Hacker News: gunalx

New comment by gunalx in "Kolibri: A Sovereign Open-Weight Model"

gunalx — Sat, 03 Oct 2026 23:02:23 +0000

LLMs can be configured to basically be deterministic. It's only really bad for performance because language is not deterministic.

New comment by gunalx in "OKI Develops 124-Layer PCB Technology"

gunalx — Sat, 03 Oct 2026 10:08:52 +0000

Wouldn't this just be a mess signal integrity wise?

New comment by gunalx in "Why should I have to pay more for buses if I don't use a smartphone?"

gunalx — Sat, 03 Oct 2026 09:52:53 +0000

Cheaper per ride over the month usually.

New comment by gunalx in "One month coding with GLM 5.3 Flash"

gunalx — Fri, 02 Oct 2026 22:48:42 +0000

If the market would be fully efficient how would it be possible to make money in the market? I font think the market is that fast at rebalancing.

New comment by gunalx in "One month coding with GLM 5.3 Flash"

gunalx — Fri, 02 Oct 2026 19:22:22 +0000

Will agree on this. From z.ai i have found the non flash to have way more consistent performance.

New comment by gunalx in "One month coding with GLM 5.3 Flash"

gunalx — Fri, 02 Oct 2026 19:21:06 +0000

When text wuality or for pure but adwansed coding is concerned i always pick glm 5.3. The flash is awesome for everything that dosent really matter though.

New comment by gunalx in "Show HN: Janus – Go binary that runs GGUF models via Vulkan on AMD/Intel/Nvidia"

gunalx — Fri, 02 Oct 2026 07:00:45 +0000

The docs are a great start.

https://github.com/ggml-org/llama.cpp/blob/master/docs/backe...

New comment by gunalx in "Show HN: Janus – Go binary that runs GGUF models via Vulkan on AMD/Intel/Nvidia"

gunalx — Fri, 02 Oct 2026 06:56:12 +0000

Yes and no. I tested llamacpp on my intel gpu with both vulkan and sycl. I measured them to be in the same ballpark even if i have se en pepole claim marger differences than i observed.

In the end i took the sligth slowdown of vulkan to have a more stable and higher development velocity backend While being able to use identical setups on both intel and and gpus.

New comment by gunalx in "Meta's Muse is fantastic for web scraping"

gunalx — Fri, 02 Oct 2026 06:11:40 +0000

Im guessing more of the internet will be login walled from now on.

New comment by gunalx in "GrapheneOS – When an app is slow"

gunalx — Mon, 28 Sep 2026 23:01:23 +0000

Have had no noticable problems with organic maps.

New comment by gunalx in "Ask HN: What would be the best local ASR model"

gunalx — Sat, 26 Sep 2026 21:52:28 +0000

Whisper is still pretty good over all. At least if you need multilingual support. Other mentions is parakeet for speed and maybe qwen3 asr.

New comment by gunalx in "Six-year-old breaks women's world Rubik's Cube record [video]"

gunalx — Thu, 24 Sep 2026 09:10:26 +0000

There is a min cap on how few moves the cube shall be possible to complete in.

New comment by gunalx in "Tokens Too Cheap to Meter"

gunalx — Wed, 23 Sep 2026 14:27:44 +0000

Grep is trivially parallelizable if you care to do it though.

https://iepathos.github.io/ripgrep/performance/#work-stealin...

New comment by gunalx in "MiMo v2.6"

gunalx — Mon, 21 Sep 2026 23:23:52 +0000

If I pay you 100$ for mowing my lawn, and you me for yours. Technically the GDP increased with 200$.

New comment by gunalx in "Raspberry Pi blocks changing RAM chips"

gunalx — Mon, 21 Sep 2026 18:08:00 +0000

Its a closed broadcom chip. Basically futile or heavily resource intensive to try.

New comment by gunalx in "Amazon blocks Meta’s new Muse AI agent from shopping on amazon.com"

gunalx — Mon, 21 Sep 2026 17:51:53 +0000

Theh will probably consolidate more. Less choices even if they are worse is also easier for llms to do. Just look at how they pick technologies to use. Almost always the same for a given model.

New comment by gunalx in "Qwen Image 2.1"

gunalx — Sun, 20 Sep 2026 14:32:50 +0000

Its happy to see a new open image model from qwen. But the license is a let down. And it dosent even beat their closed qwen3 image wich is already a bit old.

New comment by gunalx in "Gemini 3.8 Live and 3.8 Live Extended Thinking"

gunalx — Tue, 15 Sep 2026 20:10:29 +0000

Not at all in mine. Deepseek has some of the worst prose of the close to frontier models in my opinion.

New comment by gunalx in "Steam Frame starts at $1059"

gunalx — Mon, 14 Sep 2026 21:10:48 +0000

Wait I don't realise ram prices where this bad. Bought my kit for about a tenth of that a while ago. Ouch.

New comment by gunalx in "Steam Frame starts at $1059"

gunalx — Mon, 14 Sep 2026 20:35:00 +0000

If it can do innside out tracking its way more data efficent to have compute onboard. And if you already have a decent amount of compute you can scale it up just a little more to have. a fully standalone unit.