m-schuetz

Problem is, NVIDIA has so many quality of life features for developers. It's not easy getting especially smaller scale developers and academia to use other vendors that are 1) much more difficult to use while 2) also being slower and not as rich in features.

Personally I opted in to being NVIDIA-vendor-locked a couple of years ago because I just couldn't stand the insanely bonkers and pointless complexity of APIs like Vulkan. I used OpenGL before which supported all vendors, but because newer features weren't added to OpenGL I eventually had to make the switch.

I tried both Vulkan and CUDA, and after not getting shit done in Vulkan for a week I tried CUDA, and got the same stuff done in less than a day that I could not do in a whole week in Vulkan. At that moment I thought, screw it, I'm going to go NV-only now.

Agents work great for tasks that thousands of developers have done before. This isn't one of those tasks.

Vulkan has abysmal UX though. At one point I had to chose between Vulkan and Cuda for future projects, and I ended up with Cuda because a feasibilty study I couldn't get to work in Vulkan for an entire week, easily worked in Cuda in less than a day.

Man, I love Kagi. Two years ago I would never have thought I'd ever pay for a search engine, but the option to block garbage domains like userbenchmark or sites with purely AI generated content is just too good.

Security through obscurity is like a bike lock. It can be cracked with the right tools and effort, but massively improves security compared to leaving it out unlocked.

Checking different models once every quarter is exactly what made me move to claude code.

Easy and "good enough" very often trumps complex and comprehensive. Markdown is great precisely because it can't do a lot of things.

Scanning your computer is an entirely different thing than scanning browser extensions. By maximizing the expectation via "Illegally searching your computer", the truth suddenly appears harmless.

All I want from C++29 is a single-line random() function.

> For low level stuff plain C

The nice thing about C++ is that you can more or less turn it into C, if you want. My C++ code is closer to C than idiomatic, modern C++, but I wouldn't want to miss the nice parts that C++ adds, such as lambda functions and the occasional template for generalization. Pretty much the only thing I'm missing from C are order-independent designated initializers, which became order-dependent in C++, and thus useless.

> "Haven't seen anything I'd rather used" reads like "Haven't gotten over the initial learning curve with any other tool"

What an odd thing to say. I simply don't like certain design decisions in other languages that I've checked out and tried, and therefore do not see any reason to switch. E.g. I tried Rust, but it's absolutely terrible for quick&dirty prototyping, which is my main job.

C++ isn't great but so far I haven't seen anything I'd rather use.

The idea is great, the execution is terrible. In JS, modules were instantly popular because they were easy to use, added a lot of benefit, and support in browsers and the ecoysystem was fairly good after a couple of years. In C++, support is still bad, 6 years after they were introduced.

I actually prefer browser-based tools. Havent installed hex editors since hexedit for web is available; Photopea covers my occasional photo/picture editing needs, google sheets/slides/docs is my main office suit, etc. It's great to not have to install things.

From my experience, accessing system RAM from the GPU is so slow, it might as well count as "does not work". It's orders of magnitudes faster to memcpy large swaths of memory that you are going to use to the GPU, rather than accessing system mem from a kernel which then takes ages to wait for that small block/page of memory, then waits again for the next small page/block of memory, etc. Latency hiding doesnt work anymore if the latency is that large.

Personally I like C+. Picking the nice parts of C++, but skipping all the nonsense. I just wish C++ hadn't deliberately screwed up designated initializers with mandatory ordering. The C version of it that allows out-of-order assignments is clearly superior.

> Somehow vscode (including their renderers) manages to come in pretty low compared to even Apples native apps.

Because the issue isn't electron, it's not freeing resources which you can do in any language/platform.

(one who recommended surgery here)

I tried hyposensibilization therapy, and while it worked for seasonal birch pollen issues, it didn't work for dust mites, oral allergies, and chronically stuffed sinuses.

It basically means surgically removing parts of your chronically enlarged/swollen turbinates so that your airways are free again. Along with that I've had a nasal spur removed (slightly blocked airways); Septumplastic (had a slightly deviated septum that also inhibited airflow a bit); and while at it the ENT also recommended and did FESS (Opening/Widening some paths to the sinuses).

It was a pretty life changing surgery that finally allowed me to properly sleep again, and do exercises/run while breathing through the nose. For some people, the turbinates may become enlarged again after a while, but for me it's been great for two years already.

Depending on how bad it is for you, I can recommend thinking about turbinectomy. Had it done due to chronic, allergy related swellings, and it was life changing.

Evolution just needs people survive long enough to reproduce. If they get sick afterwards, it doesn't care.

I'm pretty allergic most of the time (lots of birch cross allergies and dust mites), but sometimes when I'm sick the allergic reactions appear to go down. Allergies can be pretty weird.

That's weird, I switched away from ChatGPT because I mostly got superior results from Gemini and Claude.

I liked using singletons back in the day, but now I simply make a struct with static members which serves the same purpose with less verbose code. Initialization order doesn't matter if you add one explicit (and also static) init function, or a lazy initialization check.

This is way more cumbersome than mmap if you need to out-of-core process the file in non-sequential patterns. Way way more cumbersome, since you need to deal with intermediate staging buffers, and reuse them if you actually want to be fast. mmap, on the other hand, is absolutely trivial to use, like any regular buffer pointer. And at least on windows, the mmap counterpart can be faster when processing the file with multiple threads, compared to fread.

But I agree that it's a bizarre article since mmap is not a C standard, and relies on platform-dependend Operating System APIs.

Pretty much any non-web GUI framework I tried so far has either been terrible to set up, or terrible to deploy. Or both. Electron is stupidly simple.

ImGUI is the single exception that has been simple to set up, trivial to deploy (there is nothing to deploy, including it is all that's needed), and nice to use.

> Isn't it time to throw the browser away, stop abusing HTML to make applications, and design something fit for purpose?

Not going to happen until gui frameworks are as comfortable and easy to set up and use as html. Entry barrier and ergonomics are among the biggest deciding factors of winning technologies.

Some other words that are sorely missing from dictionaries: "Warm water", "hot water", "cold water", "dirty water"

Not just recently. Google image search was semi-useless for years due to being spammed by login-walled pinterest images.

And yet I got better results with Gemini than with Claude Opus.

I tried Claude Opus but at least for my tasks, Gemini provided better results. Both were way better than ChatGPT. Haven't done any agents yet, waiting on that until they mature a bit more.