ColonelPhantom

Do you have more specifics? I just tried it on my machine (Fedora 42, Plasma 6.5.1 Wayland, Konsole 25.08.2, Radeon 780M) and it seems fine for me. Does it only occur occasionally/under specific circumstances for example?

What is broken for you? At this point, starting from roughly KDE 6, Wayland has been pretty much flawless for me. KDE 5.27 was pretty much fine already as well.

Not sure if ditching infinity cache/L3 for a bigger L2 would make sense. It seems at least plausible to me that growing the L2 cache that much would make its latency worse, thus losing most of what has been gained. In the image you linked, Nvidia's L2 latency is like twice as high as AMD's!

Keep in mind that the 'big' desktop chips have bigger caches too. The L2 on a 5080 and 'L3' on a 9070 XT are both 64 MB. Additionally, AMD has already been growing the L2, going to 8 MB on the 9070 XT vs. 4 MB on the 7800 XT and 6800 XT.

Well, sure, but from what I know, humans are way better at following 'implicit' instructions than LLMs. A human programmer can 'infer' most of the important basic rules from looking at the existing code, whereas all this agents.md/claude.md/whatever stuff seems necessary to even get basic performance in this regard.

Also, the agents.md website seems to mostly list README.md-style 'how do I run this instructions' in its example, not stylistic guidelines.

Furthermore, it would be nice if these agents add it themselves. With a human, you tell them "this is wrong, do it that way" and they would remember it. (Although this functionality seems to be worked on?)

Out of curiosity, how much of the 138K LUTs (as well as other resources like BRAM) are in use here? I wonder if there's much room to add fancy peripherals, or perhaps "growing" the CPU to achieve better IPC.

But that just raises more questions; why would Cognition be better positioned for B2B? Wouldn't Cursor have an advantage as they already have a foothold in the B2C market? Expanding from there seems a lot easier than coming for B2B directly: Cursor already has mindshare among developers.

The B2B dominance of Microsoft Office can, after all, not exist without B2C Microsoft Office.

Is 86box period-accurate for Windows XP? It seems to go up to Socket 370 (Pentium III) which is very early for XP. Additionally, from my experience using PCem (which 86box is based on), you need pretty beefy hardware to emulate a Pentium II or K6, let alone to go beyond them. And those are period-accurate for Windows 98, not XP.

I would say "kei" does pretty specifically refer to vehicles adhering to those Japanese regulations. I think "minitruck" or "compact truck" would be a better, more general name.

I guess modern crash safety does require decent crumple zones, but I'm not sure in how far Europe is different than North America in this.

If anything, small vehicles aren't a thing in NA, but extremely popular still in Europe, even though SUVification is also happening here.

There's plenty of small cars left, like the Toyota Aygo X. Renault is also working on a new electric Twingo, and the new 5 isn't huge either.

I don't think you understood what a "copilot" and a "HUD" is. Confusingly, GitHub Copilot (the original one that suggests completions) is basically a HUD. On the other hand, agents that you give a task are clearly copilots that work via a natural language interface.

The article also mentions that agents are copilots:

> Here’s another personal example from AI coding. Let’s say you want to fix a bug. The obvious “copilot” way is to open an agent chat and ask it to do the fix.

Calling GC time "additional overhead" is not entirely correct, imo. Non-GC languages often spend plenty of time dealing with malloc/free, and I consider both GC and malloc/free to be "memory management". Both are non-trivial!

GC usually also comes in "copying" variants, which can have advantages like reducing memory fragmentation.

I feel like a massive counterexample here is embedded. Hardware companies tend to laugh at maintenance (it's expensive and extends the life of their products, so you don't have to give them money as often). If Linux was not GPL, many embedded platforms like routers or smartphones would not have kernel code available.

I disagree, because there is no 'easy' way to know what the most 'significant' changes are. I can think of a few heuristics (e.g. most changed, earliest changed, as well as prioritizing things like header files for C and friends), but nothing that would work universally or particularly well.

Wouldn't it be feasible to add a tiny battery or capacitor? Assuming the radio doesn't need to transmit continuously it can be powered via those which are then powered by the "forever battery".

Why is giving precise instructions bad? I would expect LLMs to be pretty good at following instructions after three years of training them that way. Plus, if the instructions are precise enough and therefore each step is simple enough, I would expect everything it needs to do to be 'in-distribution'.

Devcontainers don't "feel" local to me, since my CLI tools are not available in them. Think ripgrep or fd, for example.

I think the other poster is confused. Both Gemma 3 and Gemma 3n are open-weight models.

Google's proprietary model line is called Gemini. There is a variant that can be ran offline called Gemini Nano, but I don't think it can be freely distributed and is only allowed as part of Android.

As for what's new, Gemma 3n seems to have some optimizations done to it that lead it to be better than the 'small' Gemma 3 models (such as 4B) at similar speed or footprint.

What do you mean by Anthropic shifting their ordering? It seems to still be consistently Opus > Sonnet > Haiku. They didn't release 4 Haiku, but they also didn't release 3.5 Opus, and pricing wise Sonnet 4 lines up with earlier Sonnets.

As for this Gemma release, I don't think Gemma 4 would be an appropriate name. 3n is limited to very small versions (like 8B total parameters) and is therefore likely less powerful than Gemma 3.

From my impression this is more like a "Gemma 3 Lite" that provides a better speed/quality tradeoff than the smaller Gemma 3 models.

I believe that PyTorch already uses Triton; I recently tried to do torch.compile on a Windows machine and it did not work because the inductor backend relies on Triton.

I believe the override for GP's 7600 is 1100 or 11.0.0 as GFX1030 is RDNA2 (6800 XT).

What is a 'procotol algebra'? I googled the term but didn't find anything that seemed relevant.

Ollama serves multiple versions, you can get Q8_0 from it too:

ollama run deepseek-r1:8b-llama-distill-q8_0

The real value from the unsloth ones is that they were uploaded before R1 appeared on Ollama's model list.

What are some cases where 'they/them/their' does not work, but 'he/him/his' does?

As for neopronouns like 'zir', I don't think I ever actually encountered anyone who uses them, either online or irl, with the exception of one Twitter user who preferred 'it'. So I mostly see them as a non-issue, although if someone does prefer a neopronoun I don't see the point of making a fuss either.

This article really grates me the wrong way. A company selling an AI product saying the following about AI sceptics:

> All you crazy MFs are completely overlooking the fact that software engineering exists as a discipline because you cannot EVER under any circumstances TRUST CODE.

is straight up insulting to me, because it effectively comes down to "use my product or you're a looney".

Also, while two years after this post (which should be labeled 2023) I've still barely tried to entirely offload coding to an LLM, the few times I did try have been pretty crap. I also really, really don't want to 'chat' with my codebase or editor. 'Chatting' to me feels about as slow as writing something myself, while I also don't get a good mental model 'for free'.

I am a moderately happy user of AI autocomplete (specifically Supermaven), but I only ever accept suggestions that are trivially correct to me. If it's not trivial, it might be useful as a guide of where to look in actual documentation if relevant, but just accepting it will lead me down a wrong path more often than not.

Yiannopoulos is an... interesting case in general. Apparently[1] he declared himself to be "ex-gay", 'demoted' his husband to housemate, and is treating his homosexuality 'like an addiction'. His future plans include 'rehabilitating conversion therapy'.

Seeing all of that, I'm really not sure his boat has been rising with the tide, so to speak. I personally don't believe anyone thinks conversion therapy is good for themselves unless they are deeply troubled.

[1] https://www.lifesitenews.com/news/activist-milo-yiannopoulos...

Calling a person by the wrong pronoun is insulting, simple as that. The social convention is: "I say 'he' if and only if the person in question is male". Thus, calling for instance a transgender woman 'he' or a cisgender man 'she', is saying you don't see them as a man/woman, thus denying a part of their identity, just like when you call them by the wrong name.

I'm sure you would agree that a man with long hair or whatever should be called 'he', so I also don't see why that doesn't extend to transgender men for example.

According to TechPowerUp, idle power is absolutely fine, but only if you enable ASPM: https://www.techpowerup.com/review/intel-arc-b580/38.html

I guess GN did not enable ASPM?

> On long roads where every other country would give right of way to the long road because it works together with the natural psychology of driving on a long continuous road, in the Netherlands they will give right of way to small side streets.

Where was this? In my experience, this only applies to 'access roads', smaller roads not intended for through traffic. Roads intended to move traffic over longer distances do generally prioritize through traffic by giving right of way.

The Max also has much more memory bandwidth. Especially for running local LLMs, that is the limiting factor much more than the number of GPU cores is.

Firefox also has a "Copy without URL tracking" option in the URL bar context menu. It's not one-keypress, but clearly a tracker stripper is not Arc-exclusive.

Also no idea what you are talking about with Zen's "Mozilla underpinning sabotaging the experience".