tucnak

23 ms·

301.
by tucnak 1y ago
I must agree that for numerical computation (and downstream optimisation thereof) Julia is much better suited than ostensibly "systems" language such as Rust. Moreover, the compatibility matrix[1] for Rust-CUDA tells a story: ther
302.
by tucnak 1y ago
> of all places, I did not expect to get personal attacks on HN You must be new to the site.
303.
by tucnak 1y ago
I'm sorry, but to blame Intel's inadequacies on political climate is comedic.
304.
by tucnak 1y ago
IPSec, no IPv6, sad.
305.
by tucnak 1y ago
How is Gerrit doing these days; is it any good for breaking away from GitHub? I'm looking for a git/mailing-list solution a small business could use on-prem but not any of that Gitlab nonsense (to throw people off-balance, if you
306.
by tucnak 1y ago
I really hope it means that 2.5 models will be available for batching in Vertex, too. We had spent quite a bit of effort making it work with BigQuery, and it's really cool when it works. There's edge-case, though, where it doesn&#
307.
by tucnak 1y ago
Is it just me or does this read a little boring and a lot longer than it should..? So much of the email is just him calling random people... whole paragraphs of it
308.
by tucnak 1y ago
Not to step on your toes, but it shall be said that instructions in a CPU "retire" in order.
309.
by tucnak 1y ago
Horrorshow. Longterm Go user and open source maintainer here.. looking at this code makes me want to puke. The whole thing is a crime against semantics. I thought the whole point was to do generics when they could do them well, right? This
310.
by tucnak 1y ago
My issue with it at the time had to do with... how to put it... simply build the extension, & carry on with my life. Version after version it would build alright, and then suddenly, boom, it doesn't work no more. Okay, I understand
311.
by tucnak 1y ago
Yeah, they would often claim that C API is at fault, but believe it or not, blaming Rust just as much. I've had an interesting discussion with pgrx people: https://github.com/pgcentralfoundation/pgrx/issues&#x
312.
by tucnak 1y ago
Rust is clearly aiming to match C++ in complexity. I don't know whether this is “Death by committee” in a new wrapper, only this time it's the "community," driven by Github emojis, as opposed to legacy processes you woul
313.
by tucnak 1y ago
https://www.google.com/search?q=is+all+of+oregon+north+of+ne... Show more -> Disclaimer and the feedback buttons are shown at the end. If you bothered enough to read the full response, you would have seen the disclaimer,
314.
by tucnak 1y ago
Your screenshot is conveniently omitting the disclaimer below: "AI responses may include mistakes. Learn more[1]" [1]: https://support.google.com/websearch/answer/14901683
315.
by tucnak 1y ago
> It does not invent plausible-looking information. This is where your analogy is falling apart; of course web browsers do not "invent plausible-looking information" because they don't invent anything in the first place! W
316.
by tucnak 1y ago
This is often overlooked. No, I don't want random cargo package to run RCE's on my machine. And there's just so many; in terms of bloat, the dependency trees are huge, rivaled only by Nodejs, if I'm being honest. I have
317.
by tucnak 1y ago
A bit off-topic, but every time I read some sophisticated Rust code involving macros, I cannot help but think that something went wrong at some point. The sheer complexity far outpaces that of C++, and even though I'm sure they would c
318.
by tucnak 1y ago
> reticent That is a $200 word right there, and you're using it wrong.
319.
by tucnak 1y ago
This is perhaps more important in the age of AI agents, but before we can tackle all these fancy ZKP constructs in the mainstream — we have to, as the industry (and so far consistently failed to) — implement Zanzibar, or whatever ReBAC, and
320.
by tucnak 1y ago
To say this has nothing to do with blockchain is like saying RADAR had nothing to do with war. Yes, people knew Maxwell's equations prior, i.e. "knew the proofs," w.r.t. ZKP but it has only really been developed much later, d
321.
by tucnak 1y ago
Not necessarily! The setup I'm discussing is explicitly non-GPU, and it's not necessarily a TPU either. Any accelerator card with NoC capability will do: the requests are queued/batched from network, trickle through the adjac
322.
by tucnak 1y ago
I think, TrueTime would constitute a "trick," insofar ordering is concerned? > Consider that talking to L1 is at least 1,000x faster than talking to the GPU. This is largely true for "traditional" architectures, but s
323.
by tucnak 1y ago
"Blending," or translating arbitrary substrings to prefixes, is a real curious one, & likely become a prerequisite for running dataset-scale LLM inferences at scale. See https://arxiv.org/abs/2405.16444v3
324.
by tucnak 1y ago
Reminds me of PG-Strom[1] which is a Postgres extension for GPU-bound index access methods (most notably BRIN, select GIS functions) and the like; it relies on proprietary NVIDIA GPUDirect tech for peer-to-peer PCIe transactions between the
325.
by tucnak 1y ago
> Sorry, you have been blocked You are unable to access beehiiv.com Well
326.
by tucnak 1y ago
> FFT plus some basic analysis Yeah, totally easier than `len(transcribe(a))/len(a)`
327.
by tucnak 1y ago
https://docs.vllm.ai/projects/production-stack/en/latest/tut...
328.
by tucnak 1y ago
Vertex is a GCP service, and it's completely in line with how any other GCP service account would work; there's multiple auth flows, some taking credentials from the environment, and others from a file. From zero-trust POV, you co
329.
by tucnak 1y ago
Your obsession with finance/marketing is exactly what I expect to see on HN. It's a shame your accusations have zero merit. In the future, please try not to embarrass yourself by attempting to get into technical discussion, and pr
330.
by tucnak 1y ago
I'd never claimed that "TPU is going to conquer everything," it's a matter of fact that the latest-generation TPU is currently the most cost-effective solution for large-scale training. I'm not even saying that NVID

More ›