bayindirh

The funny thing is the default coffee filter I always get is marked as compostable on packaging. It's probably plastic-free as well, then.

I generally fill the cup to the same level, and fill it from the same source in the house in 99% of the time, so they are implicit constants in the equation, at least for me.

Thanks for the link. I have also found a small trove as a warm-up. If anyone interested, here it is:

https://github.com/zwang4/awesome-machine-learning-in-compil...

Using small, purpose built ML "things" to make things better makes sense. We use them everywhere. I never argued against it, as well.

> This is a lack of linux driver support for firmware designed for windows laptops.

Personally, I don't care. If you don't support open drivers, you have to provide first class drivers on Linux on day 1, because NVIDIA's majority of business is running on Linux already.

On top of that, they claim to have the best Linux support, and claim to have best drivers in their class.

So, it's better to act like that and provide support. If any other company has done that, they have been incinerated with the flame of internet's hate thrice already.

ML driven compilers are different than LLM assembly generation, and former is more sound than the latter one. I believe we both have enough experience to understand that.

However, starting with LLMs directly generating machine code and moving to ML directed optimization passes, you're moving the whole stadium with a crawler-transporter, not only the goalposts.

Helping compiler in a well defined way, and working on IR or driving it repeatedly to optimize the output is one thing, generating machine code from a training set is other. Let's not mix them.

Honestly, I'd not be that hard on myself, and making bad espresso is easy in comparison to making a good one, but you don't need to go influencer levels of rituals to get good espresso either.

A good espresso/americano has more bite in taste than a good Aeropress/V60/drip. I personally love smoother profiles (less acidity, less bite, etc.), so darker roasts and espresso on the bitter side is not my kind of thing, but preferences differ. OTOH, I can make myself a cup of instant coffee if I feel lazy.

Again, there's no need to punch yourself like that. Try, experiment and build your taste profile. These kinds of hobbies needs some kind of experience to refine what you like and what you don't.

Most importantly, don't let anyone's taste dictate yours and vice versa. I never get upset just because how someone I know prefers their coffee way different than mine, but they insist that I shouldn't drink that coffee, because that doesn't fit their taste preferences.

There's no wrong, but only things you like. You can't fail at this, because there's no way to do so.

Aeropress is accelerated drip coffee. It's not espresso by any means.

When you get used to the microwave you have, you can hit the same spot every time. It's a formula depending on your cup, the wattage of he microwave and time.

I can hit the temperature I want every time on a Bosch microwave from 80s which has two power levels and a rotary, mechanical timer.

If you don't want to use paper filters, you can use Swiss Gold, generic all-steel or cloth ones as well. Also there's Stanley's all steel pour-over.

When you want to go low or no plastics, there are many choices now.

You can't PGO that code, though. Also, I'm not sure it can generate architecture specific optimized (-march, -mtune) code as well.

Considering it just pulls and remixes blocks, this is a very tall order for today's LLMs, and maybe a couple of generations as well.

Calling GCC/clang at the back and emitting ASM is both cheating and has no advantages whatsoever as well.

It's not their first rodeo. Their cards already need to authenticate with the driver to enable the card completely. When I last checked, Nouveau driver has received a key which only enabled the hardware, allowing enough features to see a desktop and install closed NVIDIA drivers, nothing more.

Remember, the kernel module is open source, but rest of the driver is closed source, and any kernel level code which can't be open sourced but had to be is moved to card firmware. Only the kernel module is open source. Not the libraries, not the driver, nothing else.

NVIDIA is holding to its pearls like a sloth holding to its branch, because their all business model and livelihood is dependent on that control. I still commend AMD's decision to allow full card enablement with any driver which talks the correct language.

That's older than current AI/LLM/NVIDIA wave. Many application specific DSPs probably carry their model weights in an encrypted manner. If not, we would have seen the extremely performant face recognition and tracking models extracted from professional cameras already.

Note, these models generally run 60/120/240 FPS depending on the camera range, and can command the connected lenses at ~100FPS (because physics).

Debian security team used to offer support for testing, but as the distro grew and attack surface increased, that was dropped. However, security incidents are patched in unstable as well and they float up to testing relatively quickly (for a desktop system, anyway). When things freeze, security updates directly land on testing as a fixup polish.

However, while Testing is "relatively secure", it's never "Server secure". I'd never use testing or testing packages in stable or any server for that matter.

> I have to add that for me, testing definitely breaks stuff sometimes.

That's interesting, can you give an example? Asking as a curious Debian user.

I don't think we disagree a bit.

I believe I need to reiterate: Stable is for servers, testing is for desktops if you're not directly exposed to internet.

Again. Stable is for servers, testing is for personal use.

I never argued otherwise, ever.

I believe RHEL/Alma/Rocky has released patches today. I have a notice about they have done so.

Debian testing is way more stable than Ubuntu stable. In Debian's parlance, Testing means "release candidate", actually.

If you want real testing, you should run "unstable". If you're in for fun, try experimental.

Running Testing on my desktops and stable on my servers since forever. None of them broke, none of them created problems.

I don't care Stable running 6.x.x. It gets security updates in less than 24h, and is applied automatically before I wake up, and doesn't ship coreutils with 104CVEs, just because it's Rust.

If your infrastructure doesn't use landscape, kernel hot patching, JuJu or similar Canonical technology, or doesn't have its own Snap storefront, there's no need to install Ubuntu. Rest is just momentum.

Debian used to move a bit slow in terms of userspace, and that's true, but for servers, that's a non-issue for the last 20 years or so (from my experience).

On the Desktop side, Fedora and Ubuntu moves faster on the userspace front, but if your system is behind a NAT, and you know what you're doing, Debian Testing is just as good, more stable and it's rolling (except when it's feature freeze, but a stable workhorse is better than a shiny but breaking one).

Debian allows me to sleep well at night, provides guaranteed updates, and works rock solid all-round, so I won't be touching Ubuntu for the foreseeable future. Esp, with controversial choices like uutils, etc.

Every bug is someone else's workflow: https://xkcd.com/1172/

Also, extensive commenting inside codebases is useful documentation, incl. these kinds of incidents. Good code documents the how, but not why. Comments shall also include the "why" where applicable.

I have skimmed the post but some of the things are not minimalism, but replacement of delivery mechanisms. On the other hand I'm also doing something similar in my digital life.

I have reduced my social media to two sites: Here and Mastodon (with a very limited follow circle). I don't follow tons of ad-ridden websites. I have a self-hosted RSS reader, following 10 feeds. Instead of following junk tech news, I read LWN.

Next step might be enabling antiproc on HN to reduce my time here, since I added time limits on applications on my phone and love that artificial limit.

I have limited energy and I want to focus it instead of spreading it around. Otherwise progressing beyond a certain point becomes impossible.

It's just a conscious being confined in a cluster of GPUs. How dangerous can it be, eh?

It made me reconsider what I think about mathematics in general. So, if it was junk food, it was a nutritious one at that.

They formed an invincible duo on stage with Ian McKellen.

See: https://www.youtube.com/shorts/phwFI-n-big

I don't think so. People love to see how far they can go in some aspect sometimes. If I had the same amount of free time 5 years ago, I'd have tried the same thing with Go, because I like the language, and love the act of programming and seeing things.

Also, I also agree that people optimize without proper testing and profiling. If there's a real bottleneck here, trying to optimize it is the right call. OTOH, optimizing without profiling opens a lot of unnecessary cans.

Lastly, Go doesn't like constant allocation and deallocation of variables, that's true, but I believe knowing this and optimizing hot-loops for that goes a very long way as a start.

Humans are unsettled by fast, drastic changes since it generally signals uncertainty ahead. It's a caveman instinct signaling that someone needs to do much more to find the new normal or navigate some hardship.

RedHat and Ubuntu has kernel hot-patching. That's a case with binaries.

The whole reverse engineering and cracking scene is built upon this premise, you know.

I have Gemini because of my Google One membership. I already was on their 2TB plan and it gave me Pro level, IIRC, and it was generous. Now I'm on Ultra mostly because of Google Drive storage I get, but Gemini also gets a nice boost, as well.

I'm on $20 Claude plan, and it's also ample for my needs, for now.

People always love to think the AI race is a duopoly (OpenAI vs. Anthropic), but from what I have experienced, Gemini was always holding on its own, and I find myself using it more than Claude.

True, I use LLMs as glorified search engines to unearth material to read myself, but Gemini is much more useful if you decide to learn yourself. It also offers interesting expansions to your questions.

While I don't offload my work to LLMs (e.g.: Write this project, I'll be watching TV), I find them useful for closing my knowledge gap by collating information, and Google's Gemini is good at that.