<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: kllrnohj</title><link>https://news.ycombinator.com/user?id=kllrnohj</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 08 Aug 2026 09:20:20 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=kllrnohj" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by kllrnohj in "2027 memory capacity is reportedly sold out"]]></title><description><![CDATA[
<p>Building your own is cheaper, although not by a ton. Valve isn't taking a loss on them.</p>
]]></description><pubDate>Sat, 08 Aug 2026 03:14:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=49218600</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49218600</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49218600</guid></item><item><title><![CDATA[New comment by kllrnohj in "AMD acquires Taalas to boost inference performance by etching models in silicon"]]></title><description><![CDATA[
<p>Well they still just do the same thing GPUs do where bad clusters are disabled. There's going to be a balance there for how big the clusters are and how many tiers they'll offer, but that's a known problem which can be well modeled.</p>
]]></description><pubDate>Fri, 07 Aug 2026 12:52:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=49209668</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49209668</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49209668</guid></item><item><title><![CDATA[New comment by kllrnohj in "AMD acquires Taalas to boost inference performance by etching models in silicon"]]></title><description><![CDATA[
<p>Larger wafers has been attempted for over a decade now and doesn't seem to be going anywhere. See the 450mm wafer section of <a href="https://en.wikipedia.org/wiki/Wafer_(electronics)" rel="nofollow">https://en.wikipedia.org/wiki/Wafer_(electronics)</a></p>
]]></description><pubDate>Fri, 07 Aug 2026 12:36:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=49209468</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49209468</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49209468</guid></item><item><title><![CDATA[New comment by kllrnohj in "AMD acquires Taalas to boost inference performance by etching models in silicon"]]></title><description><![CDATA[
<p>Cerebras is literally the entire wafer, so it can't get bigger. So where is the jump from 30x to 100x coming from? Node improvements only yield like 10-20% gains these days...</p>
]]></description><pubDate>Fri, 07 Aug 2026 03:59:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=49205760</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49205760</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49205760</guid></item><item><title><![CDATA[New comment by kllrnohj in "Getting 25 Gbps Thunderbolt Ethernet on My Mac Studio"]]></title><description><![CDATA[
<p>> Author talks about 4k video, but AI tells me that for "4K production mastering (ProRes 422 HQ, 4:2:2, 10-bit)" the rate is under 2Gbps.<p>That'd be the rate for 1 stream at 1x playback. Multiple streams because of overlays/crossfades/etc... along with playback at >1x (such as during rendering in particular) will change that significantly.</p>
]]></description><pubDate>Fri, 31 Jul 2026 18:21:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=49126809</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49126809</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49126809</guid></item><item><title><![CDATA[New comment by kllrnohj in "Keychron announces first open-source firmware for gaming mice"]]></title><description><![CDATA[
<p>it's hardly difficult to do that already. Like a Raspberry Pi Pico can be turned into a HID mouse rather easily: <a href="https://www.instructables.com/Raspberry-Pi-Pico-As-HID-Mouse/" rel="nofollow">https://www.instructables.com/Raspberry-Pi-Pico-As-HID-Mouse...</a><p>and yes, hdmi/displayport capture -> ML object detection -> fake HID mouse to cheat is absolutely already a thing.</p>
]]></description><pubDate>Wed, 29 Jul 2026 18:31:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=49101233</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49101233</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49101233</guid></item><item><title><![CDATA[New comment by kllrnohj in "Android may soon restrict on-device ADB"]]></title><description><![CDATA[
<p>So the risk is the plug for that being yanked in the ~2-5 days between ordering it and unlocking it? So... not a big deal at all. Not even slightly lol.</p>
]]></description><pubDate>Tue, 28 Jul 2026 15:37:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=49085519</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49085519</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49085519</guid></item><item><title><![CDATA[New comment by kllrnohj in "Android may soon restrict on-device ADB"]]></title><description><![CDATA[
<p>> Not before connecting it to the internet, even Pixels will go through hundreds of megabytes of data before allowing you to unlock them (see 0)<p>oookay? Hardly seems like a big deal? It would indeed be nice if the unlocked phones were set from the factory as such instead of all being the same system image as the ones that locked carriers use, sure, but hardly significant since it's not like you have to sign in or anything?<p>> Also, will Google let me browse the web with "my own OS" without their proprietary services installed on it? (see 1)<p>Why are you even asking Google at all? Just use Firefox? Don't even need to root or use a custom ROM for that, even.</p>
]]></description><pubDate>Sat, 25 Jul 2026 19:34:44 +0000</pubDate><link>https://news.ycombinator.com/item?id=49050759</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49050759</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49050759</guid></item><item><title><![CDATA[New comment by kllrnohj in "Android may soon restrict on-device ADB"]]></title><description><![CDATA[
<p>> As if you didn't need any more proof you don't own "your" devices.<p>Can you install your own OS? If yes, you own it. And Google consistently lets you do that. It other OEMs don't then direct your outrage at them.</p>
]]></description><pubDate>Sat, 25 Jul 2026 13:21:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49047381</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49047381</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49047381</guid></item><item><title><![CDATA[New comment by kllrnohj in "AMD Ryzen 7 7700X3D Review: 3D V-Cache Gaming Performance for Less"]]></title><description><![CDATA[
<p>It's primarily only the AAA <i>multiplayer</i> games that have that, not just AAA games <i>in general</i>.<p>Just browse the Deck Verified page for example: <a href="https://www.steamdeck.com/en/verified" rel="nofollow">https://www.steamdeck.com/en/verified</a> & <a href="https://store.steampowered.com/greatondeck/" rel="nofollow">https://store.steampowered.com/greatondeck/</a><p>Forza Horizon 6 and Assassin's Creed Black Flag Resync'd are both listed, and both are very recent, major AAA game launches. With top-tier Linux compatibility.<p>It is staggering how different the situation is now as compared to just a couple years ago. Steam Deck (and associated investments by Valve) are damn near a revolution. It's a perfect storm of the incumbent turning to shit at the same time the underdog really got its act together.<p>If you or your friend group is really into a game that doesn't have Linux support yeah that's a bummer, especially when it's just because of the kernel level anticheat that's horrifying in its own right (triply so when it's for a <i>PVE</i> game, cough cough, helldivers 2, cough cough). But I don't think that's <i>most people</i> anymore. The more casual and indie game scene these days is huge, and of course single player games in general continue to be super popular.</p>
]]></description><pubDate>Wed, 22 Jul 2026 14:56:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=49007810</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=49007810</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49007810</guid></item><item><title><![CDATA[New comment by kllrnohj in "Claude Code uses Bun written in Rust now"]]></title><description><![CDATA[
<p>Then I don't understand your argument. If you're just saying what could go wrong with heap corruption, then your vtable complaint also applies to storing function pointers in arena allocators in Zig? Zig doesn't have anything special here?</p>
]]></description><pubDate>Mon, 20 Jul 2026 11:30:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=48977311</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48977311</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48977311</guid></item><item><title><![CDATA[New comment by kllrnohj in "Claude Code uses Bun written in Rust now"]]></title><description><![CDATA[
<p>> You could write a JS engine with Zig-like idioms (arena allocation, static initialization)<p>Arena allocators & static initializers are not novel. You'll find them in high performance C++ projects as well, such as LLVM or JavaScriptCore. But arena allocators have the quite significant limitation that they only help when everything being allocated in them have approximately the same lifetime. So they don't help when you need to allocate memory to provide the native implementation of a JavaScript object, for example (eg, FFI).</p>
]]></description><pubDate>Mon, 20 Jul 2026 05:59:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=48974826</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48974826</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48974826</guid></item><item><title><![CDATA[New comment by kllrnohj in "Claude Code uses Bun written in Rust now"]]></title><description><![CDATA[
<p>> You can just do that, and then Zig is really no less robust than Rust.<p>If you just don't write bugs, then yes all languages are equally robust, including assembly.<p>Zig, like C, is simply not a robust language. I don't know why this feels like something contentious? It's clearly not intended to be robust?</p>
]]></description><pubDate>Mon, 20 Jul 2026 03:07:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=48973944</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48973944</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48973944</guid></item><item><title><![CDATA[New comment by kllrnohj in "Claude Code uses Bun written in Rust now"]]></title><description><![CDATA[
<p>Re #3: vtable pointers aren't mutable...?</p>
]]></description><pubDate>Mon, 20 Jul 2026 03:05:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=48973933</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48973933</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48973933</guid></item><item><title><![CDATA[New comment by kllrnohj in "The Kimi K3 Moment"]]></title><description><![CDATA[
<p>Because the transformer architecture that enabled modern LLMs wasn't invented until 2017[1]?<p>1: That's the "T" in GPT fyi, even though Google is the author of the research paper that changed everything</p>
]]></description><pubDate>Sat, 18 Jul 2026 21:11:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=48962452</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48962452</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48962452</guid></item><item><title><![CDATA[New comment by kllrnohj in "Kimi K3, and what we can still learn from the pelican benchmark"]]></title><description><![CDATA[
<p>> Bytedance and Tencent are tech giants for sure, nonetheless they’re not Google kind of giant.<p>$186 billion and $105 billion revenue in 2025 respectively vs. $402 billion? Yes, Google is larger, but they're all in that same ballpark?<p>ByteDance's 2025 net income isn't that different from Anthropic's Series H funding even ($50bn vs $65bn respectively).<p>But this is all also ignoring how much of China is state owned (25% of the GDP!), so the available resource pool is <i>dramatically</i> larger than it would appear depending on what the government decides is important</p>
]]></description><pubDate>Sat, 18 Jul 2026 12:49:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=48957690</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48957690</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48957690</guid></item><item><title><![CDATA[New comment by kllrnohj in "Kimi K3, and what we can still learn from the pelican benchmark"]]></title><description><![CDATA[
<p>Or they just don't actually have any compute access restrictions of significance? Chinese companies can just go use those GPUs in neighboring countries that aren't export-restricted, like Malaysia. Like ByteDance openly did: <a href="https://www.tomshardware.com/pc-components/gpus/chinas-bytedance-to-access-36-000-blackwell-gpu-cluster-through-malaysia-cloud-operator-nvidia-confirms-no-objections-deal-is-in-line-with-us-export-controls" rel="nofollow">https://www.tomshardware.com/pc-components/gpus/chinas-byted...</a><p>and Tencent is rumored to have done via Japan: <a href="https://wccftech.com/china-tencent-gains-access-to-nvidia-blackwell-ai-chips-by-leveraging-the-rental-loophole/" rel="nofollow">https://wccftech.com/china-tencent-gains-access-to-nvidia-bl...</a><p>And that's not even considering just smuggling the GPUs in by eg buying them in Singapore.<p>AI-specific chips also seem to be on the easier side to design & create relative to high performance CPUs & GPUs, so there's no particular reason to expect Chinese domestic designs to continuously lag behind. They have access to the same fabs, after all</p>
]]></description><pubDate>Fri, 17 Jul 2026 21:47:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=48952598</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48952598</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48952598</guid></item><item><title><![CDATA[New comment by kllrnohj in "Rob Pike – 'Concurrency Is Not Parallelism' [video] (2012)"]]></title><description><![CDATA[
<p>> and a proper way to write parallel programs for when async would lead to all the bad stuff async leads to.<p>you need real threads for parallelism though? And real threads scale just fine such that m:n threading isn't typically needed.</p>
]]></description><pubDate>Sun, 05 Jul 2026 19:23:14 +0000</pubDate><link>https://news.ycombinator.com/item?id=48797143</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48797143</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48797143</guid></item><item><title><![CDATA[New comment by kllrnohj in "Rob Pike – 'Concurrency Is Not Parallelism' [video] (2012)"]]></title><description><![CDATA[
<p>That's because they are very rarely useful. This was true then and it's still true now. There's just not many workloads where it makes sense to need to rapidly launch a thread that doesn't need to do much of anything but does need to exist for a while before terminating.<p>What is useful is the state machine aspects of things like coroutines or async/await, but those aren't quite fibers and very much aren't M:N threading. A major use of them is in UI where they have strict thread requirements even.</p>
]]></description><pubDate>Sat, 04 Jul 2026 19:33:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=48788223</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48788223</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48788223</guid></item><item><title><![CDATA[New comment by kllrnohj in "Qwen 3.6 27B is the sweet spot for local development"]]></title><description><![CDATA[
<p>> maybe tell us how much a non-Apple system that you can run that (probably similarly or faster) would cost?<p>Ryzen AI Max 395+ with 128GB of unified memory can be found around $3-4k.<p>But 27B isn't <i>that</i> large, either, especially if you are ok with the quantized models. So this laptop choice seems to more be a "because they had it" rather than "this is what's necessary for this particular workflow"</p>
]]></description><pubDate>Mon, 29 Jun 2026 17:49:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=48722559</link><dc:creator>kllrnohj</dc:creator><comments>https://news.ycombinator.com/item?id=48722559</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48722559</guid></item></channel></rss>