposes Vulkan in the guest. Both of these work without dedicating the host GPU to the guest, it gives mediated access to the GPU without any specific hardware.
There's also another path known as vDRM/host native context that proxies the direct rendering manager (DRM) uAPI from the guest to the host over virtio-gpu, which allows the guest to use the native mesa driver for lower overhead compared to virgl/Venus. This does, however, require a small amount of code to support per driver in virglrenderer. There are patches that have been on the QEMU mailing list to add this since earlier this year, while crosvm already supports it.
Why do you have to install Windows? You could put bazzite or any other distro of your choosing on this machine and have a similar experience to the official Steam Machine.
Oddly, tameness has also long been associated with traits such as a shorter face, a smaller head, floppy ears and white patches on fur—a pattern that Charles Darwin noted in the 1800s.
Hmm, so evolutionary pressure of existing around humans makes animals cuter.
I wonder why we find these features endearing?
It would make sense, but Apple has large amounts of disdain for people having fun with their products. This evidenced by the large amounts of engineering they've put into very large, capable, and efficient GPUs, only to squander them on rendering web pages and liquid glass.
They released Apple Vision Pro with no ability to play popular PC games on it.
A VR headset. That doesn't play games.
As far as I understand, there's actually an intermediate driver on macOS that implements Vulkan on top of Metal, similar to how Proton implements Direct3D on top of Vulkan.
The available low-level API is Metal, and the existing software stack is written for Vulkan, so it makes more sense to implement Vulkan than to write a new Metal backend.
It's not at all about looks, it's about a different kind of handling, for off road, that's mutually exclusive with on road handling.
Yes, some people choose to emulate off road appearances, such as with fake bead locks and then only ever drive their vehicle on road. That doesn't discount the fact that there are a great many explicit choices you can make in designing and building a vehicle that sacrifice on road performance for off road performance.
What current mass production EVs use hub motors? It seems a lot more sensible to have the motors inboard, mounted to the chassis, and drive the wheel(s) with axle shafts. It seems in my searching this is how nearly all EVs are currently designed and produced.
Lifting an off road vehicle isn't ironic at all, nearly every characteristic that makes a vehicle good on road makes it bad off road and vise versa.
Increased height makes for increased ground clearance and improved break over angle. Sway bars are another suspension component that's great for reducing body roll on road at speed, but reduces articulation and ground contact off road. Differential lockers also negatively impact turning radius, and cause tire chirp, wear, and oversteer under throttle on road, while increasing traction off road.
What's silly is daily driving an off road vehicle on road, especially if you never take it off road.
Practically speaking, very few systems actually support SystemReady. There's an experimental port of edk2 for the Raspberry Pi, but some hardware is unavailable when using it.
It's not the fault of Linux for refusing to allow this.
Linux doesn't refuse anything, it's free and open source software. If publishers want to offer anti-cheat software developed for Linux, it will run. In fact, many games do have anti-cheat, like Insurgency: Sandstorm, which uses EAC through Proton.
Publishers can even develop invasive kernel-level anti-cheat just like they do for Windows. They don't because it's a small portion of the market currently, and I assume they consider it not worth the investment as of now. To what extent existing Linux users would willingly allow such software to run is also an open question.
There's also the point that even invasive kernel level anti-cheat on Windows with requirements for secure boot continues to be inadequate to stop cheaters in competitive online games.
Learning things is good. I also find I care more about the outcome and timeline than many professionals do, because I have to deal with the end result. That's not to say professionals can't or won't do the job better, or that they don't have more applicable experience to do the job more efficiently, but evaluating professionals is often as much or more work as learning how to do the job myself and just doing it. On average, the end result is at least better than a poor professional, sometimes as good or better than an average professional.
I wasn't trying to make any jabs, just an observation that getting outside of your comfort zone can be novel and interesting, even if it's mundane to people that commonly spend lots of time there.
I suppose when your accustomed level of abstraction is interpreted languages like JavaScript, and "the web", "gluing some libraries together in C" is a somewhat novel and interesting endeavor.