• 0 Posts
  • 30 Comments
Joined 3 years ago
cake
Cake day: July 1st, 2023

help-circle
  • Adding to this (which is a solid recommendation and answer BTW), you can try out podman kube play <your-file>.yaml (see here) before going full k8s or k3s setup to familiarize yourself with the concepts, without moving too far away from the docker-compose ease of use.

    Regarding question 1, any distro works, but if your are looking specifically for a lightweight, fast to deploy node host os, I recommend opensuse microOS/leap micro or similarly, fedora coreOS. With both you can drop a combustion/butane/ignition config file in a usb installer partition, so you can quickly integrate fresh installs in your cluster (ssh, network config, user accounts, package installs) see https://opensuse.github.io/fuel-ignition/





  • Hmm. I had pretty much the same experience, and wondered about having multiple conversation agents for specific tasks - but didn’t get around to trying that out. Currently, I am using it without LLM, albeit with GPU accelerated whisper (and other custom CV tasks for camera feeds). This gives me fairly accurate STT, and I have defined a plethora of variable sentences for hassil (intent matcher), so I often get the correct match. There is the option for optional words and or-alternatives, for instance:

    sentences:
     - (start|begin|fire) [the] [one] vaccum clean(er|ing) [robot] [session]
    

    So this would match “start vacuum”, but also “fire one vacuum cleaning session”

    Of course, this is substantial effort initially, but once configured and debugged (punctuation is poison!) works pretty well. As an aside, using the atom echo satellites gave me a lot of errors, simply because the microphones are bad. With a better quality satellite device (the voice preview) the success rate is much higher, almost flawless.

    That all said, if you find a better intent matcher or another solution, please do report back as I am very interested in an easier solution that does not require me to think of all possible sentence ahead of time.


  • With this new driver, system LEDs, audio LEDs, extra keyboard keys, keyboard backlight, and other features are working.

    And this over a year after this was released. Our whole office skipped this gen for new hardware because Linux support wasn’t ready. Additionally, reports were that performance on Linux was/is abysmal for the capabilities. Generally, I feel it was a mistake to prioritize all new ARM and AI CPUs for windows, with lagging and shit linux support until now, as mostly enthusiastic would-be customers are AI devs/researchers, and they often prefer some Linux variant as it “just works” with most tooling. The ‘normie’ office windows user does not give 2shits about locally accelerated inference. Why chipmakers fumbeled the ball so badly with the new AI accelerator / NPU CPUs is beyond me.






  • Maybe, but as someone who spent a summer school breaks worth of time in 2002 getting drivers for a Nvidia GeForce 2 card to run under Mandrake (oh the kernel panics…) to play counter-strike 1.X on wine… It’s come a long fucking way.

    I use Debian for everyday work and on my private machine nowadays and struggle with the shitty experience of windows when helping someone out now and then. Granted, I don’t have much time for games these days, and often fire up the PS for that, but I feel experience can vary as long as you know what you want and manage expectations.




  • PostgreSQL is definitely a boost to performance, especially if you offload the DB to a dedicated server (depending on load, can even be a cluster)

    Nevertheless, it probably has much to do with how it’s deployed and how many proxies are in front of it, and/or VPN. If you have large numbers of containers and small CPU/low memory hardware, and either running everything on one machine or have some other limitations, it’ll be slow.

    Admittedly, I’m not very familiar with the codebase, but I feel Apache isn’t improving the speed either. Not exactly sure how PHP is nowadays with concurrency and async, but generally a microservice type architecture is nice because you can add more workers/instances wherever a bottleneck emerges.