Hacker News: nikonyrh

New comment by nikonyrh in "Your phone is about to stop being yours"

nikonyrh — Wed, 29 Apr 2026 12:11:26 +0000

At least in Finland's Nordea bank you can order a physical code calculator, they used to be small enough to keep on your wallet but the new one is the size of an old small phone. It even has a QR scanner. So I just keep it at home.

New comment by nikonyrh in "Ask HN: Does anyone have interests in anything besides AI?"

nikonyrh — Thu, 05 Feb 2026 06:18:17 +0000

Wood / metal working, MTB and Factorio. Or did you mean within the IT sector? Well ML is interesting ;)

New comment by nikonyrh in "Nvidia is dominating the S&P 500 more than any company in at least 44 years"

nikonyrh — Sat, 09 Aug 2025 21:51:04 +0000

AMD already has Composable Kernels[1], and supports for example Triton[2]. Then there is also HIP[3], and there are tools to automatically convert from from CUDA to HIP. But since CUDA is the de-facto standard, there is always friction to use something else (unless you need to support also AMD stack).

Making something just CUDA-compatible is non-trivial, and since Nvidia decides its direction and new features then the alternatives would always be lagging behind. Currently there are also major hardware differences between Nvidia and AMD, which may make highly optimized CUDA code inefficient or even buggy.

  [1] https://github.com/ROCm/composable_kernel?tab=readme-ov-file#composable-kernel
  [2] https://github.com/triton-lang/triton?tab=readme-ov-file#triton
  [3] https://github.com/ROCm/HIP?tab=readme-ov-file#what-is-this-repository-for

New comment by nikonyrh in "The Most Frustrating Customer Service Call of All Time"

nikonyrh — Sat, 09 Aug 2025 21:30:10 +0000

He should have asked what is the price for 1 MB, or 1000 kB. 2 cents, right? insert meme here

New comment by nikonyrh in "Modded-NanoGPT: NanoGPT (124M) quality in 3.25B tokens"

nikonyrh — Mon, 14 Oct 2024 00:27:59 +0000

I suppose you aren't a fan of the https://en.wikipedia.org/wiki/Hutter_Prize .

> The goal of the Hutter Prize is to encourage research in artificial intelligence (AI). The organizers believe that text compression and AI are equivalent problems.

New comment by nikonyrh in "How much energy does desalinisation use? Is it "absurdly cheap"?"

nikonyrh — Mon, 23 Sep 2024 21:51:45 +0000

I don't think the distinction between indoors or outdoors usage is relevant, in the end we are still talking about the volume of drinkable water people use. Whether they use it for drinking, cooking, washing, or watering the lawn.

New comment by nikonyrh in "Is it feasible for a regular American to keep their physical address private?"

nikonyrh — Thu, 04 Jul 2024 17:18:37 +0000

I always found it interesting that you cannot have a domain name without publishing your address and phone number, or paying for an extra service.

New comment by nikonyrh in "Ask HN: How can I sleep more and deeper?"

nikonyrh — Fri, 14 Jun 2024 01:15:21 +0000

This reminds me of the meme where you read up instructions of good sleep, at 4am before getting ready to go to bed.

New comment by nikonyrh in "Notebooks Are McDonalds of Code"

nikonyrh — Fri, 14 Jun 2024 00:37:05 +0000

"In a notebook, you can go back and edit and run individual lines of the notebook without re-running the whole notebook from the start and without re-computing everything that depends on what you just edited."

Isn't this a standard on REPLs as well? You can select the code you wish to run, and press Ctrl+Enter or what ever. I must admit, I've programmed Python for about 10 years in Spyder and VS Code now, but I haven't used notebooks at any point. Just either ad-hoc scripts or actual source files.

My definition of a "notebook" is an ad-hoc script, split into individual "cells" which are typically run as a whole. On my workflow, I just select the code I wish to run. Sometimes it is one expression, one line, 100 lines or 1000 lines depending what I've changed on the script.

New comment by nikonyrh in "Why today's phones are so boooooring?"

nikonyrh — Tue, 11 Jun 2024 19:56:20 +0000

I bought Unihertz Pocket Titan, at least it was different! Quite thick and heavy, but the battery lasts about 6 days even without any power-saving mode.

New comment by nikonyrh in "Ask HN: As someone with obscure CUDA knowledge, how to become AI developer?"

nikonyrh — Tue, 11 Jun 2024 18:18:56 +0000

Since with CUDA you are programming so close to hardware, and hardware (and CUDA itself) has advanced so much, I recommend you to go trough very carefully all the major CUDA versions and see how it has evolved. Well, strictly speaking I'm talking about different versions of "compute capabilities". Of course Wikipedia has a good summary: https://en.wikipedia.org/wiki/CUDA#Version_features_and_spec...

An other point is that you don't need to write any CUDA code to be able to utilize GPU computing. If you need ML models, you have frameworks like PyTorch and Tensorflow. You just need to express your mathematical problem, and the framework will take care of the rest.

Even if you needed to write custom GPU code, you don't need to do it in C anymore! For example you can JIT Python, using Numba or Triton.

Usually writing custom code is only required when:

- You are doing something novel, like PhD level stuff

- You must optimize the ML project for performance and trough-put at interference time

- You need to brute-force solutions (be it crypto-hashes, passwords, NP-complete problems, ...)

My last point to you is that do you want to learn to use these pre-existing frameworks and libraries, or learn to develop them or maybe even create new ones? What ever your answer is, I'd say that the first option is a great stepping-stone to advance to the second one.

New comment by nikonyrh in "Ask HN: Diverse Challenging Programming Projects"

nikonyrh — Mon, 10 Jun 2024 23:25:22 +0000

- Photo editor

- In-memory database

- Data compression algorithm

- 2D or 3D physics engine / game

- Game engines in general

New comment by nikonyrh in "A ChatGPT mistake cost us $10k"

nikonyrh — Sun, 09 Jun 2024 22:47:48 +0000

Aah that makes sense, thanks!

New comment by nikonyrh in "A ChatGPT mistake cost us $10k"

nikonyrh — Sun, 09 Jun 2024 21:33:47 +0000

I'm not familiar with this library, how does `text("(now())")` evaluate and why there are extra parentheses? And should that be a lambda expression as well, so that `create_date` isn't just the timestamp when the python process was started?

New comment by nikonyrh in "Ask HN: What does experience of data structures and algorithms mean?"

nikonyrh — Sun, 09 Jun 2024 21:12:33 +0000

I suppose all SW engineers have experience with data structures, when you are choosing whether to store the data into a dictionary, array or linked list. It is a bit more stretch to ask whether you've implemented a database before or not. On a more advanced level you'd dive into topics such as copy-on-write and software transactional memory (hobby-wise I come from Clojure background).

As an example, do you know how git works?

Edit: Oh an other great example of data structures is torrent files (and magnet links). Hint: https://en.wikipedia.org/wiki/Merkle_tree

New comment by nikonyrh in "Public URLs for Localhost via SSH"

nikonyrh — Fri, 31 May 2024 15:52:50 +0000

Wow, very cool and easy to set up!

New comment by nikonyrh in "Ask HN: What is the project that you are most excited about today?"

nikonyrh — Mon, 13 May 2024 22:16:30 +0000

Building a custom wall mount for my downhill bike. It requires some extra engineering and planning, the space is tight since there is already one bike on the hallway. It is also important to estimate the forces involved, so that there is sufficient safety margin.

New comment by nikonyrh in "Ask HN: Are You Training a Model?"

nikonyrh — Mon, 13 May 2024 11:50:49 +0000

That is what I usually do, I don't find very interesting to just load pre-existing models and feed data into them. But sometimes it is a necessity, if the network is too large (like Stable Diffusion, ResNet and its variants, ...).

New comment by nikonyrh in "You Don't Want to Use CSV Files"

nikonyrh — Sun, 05 May 2024 01:37:10 +0000

I remember not being able to read Pandas parquet files using for example Vaex (https://pypi.org/project/vaex/). And it is a bit disgenuine not to enlist all supported arguments of `pandas.read_parquet`'s `*kwargs`. But I agree that Parquet is already 100x better than CSV.

New comment by nikonyrh in "China's Temu Takes over 17% of US Market Share"

nikonyrh — Sun, 28 Apr 2024 22:20:46 +0000

I thought this was 17% of all online sales! Like all of the stuff that Amazon sells.