boulos

0 x inf is also NaN. It's not just division.

But it's not clear that using Sonnet or any other LLM as a "grader" would result in the same improvement. For objective grading, you could be sure that the additional adaptive support is helping. For subjective things like writing style, literature, poetry, you end up with whatever Sonnet thinks is good (and randomly so).

It still could be better for students, but it's not obvious that it would be (or maybe not as strongly?).

Do you have a larger study planned for the Fall? It definitely seems promising.

I'm curious how well you feel this worked because the subject was Statistics (objective grading) versus something more subjective like Civics or Literature.

PS - I'd say this qualifies for Show HN, too!

Do you

And it made a classic optimization: removing New Zealand :).

This is nice work, but I found the bug finding example to be weird:

One such bug was in the sign function for zigzag decoding of the datrs/varinteger library. On input Std.U64.MAX, the expression (value + 1) overflowed, causing crashes in debug mode and silent corruption in release mode—an edge case that testing and fuzzing would typically miss.

In what way would this boundary condition case be considered something that "testing [...] would typically miss"? It's certainly something that bad tests would miss or not think about, but I find that (a) careful people and (b) ML coding systems are actually really good at "oh, I should test the extreme values". Especially for things that parse user input.

I'm curious if they found other bugs that were more interesting, but found them too hard to explain quickly.

I appreciate the focus on amusement in:

Some of the newly added features may come as a surprise to those of you who keep a close track of ZLUDA development. Most of them were previously explicitly outside ZLUDA's roadmap. There has been a change of plans. ZLUDA development is no longer commercially funded, so it's back to being my weekend project. This means that the priority is no longer what makes commercial sense, but what I find the most entertaining. That's why the sudden addition of textures, PhysX and better Windows support.

Outside of the time commitment that full-time support would offer, I generally think focusing on amusement is a good life strategy.

Yep. Google does this in Finland for the Hamina datacenter: https://blog.google/company-news/inside-google/around-the-gl...

529 and not 429?

Technically, we did RTSL before Ingo and then Matt did their stuff at Intel. LLVM wasn't particularly usable in 2007, but Austin and I did try it (this isn't in the paper!). A few years later, I did an actual LLVM JIT for OSL, but we didn't have enough coherence for vectorized shading at the time at Sony.

But none of it was surprising. The original RenderMan shading language was "vectorized" and it even used SIMD instructions on modern processors to run the "interpreter loops". That is, a single "color add" in RSL might have looked like:

  for (int i=0; i < grid_points; i++) {
    out_color[i] += foo;
  }

The point of ISPC et al. was to give people CUDA-like easy mode for "trust me, just vectorize the whole thing and deal with the masking for me". It goes beyond a hardcoded shading language (easier target!) though didn't reach as far as CUDA with complex structs / C++ capabilities.

Because it's complex?

Perhaps you missed work like https://crfm.stanford.edu/2025/05/28/fast-kernels.html ?

That's not the "edge" in the title. It means competitive advantage in this case.

How do you like the brikka vs the classic?

How's your experience with Fil-C been? Is it materially useful to you in practice?

This used to be true, but GCP forced their hand in January of 2024 with https://cloud.google.com/blog/products/networking/eliminatin...

AWS matched a few months later:

https://aws.amazon.com/blogs/aws/free-data-transfer-out-to-i...

I'm not trying to convince you to stay (I work for neither anymore!), just wanted to note that you can technically request a waiver. I'm not sure how this works in practice though. Like, if you want to leave Athena and move to something on-premise is that enough to have just that workload? Maybe!

Edit: I also didn't follow this at the time, but the AWS wording suggests that the "EU Data Act" is also involved.

Yeah, but that missing context is super important.

If they want it for local dev work, that's pretty different from wanting a high-performance air gapped object store without rewriting clients.

They seem to know what they're doing (having complained about a methodology problem in MinIO), and yet don't personally want to throw their hat in the ring not maybe pay anyone...

Context matters!

Neat. I wish I could select just a couple lines at once though. I feel like the 1/2/3 plus one of the other lines would make something more appealing.

Compared to roll-your-own with S3 or GCS it does :)

Will do. I wanted mostly to (re)inform the submitter.

Please don't editorialize titles unless they're clearly clickbait.

"Designing AI for Disruptive Science" is a bit market-ey, but "AI Risks 'Hypernormal' Science" is just a trimmed section heading "Current AI Training Risks Hypernormal Science".

Our sixth generation sensor suite is intended for winter driving: https://waymo.com/blog/2025/10/creating-an-all-weather-drive... for some nice visuals.

On recent podcasts, Ilya says he's no longer assuming they can jump straight there.

Maybe the author isn't aware that Chainguard is going to keep patching MinIO for CVEs:

https://www.chainguard.dev/unchained/secure-and-free-minio-c...

You wouldn't get the other changes in this post (e.g., restoring the admin console) but that's a bit orthogonal.

The entire global software industry is worth less than $1 trillion dollars

Are you saying "worth" as a shorthand for something like annual profit? If you sort the 2025 data by earnings, you get pretty large numbers quickly: https://en.wikipedia.org/wiki/List_of_largest_technology_com...

That's not how you should measure "worth". In that world, you'd have a P/E ratio of 1. Comparing to a bond, it would be like expecting to get paid the face amount in a single year. Many people are quite happy with 5-10% interest as a risky benchmark, so 10-20 P/E isn't wild. That puts the market cap for tech itself at 10-20T as a reasonable baseline.

I feel vindicated :). We put in a lot of effort with great customers to get nested virtualization running well on GCE years ago, and I'm glad to hear AWS is coming around.

You can tell people to just do something else, there's probably a separate natural solution, etc. but sometimes you're willing to sacrifice some peak performance just have that uniformity of operations and control.

This isn't strictly correct: you probably mean wrt compressed size. Compression is a tradeoff between size reduction and compression and decompression speed. So while things like Bellard's tz_zip (https://bellard.org/ts_zip/) or nncp compress really well they are extremely slow compared to say zstd or the much faster compression scheme in the article. It's a totally different class of codec.

IIRC, Google contributed +-400 racks last year to Open Compute: https://cloud.google.com/blog/topics/systems/enabling-1-mw-i...

I haven't benchmarked them myself yet, but the C++23 flat map containers are supposed to finally have fixed this. Chrome lists them as TBD: https://chromium.googlesource.com/chromium/src/+/main/styleg... .

Wait, really? I haven't been up this season, but it's always been there! I understand removing the printed ones when the bars have them (and the big boards at the top). Is it all just ads now?

Every lift at Snowbird has the map printed on the bar. So you can plan your route on the way up. I agree that when you get lost, that map won't save you, but I think an offline PDF is also fine.