There is actually less to ’xkill’. It nukes the X window from orbit in a very violent manner. The owning process(-tree) will usually just instantly curl up and die.
The main benefit is that it doesn’t actually kill the process, it only nukes the window. As such, you can get rid of windows belonging to otherwise unkillable processes (zombies, etc).
Also, it’s fun. Just don’t miss the window and accidentally kill your WM. (Beat that Wayland)
Now consider that most enterprises are about five years behind that. Takes a few years before what’s available in Fedora trickles down to RHEL, and a few more years before it’s rolled out to clients. Ubuntu is on a similar timeline.
The fixes you got two years ago might be rolled out in 3 years in these places. Oh, and these are the people forking up much of the money for the Wayland development efforts. The current state of Wayland if you pay for it is kinda meh.
I’ll bite. It’s getting better, but still a long way to go.
But what do I know, I’ve only deployed and managed desktop linux for a few thousand people. People were screaming about these design flaws back in 2008 when this all started. The criticisms above were known and dismissed as FUD, and here we are. A few architectural changes back then, and we could have done this migration a decade faster. Just imagine, screen sharing during the pandemic!
As an example, see Arcan, a small research project with an impressively large subset of features from both X11 and Wayland (including working screen sharing, network transparency and a functioning security model). I wouldn’t use it in production, but if it was more than one guy in a basement working on it, it would probably be very usable fairly fast, compared to the decade and half that RedHat and friends have poured into Wayland thus far. Using a good architecture from the start would have done wonders. And Wayland isn’t even close to a good architecture. It’s just what we have to work with now.
Hopefully Xorg can die at some point, a decade or so from now. I’m just glad I don’t work with desktops anymore, the swap to Wayland will be painful for a lot of organisations.
Rough start? It’s been over a decade and it’s still rough.
Biomedical AI literally won the Nobel prize last year. But LLMs won’t help at all.
Tangentially related, any biomedical outfit that hasn’t bought a shitton of GPUs to run alphafold on is probably mismanaging money.
You have FreeIPA if you want a ”product”.
But honestly, if I, as a Linux admin, would do this kind of thing at this scale, I’d probably elect to remain on AD.
Here be dragons. But basically:
Run a VM from contents of a physical disk: use ’dd’ to create disk image. If on linux, try to boot and fix all the errors, hopefully few.
Run VM as physical machine: other way around.
You won’t find this in a tutorial. You need to understand concepts, read manuals, fit everything together, execute, fail and retry until it works.
For Windows, I have no idea. Conceptually, I figure it’s similar.
You mean a transparency log? Just sign and publish. Or if it’s confidential, have a timestamp authority sign it, but what’s the point of a confidential blockchain? Sure, we han have a string of hashes chained together á la git, but that’s just an implementation detail. Where does the trust come from, who does the audit? That’s the interesting part.
If your blockchain isn’t distributed, it doesn’t need to be a blockchain, because then you already have trust established.
It mostly affects people working with ”fun” enterprise hardware or special purpose things.
But to take one example, proprietary drivers for high performance network cards, most likely from Nvidia.
You assume a uniform distribution. I’m guessing that it’s not. The question isn’t ”Does the model contain compressed representations of all works it was trained on”. Enough information on any single image is enough to be a copyright issue.
Besides, the situation isn’t as obviously flawed with image models, when compared to LLMs. LLMs are just broken in this regard, because it only takes a handful of bytes being retained in order to violate copyright.
I think there will be a ”find out” stage fairly soon. Currently, the US projects lots and lots of soft power on the rest of the world to enforce copyright terms favourable to Disney and friends. Accepting copyright violations for AI will erode that power internationally over time.
Personally, I do think we need to rework copyright anyway, so I’m not complaining that much. Change the law, go ahead and make the high seas legal. But set against current copyright laws, most large datasets and most models constitute copyright violations. Just imagine the shitshow if OpenAI was an European company training on material from Disney.
Stability and standardisation within the kernel for kernel modules. There are plenty of commercial products that use proprietary kernel modules that basically only work on a very specific kernel version, preventing upgrades.
Or they could just open source and inline their garbage kernel modules…
Document databases are the future /s
Or you know, trusted timestamps and cryptographic signatures via normal PKI. A Merkle tree isn’t worth shit legally if you can’t verify it against a trust outside of the tree.
All of the blockchain bullshit miss that part - you can create a cryptographic representation of money or contracts, but you can’t actually enforce, verify or trust anything in the real world without intermediaries. On the other hand, I can trust a certificate from a CA because there are verifiable actual real-world consequences for someone if that CA breaks legal agreements.
I’ll use a folder of actual papers, signed using a pen. Have some witnesses, make sure they have a legal stake and consequences, and you are golden.
Distributed blockchains are useful when all of the below are fulfilled:
Here, we have a single peer creating entries in a ledger. We can get away with a copy of the ledger and one or more trusted timestamping authorities.
There is an argument that training actually is a type of (lossy) compression. You can actually build (bad) language models by using standard compression algorithms to ”train”.
By that argument, any model contains lossy and unstructured copies of all data it was trained on. If you download a 480p low quality h264-encoded Bluray rip of a Ghibli movie, it’s not legal, despite the fact that you aren’t downloading the same bits that were on the Bluray.
Besides, even if we consider the model itself to be fine, they did not buy all the media they trained the model on. The action of downloading media, regardless of purpose, is piracy. At least, that has been the interpretation for normal people sailing the seas, large companies are of course exempt from filthy things like laws.
What? Just base64 encrypt it before you store it in the git hub
I guess we can formulate a law:
Pick any two.
I’m using ”Commercially deployed” in the context of ”company you interacted with had an AI represent them in that communication”. You don’t use AI for that to increase costumer satisfaction. (I wonder why I haven’t seen any AI products targeted at automated B2B sales?)
I won’t argue that GenAI isn’t useful for end consumers using it properly. It is.
(As an aside, I hope you and your grandfather get better!)
Please note that the nominal FLOP/s from both Nvidia and Huawei are kinda bullshit. What precision we run at greatly affect that number. Nvidias marketing nowadays refer to fp4 tensor operations. Traditionally, FLOP/s are measured with fp64 matrix-matrix multiplication. That’s a lot more bits per FLOP.
Also, that GPU-GPU bandwidth is kinda shit compared to Nvidias marketing numbers if I’m parsing correctly (NVLink is 18x 10GB/s links per GPU, big ’B’ in GB). I might read the numbers incorrectly, but anyway. How and if they manage multi-GPU cache coherency will be interesting to see. Nvidia and AMD both do (to varying degrees) have cache coherency in those settings. Developer experience matters…
Now, the real interesting thing is power draw, density and price. Power draw and price obviously influence TCO. On 7nm, I guess the power bill won’t be very fun to read, but that’s just a guess. The density influences network options - are DAC-cables viable at all, or is it (more expensive) optical all the way?