Hacker News: bonzini

New comment by bonzini in "AI assistance when contributing to the Linux kernel"

bonzini — Sat, 11 Apr 2026 18:06:28 +0000

The wording might be more or less lawyerly but the idea is fairly common, e.g. https://openinfra.org/legal/ai-policy (OpenStack).

New comment by bonzini in "AI assistance when contributing to the Linux kernel"

bonzini — Sat, 11 Apr 2026 18:05:08 +0000

Signed-off-by is a chain. The second person asserts that they delegate to the first person for the parts contributed by the first, and signs off on the ones that were contributed personally.

Hypothetically in court you'd go to the last, ask "did you write this" and only if not go up.

New comment by bonzini in "AI assistance when contributing to the Linux kernel"

bonzini — Sat, 11 Apr 2026 18:02:58 +0000

You can write in AGENTS.md to ask the user for explicit sign off and to explain the document to the user.

New comment by bonzini in "Surelock: Deadlock-Free Mutexes for Rust"

bonzini — Sat, 11 Apr 2026 17:48:10 +0000

Usually a global lock is a lock that is taken outside all others and is taken for large parts of the runtime (or even, everywhere the thread isn't waiting on a condition variable, file descriptor and the like).

Mutex::new(AppConfig::default()) might very well be a small, leaf mutex.

New comment by bonzini in "Moving from WordPress to Jekyll (and static site generators in general)"

bonzini — Fri, 10 Apr 2026 04:16:41 +0000

I use a SSG for a conference. The git repository guides me in doing exactly the same changes every year. I have pre built blocks for the various stages of organization and go through the commits for the previous year to help me understand what to do next.

New comment by bonzini in "Show HN: M. C. Escher spiral in WebGL inspired by 3Blue1Brown"

bonzini — Mon, 06 Apr 2026 04:59:00 +0000

You would have to know the position of the smaller copy in the uploaded image for the effect to work.

New comment by bonzini in "Show HN: M. C. Escher spiral in WebGL inspired by 3Blue1Brown"

bonzini — Sun, 05 Apr 2026 11:47:00 +0000

It's cited by 3b1b themselves, who used Leiden's un-spiralized image to describe the effect.

New comment by bonzini in "IBM Announces Strategic Collaboration with Arm"

bonzini — Thu, 02 Apr 2026 12:26:10 +0000

This certainly has been in the making for longer than the "everything we do must be for AI" bubble. In fact s390 has its own on-die inference engines and they have access to the same caching mechanisms as the main processor (which are quite insane).

IBM Announces Strategic Collaboration with Arm

bonzini — Thu, 02 Apr 2026 08:48:44 +0000

Article URL: https://newsroom.ibm.com/2026-04-02-ibm-announces-strategic-collaboration-with-arm-to-shape-the-future-of-enterprise-computing

Comments URL: https://news.ycombinator.com/item?id=47611721

Points: 282

# Comments: 191

New comment by bonzini in "Scientists crack a 20-year nuclear mystery behind the creation of gold"

bonzini — Wed, 01 Apr 2026 21:40:02 +0000

It's about refining theoreticals models that are used to predict nucleosynthesis of heavier elements. The researchers used indium because we can obtain the required neutron-heavy isotopes for indium but not for heavier elements such as gold or platinum. But improving the model with data from indium, they say, makes it more accurate for gold as well?

Why then gold in the title? Probably just because it's shiny.

New comment by bonzini in "Monado became the foundation for OpenXR runtimes"

bonzini — Sun, 29 Mar 2026 13:54:42 +0000

> The OpenGL model is having a specification that everyone implements, not a single foundational implementation??

They refer to Mesa. Otherwise unreadable I agree...

New comment by bonzini in "Looking at Unity made me understand the point of C++ coroutines"

bonzini — Wed, 25 Mar 2026 23:36:46 +0000

Stackful coroutines also can't be used to "send" a coroutine to a worker thread, because the compiler might save the address of a thread local variable across the thread switch (happened in QEMU).

New comment by bonzini in "Slovenian officials blame Israeli firm Black Cube for trying to manipulate vote"

bonzini — Wed, 25 Mar 2026 22:13:37 +0000

> If none of it mattered, why bother?

Hope for the best, plan for the worst. Republicans and missiles being the best.

New comment by bonzini in "Data centers are transitioning from AC to DC"

bonzini — Wed, 25 Mar 2026 20:13:35 +0000

> all the inputs and outputs share the same ground, it's not just the values for that pair of wires?

No, it depends on the converter. There are converters that leave 160V on the DC power rail for a 110V AC input, and 155V on the DC "ground" rail.

They are economic and you could find then when galvanic isolation is at least in theory not important, but they're terribly unsafe when used on PCBs that people might muck with.

If you have some "normal" converters and some of this kind, sharing the ground would be quite dangerous.

New comment by bonzini in "An ode to bzip"

bonzini — Sun, 15 Mar 2026 07:24:37 +0000

> meant to have good ratios and good (de)compression speeds as compared to other tools

That does not mean it's Pareto optimal; Pareto-optimality forms a curve and while zstd, LZMA, LZ4, ZPAQ all want to be as close as possible to the curve, they focus on different parts of it. In particular zstd tries to stay on the middle part of the curve, while LZMA and LZ4 focus on opposite sides

          ___.--- higher throughput
        /     LZ4
       / zStd
      |
      ; LZMA
     |
     | ZPAQ
    lower size

Also, the Pareto curve is not necessarily known in advance. All you can do is add more and more algorithms or tweaks to understand what it looks like. For example, this blog post [https://insanity.industries/post/pareto-optimal-compression/] shows that prior to zstd, bzip2 and gzip2 were both pretty much Pareto optimal in the same area for ratio vs. compression speed. LZMA at low settings was a bit better but much slower. There was a huge gap between LZMA and LZ4, and bzip2/gzip filled it as best as they could.

The same blog post shows that zstd is an absolute speed demon at decompression; while not all zstd settings are Pareto optimal when looking at size vs compression speed (in particular LZMA wins at higher compression ratios, and even considering zstd only there's hardly a reason to use levels 11-15), zstd is pretty much Pareto optimal at all settings when looking at size vs. decompression speed. On the other hand at intermediate settings zstd is faster and produces smaller files than gzip, which therefore is not Pareto optimal (anymore).

New comment by bonzini in "RISC-V Is Sloooow"

bonzini — Wed, 11 Mar 2026 07:28:59 +0000

I think they have already added hardware page table walks.

https://lwn.net/Articles/932048/

New comment by bonzini in "RISC-V Is Sloooow"

bonzini — Tue, 10 Mar 2026 21:47:41 +0000

LoongArch is, on a first approximation, an almost RISC-V user space instruction set together with MIPS-like privileged instructions and registers.

New comment by bonzini in "JSLinux Now Supports x86_64"

bonzini — Tue, 10 Mar 2026 10:29:14 +0000

Wow, with AVX512 too?? Now I really want to add it to QEMU. :)

(For APX I have patches at https://lore.kernel.org/qemu-devel/20260301144218.458140-1-p... but I have never tested them on system emulation).

New comment by bonzini in "Mondrian Entered the Public Domain. The Estate Disagrees"

bonzini — Tue, 03 Mar 2026 06:10:27 +0000

The article is about the Mondrian estate picking and choosing between the 95 year, life+70, and life+80 rules—and being wrong in all three cases.

So I wasn't doing a legal analysis of which rule is the correct one, but rather observing how their arguments are ridiculous.

New comment by bonzini in "Mondrian Entered the Public Domain. The Estate Disagrees"

bonzini — Mon, 02 Mar 2026 11:35:22 +0000

I read "The duration of the U.S. protection for all other works… was for 70 years from the artist’s date of death" and thought wow, did Mondrian really live into the 1960s or so?

Next paragraph: "Mondrian died in 1944. Any of his works subject to a life-plus-70 regime would have entered the public domain" 10 years ago. Who even thought of including that in a legal argument??