Hacker News: ydj

New comment by ydj in "Qwenputer: A functioning, deterministic RISC-V CPU inside Qwen3.5"

ydj — Thu, 29 Oct 2026 03:50:52 +0000

I dare say that’s the joke.

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

ydj — Sat, 03 Oct 2026 04:29:49 +0000

Agree, this post feels like it said very little. “Failed to use glm for a month because sometimes it’s slow and want to try other models too.” Okay, so I guess the only complaint about the model (or the service rather) is that it’s a bit slow.

New comment by ydj in "Deser: Rethinking Rust Serialization"

ydj — Wed, 30 Sep 2026 00:26:12 +0000

deser might also just be deserialize shortened, and I was wondering if it was something focused on deserialization only. As such, I didn’t find a quick little note on the etymology to be a particularly egregious waste of words.

New comment by ydj in "Quake Shareware, a CD-ROM just a little too full"

ydj — Mon, 17 Aug 2026 23:40:58 +0000

The computers in the photo and the layout of the store and signage are also quite strange.

New comment by ydj in "LG monitors silently install software through Windows Update without consent"

ydj — Sun, 19 Jul 2026 03:04:48 +0000

I used to think LG monitors were a good option for high quality 4k high refresh rate monitors. But having owned one of their expensive gamery offerings, it is basically worse in every way compared to an older acer. Worse DP1.4 stability (needs really nice cables to be stable), slow/broken ddc implementation, weird tearing with hdmi output from a Mac.

And I also had an older lg 4k monitor that bricked itself after a power outage, which is apparently a known eeprom issue.

Now this. No more LG for me.

New comment by ydj in "SearXNG: A free internet metasearch engine"

ydj — Sun, 05 Jul 2026 02:45:28 +0000

Yes, if it does archiving too that would be perfect.

New comment by ydj in "How We Made IPFS Content Publishing 10x Faster"

ydj — Sat, 04 Jul 2026 04:31:41 +0000

Not sure, there was a lot of complaints when it rolled out because it was actually peer to peer syncing to other laptops which was causing bandwidth issues for people working on vpn, from planes etc.

I’m not sure if it was actually a well thought out idea, since it’s not like fb has a lack of servers or bandwidth, but I didn’t work on it, so I don’t want to throw too much shade at it.

New comment by ydj in "SearXNG: A free internet metasearch engine"

ydj — Sat, 04 Jul 2026 03:32:45 +0000

Hister sounds like something I wanted for a while, but never got around to building. Searching stuff I’ve seen before is most of what I do with a search engine, so having it local and fast would be amazing. Eager to give it a try.

New comment by ydj in "How We Made IPFS Content Publishing 10x Faster"

ydj — Wed, 01 Jul 2026 19:22:49 +0000

At meta, there was a project for delivering binaries of internally built libraries / binaries to dev laptops using a private ipfs network. This was live for at least some period of time.

New comment by ydj in "A data race that doesn't compile"

ydj — Fri, 26 Jun 2026 05:09:00 +0000

Surely you can express this in C++ just as you could in rust…

New comment by ydj in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"

ydj — Mon, 15 Jun 2026 01:07:49 +0000

Yeah, I’m doing TP with two cards. The topology is configured based on yaml files, and if you are not using a predefined config you can just create a new config with your topology.

I’m not even using a 800G cable since they are expensive and I don’t think I need the bandwidth, opting for 400G instead. This just needs a config change for the number of Ethernet links it uses internally. (Apparently these cables are just many 200G links put together.)

New comment by ydj in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"

ydj — Sun, 14 Jun 2026 07:41:39 +0000

I think it’s not really worth it compared to just buying tokens or a coding plan.

My setup has 4090 handling attention while TT accelerators handles MLP. With just a 4090 you can have CPU handle the MLP layers and use a MoE model, assuming sufficiently powerful cpu. I tried that setup with minimax 2.5 before, and was able to eke out around 10 to 15 tps (albeit with a 7965wx cpu)

New comment by ydj in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"

ydj — Sun, 14 Jun 2026 07:15:38 +0000

The software stack is pretty immature, definitely very DIY. Their officially supported models are pretty old at this point, though there’s community support for gemma4, and models with GDN like qwen3.6 is supposedly very close.

The entire stack (minus some binary blobs in firmware) is open source, so if you have the time and persistence you can get whatever you want done.

A few community members have been working on support with llamacpp, where we can have supported operations offloaded to the TT cards, while having unsupported ops running on GPU or CPU. Llamacpp is pretty good at that. The existing kernels could definitely be better, and I’ll try my hand at writing some kernels some time.

New comment by ydj in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"

ydj — Sun, 14 Jun 2026 07:09:29 +0000

Yeah that’s definitely the smarter buy if you want to just have models running quickly. But the cost of 2 p150 and a 4090 was <$5000 for me.

The main issue is the immature software, and somewhat baroque way of writing kernels. Please, buy one and join us.

New comment by ydj in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"

ydj — Sat, 13 Jun 2026 16:26:26 +0000

80tp/s with 5080 3090 combo is wild. I’ve been working with a 4090 and two Tenstorrent p150 cards, and manage only about 30 tps utilizing all three for qwen3.6 27b q8. Guess I got more optimization to do.

Would like to see the perf of their setup with and without mtp and ngram speculative decoding though, as well as parallel decode performance (once llamacpp mtp plays well with multiple slots).

Being in California electricity alone puts this non-competitive with just paying a cloud though.

New comment by ydj in "Show HN: Number Gacha, a gacha game distilled to its essence"

ydj — Tue, 19 May 2026 03:01:19 +0000

I think some people do like the collection mechanics. And some games require specific team comps in order to beat the harder content.

New comment by ydj in "What political censorship looks like inside an LLM's weights (Qwen 3.5)"

ydj — Tue, 19 May 2026 02:57:54 +0000

How do you determine that the model was reasoning in Chinese in layer X? I would think the middle layers do not map into any tokens.

New comment by ydj in "The Future of Obsidian Plugins"

ydj — Tue, 12 May 2026 23:50:48 +0000

The thing I always wondered regarding obsidian plugins is how they are able to offer them on iOS, given that iOS has rules against downloading code that alters functionality of the software.

New comment by ydj in "Accelerating Gemma 4: faster inference with multi-token prediction drafters"

ydj — Wed, 06 May 2026 04:09:32 +0000

Not necessarily. Servers serving the model likely has enough traffic that they are batching decodes already. MTP reduces latency and increase efficiency only when the server can’t batch enough concurrent streams to be compute bound rather than memory bound.

New comment by ydj in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

ydj — Thu, 16 Apr 2026 19:52:36 +0000

Running the mxfp4 unsloth quant of qwen3.5-397b-a17b, I get 40 tps prefill, 20tps decode.

AMD threadripper pro 9965WX, 256gb ddr5 5600, rtx 4090.