<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: gwenzek</title><link>https://news.ycombinator.com/user?id=gwenzek</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 15 Jun 2026 13:13:58 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=gwenzek" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by gwenzek in "Zig 0.16.0 Release Notes"]]></title><description><![CDATA[
<p><a href="https://ziglang.org/download/0.16.0/release-notes.html#Loop-Vectorization-Disabled-to-Work-Around-Regression" rel="nofollow">https://ziglang.org/download/0.16.0/release-notes.html#Loop-...</a><p>Seems like a big one! I wonder how it will impact real code performance</p>
]]></description><pubDate>Wed, 15 Apr 2026 00:53:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=47773364</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=47773364</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47773364</guid></item><item><title><![CDATA[New comment by gwenzek in "As Android developer verification gets ready to go, a new reason to be worried"]]></title><description><![CDATA[
<p>I really don't like this move. I always rooted for Android because of its openness. It was also a reason why I was proud to work at Google.<p>The PlayStore is full of crapware, and I'm using a lot of apps from Fdroid, including basics like keyboard and text.<p>If you're at Google try to push back on this please.</p>
]]></description><pubDate>Sat, 20 Sep 2025 05:47:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=45310753</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=45310753</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45310753</guid></item><item><title><![CDATA[New comment by gwenzek in "Unexpected productivity boost of Rust"]]></title><description><![CDATA[
<p>Andrew actually agrees with the general sentiment.
And this will be made into a compile error.<p>What's happening is that compiler knows the two errors come from disjoint error set, but it promotes them both to anyerror<p>Details at <a href="https://github.com/ziglang/zig/issues/25046" rel="nofollow">https://github.com/ziglang/zig/issues/25046</a></p>
]]></description><pubDate>Thu, 28 Aug 2025 17:20:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=45054675</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=45054675</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45054675</guid></item><item><title><![CDATA[New comment by gwenzek in "Implementing a Struct of Arrays"]]></title><description><![CDATA[
<p>Interesting to see the new C++ at work.
But also I'm sure it's easier to learn Zig full language that just the new C++ metaprogramming</p>
]]></description><pubDate>Fri, 09 May 2025 21:46:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=43941206</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=43941206</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43941206</guid></item><item><title><![CDATA[New comment by gwenzek in "Zig 0.14.0 Release Notes"]]></title><description><![CDATA[
<p>Really liking the promotion of "Unmanaged" containers.
It's the ones I'm using everywhere already</p>
]]></description><pubDate>Thu, 06 Mar 2025 04:09:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=43276325</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=43276325</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43276325</guid></item><item><title><![CDATA[New comment by gwenzek in "Show HN: CLR, POC borrow checker for Zig"]]></title><description><![CDATA[
<p>Nice write up !
I think it would be nice for Zig to try to catch some of those errors like return of stack pointers, 
Cause it's painful for beginners</p>
]]></description><pubDate>Wed, 05 Feb 2025 12:15:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=42947425</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=42947425</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42947425</guid></item><item><title><![CDATA[New comment by gwenzek in "Why is C the safest language? [pdf]"]]></title><description><![CDATA[
<p>If you think this paper is pro-C++, read it again:<p>> many languages like C++ or Java are now old enough that if they would have
> offered significant advantage over C, they should have supplanted C by now,
> this clearly hasn’t happened.<p>I don't agree with all the points though and the rhetoric is pretty weak IMO, like the point about cars being safe for pedestrians.<p>But I agree with the points about C being simple and having great tooling (thanks to being simple).</p>
]]></description><pubDate>Thu, 09 Jan 2025 08:47:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=42643205</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=42643205</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42643205</guid></item><item><title><![CDATA[New comment by gwenzek in "ZML - High performance AI inference stack"]]></title><description><![CDATA[
<p>ZML the zig library is mostly a wrapper of StableHLO/pjrt.
But it's a high quality wrapper, and the tagged tensor syntax is really helpful to write complex ops like dot, or gather.<p>And ZML the framework also resolve issues with the complex dependency graph of stablehlo/pjrt.</p>
]]></description><pubDate>Thu, 19 Sep 2024 07:34:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=41589461</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=41589461</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41589461</guid></item><item><title><![CDATA[New comment by gwenzek in "ZML - High performance AI inference stack"]]></title><description><![CDATA[
<p>Zig has been relatively stable for the past few years for the main Zig code.
What has changed the most is the `build.zig` build system (which we aren't using).<p>We are also looking ahead at Zig roadmap, and trying to anticipate upcoming breaking changes,
and isolate our users from that.</p>
]]></description><pubDate>Wed, 18 Sep 2024 07:51:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=41576970</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=41576970</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41576970</guid></item><item><title><![CDATA[New comment by gwenzek in "ZML - High performance AI inference stack"]]></title><description><![CDATA[
<p>It' s a bit early to compare directly to TensorRT because we don't have a full-blown equivalent.<p>Note that our focus is being platform agnostic, easy to deploy/integrate, good performance all-around, and ease of tweaking.
We are using the same compiler than Jax, so our performances are on par.
But generally we believe we can gain on overall "tok/s/$" by having shorter startup time, choosing the most efficient hardware available, and easily implementing new tricks like multi-token prediction.</p>
]]></description><pubDate>Wed, 18 Sep 2024 07:39:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=41576897</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=41576897</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41576897</guid></item><item><title><![CDATA[New comment by gwenzek in "ZML - High performance AI inference stack"]]></title><description><![CDATA[
<p>Hi co-author here. Zig is way simpler than C++. Simple like in an afternoon I was able to onboard in the language and rewrote the core meat of a C++ algorithm and see speed gains (fastBPE for reference).<p>Coming from Python, the hardest part is learning memory management. What helps with ZML is that the model code is mostly meta programming, so we can be a bit flexible there.<p>We have a high level API, that should feel familiar to Pytorch user (as myself), but improves in a few ways</p>
]]></description><pubDate>Tue, 17 Sep 2024 15:48:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=41568948</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=41568948</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41568948</guid></item><item><title><![CDATA[New comment by gwenzek in "Zig and Emulators"]]></title><description><![CDATA[
<p>Very nice read.<p>I agree with the author that Zig should improve a bit the readability of integer and vector arithmetics.<p>Small tips is to use more lines.<p><pre><code>  fn addOffset(base: u16, off: u8) u16 {
    const ibase: i16 = @bitcast(base);
    const ioff: i8 = @bitcast(off);
    return @bitCast(ibase +% ioff);
}</code></pre></p>
]]></description><pubDate>Sun, 25 Aug 2024 13:14:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=41347063</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=41347063</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41347063</guid></item><item><title><![CDATA[New comment by gwenzek in "It's Time to Tax Billionaires"]]></title><description><![CDATA[
<p>Why do we tax income though? Especially for billionaire in a globalized world, it's easy to hide incomes. Wealth is harder to hide.</p>
]]></description><pubDate>Mon, 23 Oct 2023 20:33:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=37991184</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=37991184</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=37991184</guid></item><item><title><![CDATA[New comment by gwenzek in "Great Male Renunciation"]]></title><description><![CDATA[
<p>Graphic 5.2 in <a href="http://piketty.pse.ens.fr/files/Piketty2019GraphiquesTableaux.pdf" rel="nofollow noreferrer">http://piketty.pse.ens.fr/files/Piketty2019GraphiquesTableau...</a><p>In France it was down to less than 1% at the time of Revolution, but some countries like Spain that had  centuries long wars, was closer to 6-8% range.</p>
]]></description><pubDate>Sun, 22 Oct 2023 22:25:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=37979637</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=37979637</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=37979637</guid></item><item><title><![CDATA[New comment by gwenzek in "Problems of C, and how Zig addresses them"]]></title><description><![CDATA[
<p>I wrote a similar article but focused on how Zig design enable better optimizations than C. <a href="https://zig.news/gwenzek/zig-great-design-for-great-optimizations-638" rel="nofollow noreferrer">https://zig.news/gwenzek/zig-great-design-for-great-optimiza...</a></p>
]]></description><pubDate>Mon, 03 Jul 2023 16:15:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=36575068</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=36575068</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36575068</guid></item><item><title><![CDATA[Zig: Great Design for Great Optimizations]]></title><description><![CDATA[
<p>Article URL: <a href="https://zig.news/gwenzek/zig-great-design-for-great-optimizations-638">https://zig.news/gwenzek/zig-great-design-for-great-optimizations-638</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=36573929">https://news.ycombinator.com/item?id=36573929</a></p>
<p>Points: 11</p>
<p># Comments: 3</p>
]]></description><pubDate>Mon, 03 Jul 2023 15:04:35 +0000</pubDate><link>https://zig.news/gwenzek/zig-great-design-for-great-optimizations-638</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=36573929</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36573929</guid></item><item><title><![CDATA[New comment by gwenzek in "The Precipitation Problem"]]></title><description><![CDATA[
<p>Hard question. Probably because we were in a sweet spot without realizing it. And that lands are also pretty damaged. Swamp are a key part of regulating water flow, but humanity have been destroying those for hundreds of years. Also grounds are probably over-plowed, which make them less permeable. See "dust bowl".</p>
]]></description><pubDate>Tue, 27 Jun 2023 16:41:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=36495303</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=36495303</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36495303</guid></item><item><title><![CDATA[New comment by gwenzek in "Japanese biomass venture using microorganisms to tackle waste disposal"]]></title><description><![CDATA[
<p>I know farmers that are using city compost to replace chemical fertilizers for their vegetables. There is probably an over supply of compost in some places, but it feels this will probably reverse over time as farmers adapt their techniques.</p>
]]></description><pubDate>Sun, 18 Jun 2023 07:00:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=36377982</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=36377982</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36377982</guid></item><item><title><![CDATA[New comment by gwenzek in "Using mmap to make LLaMA load faster"]]></title><description><![CDATA[
<p>It's indeed a very different world. This model was trained on thousands of GPUs. The weird file format corresponds to the train time sharding of the weights. And really nobody is doing CPU inference with all the GPU we have. And also the "CLI" use case seems contrieved to me. If you plan to interact several times with the model and want to keep the weights in RAM, why don't you start a REPL or spin up a server?</p>
]]></description><pubDate>Fri, 07 Apr 2023 06:51:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=35478882</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=35478882</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=35478882</guid></item><item><title><![CDATA[On the value of going deep, or How a broken keyboard led me to fix bugs in Zig]]></title><description><![CDATA[
<p>Article URL: <a href="https://zig.news/gwenzek/on-the-value-of-going-deep-or-how-a-broken-keyboard-led-me-to-fix-bugs-in-zig-288o">https://zig.news/gwenzek/on-the-value-of-going-deep-or-how-a-broken-keyboard-led-me-to-fix-bugs-in-zig-288o</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=34021421">https://news.ycombinator.com/item?id=34021421</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 16 Dec 2022 22:25:36 +0000</pubDate><link>https://zig.news/gwenzek/on-the-value-of-going-deep-or-how-a-broken-keyboard-led-me-to-fix-bugs-in-zig-288o</link><dc:creator>gwenzek</dc:creator><comments>https://news.ycombinator.com/item?id=34021421</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=34021421</guid></item></channel></rss>