TIL about Pixelfed :) I will be signing up.

I don't know what I want for dinner and that never stopped me from identifying myself.

Yes. The intelligent multi-device-type feature is a huge improvement for any workload that needs more space than what an SSD can affordably provide, even moreso with the reliability of eg RAID1.

Before that I had to use BTRFS (RAID1) on bcache (not fs) devices, but half of the cache space was being wasted on the redundant copies because the two systems operate independently.

Yes

PyTorch actually works pretty well on Arch with opencl-amd and opencl-amd-dev (the official packages didn't work for me). I'm extremely happy with my new Radeon.

They didn't say anything about Jewish people

What's wrong with XMPP? I've been using it for many years, it's by far been the greatest experience and it has OMEMO encryption.

I wouldn't go as far as moreso but yes!

All the advantages of btrfs + the ability to combine SSDs and HDDs in a way that maximizes speed and space.

I am very happy with my ASUS ROG Zephyrus G14 GA402RJ. It's small, powerful, not too expensive, has a recent AMD GPU (RX6800S) / CPU (Ryzen 9 6900), upgradeable RAM, good resolution, and a gorgeous screen (great for photo development too).

I use Zim, a local wiki which uses text files with a markdown-like syntax, Syncthing to synchronize the notes, and Markor on Android supports this syntax.

It's pretty much never advantageous having Windows to run AI stuff, it runs like dog shit and the drivers are not prioritized because no one does serious AI research using Windows.

There are open models that one can download on HuggingFace and run locally, but they are not as good as ChatGPT4 which has had insane(r) amounts of resources thrown at.

That is creepy.

I don't think my grandma was a sysadmin.

I haven't had an issue with gaming on Linux in ages. Since the Steam Deck came out checking the compatibility of a game is an afterthought I do not need to worry about.

ChatGPT4

I don't understand your question. GPT is proprietary and hosted by OpenAI. There are other large language models (LLM) that one can download (or even train if they are open-source or at least have a descriptive scientific paper and open training data) and host themselves, but they are not as powerful.

That's not how it works anywhere I've heard of (and I've lived in the USA and Europe and worked in an American grocery store). Maybe if it's a refrigerated product in the US (or vandalized somehow), but that's not the case with dates.

Correct. I game a lot and I never have to worry about Linux compatibility anymore.

How does it compare to Zim?

Yes, lots of storage space with redundancy and the speed advantage of an SSD. If you have enough data where a pair of reasonably priced SSDs is not enough then it is highly advantageous to combine them with (cheaper/bigger) HDDs.

Personally I would not consider a filesystem without data redundancy for my personal files, and I have enough pictures to fill some hard drives but I don't like waiting for them to load.

It has RAID modes and it intelligently rearranges data s.t. commonly used files are stored in a fast drive and fetched from there, whereas BTRFS will write to and read from a "random" drive regardless of its speed.

The previous solution of using btrfs raid1 + bcache (not the FS) separately was very wasteful because the cache had to store both/all copies of the data since btrfs picks a random drive to read from.

I was very confused as to why you would need regular updates of, say, a SofleRGB, then I opened your links.

I use PyTorch daily at work and I'm happy to have gone AMD. Initial support for last gen was lacking but it's there now. It mostly works , the performance is great but it crashes on some unusual tasks.

*except games that don't run on proton/wine. I don't have a Windows installation and I never have to worry about a game not running out of the box on Linux anymore.

Install the ROCm version of packages and it mostly works (at least on Arch, and I assume Ubuntu since it has official packages from AMD.)

I had bcache + btrfs (RAID1) before this but it was a huge waste of space because bcache had to cache two identical copies of the data in order to be effective (since BTRFS and bcache don't communicate and BTRFS picks from a random disk); that's half as much cache.

With Bcachefs everything is integrated so it knows to cache only one copy in RAID1 (and it doesn't even need to hold two HDD copies, the fast/"cached" copy counts). Data is read from the fastest source and every resource is best utilized.

I tried that app but it crashes all the time on my phone from the moment it asked me to upload a photo.

Quick update to let you know that it works on all my tasks with ROCm 5.7

🖕