Latest activity

Which one of them can compile itself from source like the BSD's? They all don't seem to like that. Afraid of a fork?

Well if you track "latest" in packages, the difference should be only a couple of days max (assuming that you build ports every day).

Neither is the network_interfaces, it gets those automatically too. The OP doesn't have those settings in his rc.conf though, so that can't be it.

Not what I think people will reply to that and my followups. In any case, use of an advanced local LLM is so slick that it is a 2-liner *assuming you start from working GPU drivers). Not much for the initial HOWTO post. I kindly ask to give...

You sure? The one I checked didn't have that routing error message anywhere. Do you have any jails running?

This sounds like a candidate for "Howtos and FAQs".

This seems certain to be true. Gotta keep the brain juices flowing, whether you like your chatbot or not.

I plan to post how to run GPU-accelerated local LLMs on FreeBSD right now, which models to use for coding, how different hardware trades off. How huggingface works, how llama.cpp moved the caches around. Moving into post-training (on FreeBSD) soon.

View: https://www.youtube.com/watch?v=lXg0Efj4EW8

This nonsense wouldn't have existed if p2p filesharing networks weren't killed for the profit loss of media giants. I believe the OSS world has to seize control again like in limewire/dc++ times. If companies are legally permitted to download...

I found the article about connecting a normal phone to a FreeBSD server (via an ATA adapter and Asterisk) very interesting. I love projects like this!

do you have pf_enable="YES" in /etc/rc.conf? This ends up executing pfctl -e among other things. You can check pfctl -si to see if it is enabled.

Might make cool wallpapers, too. https://www.nasa.gov/gallery/artemis/

Original article here. Consider this when replying. FreeBSD, The FreeBSD Foundation, and The FreeBSD Forums are not associated with the content of this article.

Yes.

Thank you for posting a summary and not just a link!

Original article here. Consider this when replying. Open source infrastructure depends on more than new features. It also depends on the steady, often unseen work of identifying risks, improving processes, and making systems easier to maintain...

Just tried that on a dual screen system: A window manager is already running on screen 0. So this needs a special xorg.conf with 2 screen sections?

What you say seems to contradict itself. Either way, llama.cpp with hardware acceleration through Vulkan works out of the box with the packge.

Nightmares on wax... View: https://www.youtube.com/watch?v=oRIgjKn2v9s

Might make cool wallpapers, too. https://www.nasa.gov/gallery/artemis/

I'm not using IPFW myself, but what I gather from the FreeBSD handbook this service can use 2 types of logging: either through syslog or using ipfw0. So my guess is that your service uses its own logging.

Wrong direction... View: https://www.youtube.com/watch?v=3gWzt4x_FG4

I used Zaphod mode until recently. With fvwm2. Things can be confusing since there is only one window manager for both displays.

also there are lots of tools like this in ports / packages, net-mgmt/darkstat for example.

Hi! I've been in Linux for 5/6 years. I started quite early with Arch, been a while, and than switch to Artix (no systemd). Recently I discovered Alpine Linux, that is a great OS, but I started feeling the Linux universe was not enought for me...

Never tried 2 Xorg servers 1 computer but with a root window overlapping 2 displays, I comfigured openbox to start different programs always on the same monitor depending on the executable name. It also supports keyboard shortcuts to move...

bakul, Emrion, guys, thank you so much for such valuable information! That's really interesting, I never even thought that just using a pointer can make a considerable difference.

When in doubt about something then don't forget about the -o option for ls: ls -lo: peter@zefiris:/var $ ls -lod e* dr-xr-xr-x 2 root wheel schg 2 Nov 29 2024 empty/ See?

How do you produce that graph?

I plan to post how to run GPU-accelerated local LLMs on FreeBSD right now, which models to use for coding, how different hardware trades off. How huggingface works, how llama.cpp moved the caches around. Moving into post-training (on FreeBSD) soon.

Once, I stumble on this question: why a pointer of pointer and not a pointer for the next element? I talk about double chained lists. I read that to write the base functions of such a list needs less code and will be faster. So, I coded myself...

The best way to understand this is by drawing a picture of such a queue. The head has two ptrs, one to head and one to the next field of the last entry. Each entry has a next pointer *after* 3 integer fields, for example. Now draw pointer...

That can stay in offtopic. The forums are here to help people installing and configuring various components on FreeBSD. We're not a generic discussion platform. That's not how it works.

That's not possible. Regardless of whether the pool is manually or automatically imported, the zpool-import log is recorded in any case in the pool history. There is no option to exclude it.

I'm honestly looking for posts that would fit, it's rather pointless to create a new forum if there's no content in it. Most of the AI posts have been rather generic. Not that much relating to configuration or setting it up on FreeBSD. Even...

Or a firewall that's blocking it, or the packets aren't NAT'ed and thrown onto the internet with an RFC1918 source address (so you will never receive the responses). In any case, the local network works (you can ping the gateway and other hosts...

Dodgy internet connection? And a service that tries to call home, but ends up with a couple of retransmissions due to the bad connection? Not everything is within your control.

I had two brothers. Both passed away suddenly (a few years apart), but fortunately. my nieces and nephews knew their passwords. I have a death document for my wife, and a script so she can reboot into xfce and click to get to necessary...

I guess I start to sort of understand it, but it still is not very clear to me, where the optimization is... Could you please elaborate a bit? Yeah, I already did that, that's why I actually ended up having this question - this is how I saw that...

stqh_last points to the next *field* of the last entry. This is a slight optimization. The macros are hard to read but you can define a sample tail queue and operations on them using these macros, then look at the output of cc -E to make sense of...

Original article here. Consider this when replying. FreeBSD, The FreeBSD Foundation, and The FreeBSD Forums are not associated with the content of this article.

Original article here. Consider this when replying. Open source infrastructure depends on more than new features. It also depends on the steady, often unseen work of identifying risks, improving processes, and making systems easier to maintain...

It's PCI/PCIe, the days of messing with IRQs, DMA and I/O addresses are long gone.

When in doubt about something then don't forget about the -o option for ls: ls -lo: peter@zefiris:/var $ ls -lod e* dr-xr-xr-x 2 root wheel schg 2 Nov 29 2024 empty/ See?

I'm exploring queue framework - both its public API and internal implementation in sys/queue.h and I'm a little bit confused about one thing in structure that is used for {S}TAILQ head: #define STAILQ_HEAD(name, type)...

Might make cool wallpapers, too. https://www.nasa.gov/gallery/artemis/

https://www.nasa.gov/gallery/artemis/ I know I posted this elsewhere cause I forgot about this thread.

As a hobbyist some years ago now I built 3 telecaster clones with great inspiration from TDRPI. Did all woodwork including neck and also winded the pickups. My amps these days are all software. Trying to migrate from Mixcraft with Guitar Rig...

Might make cool wallpapers, too. https://www.nasa.gov/gallery/artemis/