<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: kvark</title><link>https://news.ycombinator.com/user?id=kvark</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Fri, 20 Nov 2026 18:35:18 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=kvark" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by kvark in "How I Fight AI Brain Fry"]]></title><description><![CDATA[
<p>Why inventing your own issue/work tracking system based on git? Wouldn't it be easier to work with, say, github issues the same way? You create it normally, agent picks up, works, updates the status on it or in the relevant wip PR, then potentially adding "needs more info" label. You'd get all the same tracking benefits and more: better UI, kanban, more tools.</p>
]]></description><pubDate>Sun, 08 Nov 2026 20:33:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=50416878</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=50416878</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50416878</guid></item><item><title><![CDATA[New comment by kvark in "The end of Microsoft Research's AI Frontiers group"]]></title><description><![CDATA[
<p>Autogen is open source. I don't know the availability of other projects.
Generally engineers in megacorps aren't shy of picking up the latest AI tools. I suppose Microsoft may have barriers around this though, curious to discover more details.<p>Edit: Fara is open weight as well. Seems like a great match for <a href="https://www.microsoft.com/en-us/windows-365/agents" rel="nofollow">https://www.microsoft.com/en-us/windows-365/agents</a> ?</p>
]]></description><pubDate>Sat, 07 Nov 2026 19:20:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=50407190</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=50407190</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50407190</guid></item><item><title><![CDATA[New comment by kvark in "The end of Microsoft Research's AI Frontiers group"]]></title><description><![CDATA[
<p>His complaints sound childish to me. If the research products were truly innovative and useful, he wouldn't need to force Microsoft to adopt them.</p>
]]></description><pubDate>Sat, 07 Nov 2026 19:10:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=50407124</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=50407124</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50407124</guid></item><item><title><![CDATA[New comment by kvark in "Valen's Memory Safety: A New Kind of Borrow Checking"]]></title><description><![CDATA[
<p>It sounds more like "path" borrowing than "group" borrowing to me, but the idea is great! Quite eye opening!<p>One minor thing is that I'm not sure why they had to have "in" keyword for sub-borrows. Wouldn't it be more consistent to see "entity &world.entities[?]" instead of  "entity in world.entities[]"? We'd consistently get the borrow "&" symbol and open the door for some contracts on what index (range) is affected.</p>
]]></description><pubDate>Sun, 11 Oct 2026 15:11:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=50044156</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=50044156</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50044156</guid></item><item><title><![CDATA[Rendering Shooter]]></title><description><![CDATA[
<p>Article URL: <a href="https://drawcall.grok.me/">https://drawcall.grok.me/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49424025">https://news.ycombinator.com/item?id=49424025</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 24 Aug 2026 18:32:43 +0000</pubDate><link>https://drawcall.grok.me/</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=49424025</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49424025</guid></item><item><title><![CDATA[New comment by kvark in "Is it all just vapourware?"]]></title><description><![CDATA[
<p>Or maybe open-source development is just poorly compatible with AI workflow? Today, projects may need a community, an issue tracker, but pull requests are becoming less important.<p>I suspect the ban of AI in established projects to be a very complex decision. Even if core developers would like to use AI, they don't want to review all of the AI-generated code from the larger community. So the only consistent way to preserve sanity is to declare that nobody would use it. Good luck enforcing it though!</p>
]]></description><pubDate>Sun, 09 Aug 2026 21:52:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=49236440</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=49236440</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49236440</guid></item><item><title><![CDATA[New comment by kvark in "You can't solve computer use by ignoring the interface"]]></title><description><![CDATA[
<p>We have 2 paths here:<p>- either evolve our UI systems for computers and let them operate efficiently, i.e. converting everything to API calls
. Frontier labs are pushing for this, starting with the Chrome plugin that Claude Cowork mandates.<p>- or make AI systems use the screen like humans do - what the article is arguing for<p>Both of these paths are possible. One of them is much more efficient than the other in the long term. Another one may be inevitable.</p>
]]></description><pubDate>Thu, 30 Jul 2026 15:05:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49111099</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=49111099</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49111099</guid></item><item><title><![CDATA[New comment by kvark in "I Am Not a Reverse Centaur"]]></title><description><![CDATA[
<p>We had a process at one company where you had to create an issue before filing a PR. I found it most non-sensical and introducing friction for no good reason. Very surprised to see the author suggesting it in the article.<p>Review is indeed the main bottleneck now for open source, and we need to solve it. Introducing more friction is hardly helping.</p>
]]></description><pubDate>Fri, 12 Jun 2026 18:35:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=48507767</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=48507767</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48507767</guid></item><item><title><![CDATA[New comment by kvark in "Local ML inference benchmark: PyTorch vs. llama.cpp vs. the Rust ecosystem"]]></title><description><![CDATA[
<p>Author of the benchmark here, as well as Meganeura - one of the tested frameworks. Results are constantly being refined, gaps are filled, but it's hard to get more hardware coverage by myself. Would appreciate someone running it on Intel Arc GPUs, both discreet and integrated. Happy to answer questions about methodology!</p>
]]></description><pubDate>Mon, 20 Apr 2026 14:10:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=47834640</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47834640</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47834640</guid></item><item><title><![CDATA[Local ML inference benchmark: PyTorch vs. llama.cpp vs. the Rust ecosystem]]></title><description><![CDATA[
<p>Article URL: <a href="http://kvark.github.io/ai/performance/2026/04/19/tales-from-the-inference-land.html">http://kvark.github.io/ai/performance/2026/04/19/tales-from-the-inference-land.html</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47834585">https://news.ycombinator.com/item?id=47834585</a></p>
<p>Points: 2</p>
<p># Comments: 1</p>
]]></description><pubDate>Mon, 20 Apr 2026 14:07:02 +0000</pubDate><link>http://kvark.github.io/ai/performance/2026/04/19/tales-from-the-inference-land.html</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47834585</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47834585</guid></item><item><title><![CDATA[Inferena: Local benchmark of PyTorch vs. Llama.cpp vs. Rust frameworks]]></title><description><![CDATA[
<p>Article URL: <a href="http://inferena.tech/">http://inferena.tech/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47820109">https://news.ycombinator.com/item?id=47820109</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Sat, 18 Apr 2026 22:33:58 +0000</pubDate><link>http://inferena.tech/</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47820109</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47820109</guid></item><item><title><![CDATA[New comment by kvark in "Inference Arena – new benchmark of local inference and training"]]></title><description><![CDATA[
<p>Thank you for support!<p>Part of the problem with timing variety is frameworks not always picking the right gpu/backend.<p>If you want to inspect or tweak the setup, be my guest at <a href="https://github.com/kvark/inferena" rel="nofollow">https://github.com/kvark/inferena</a></p>
]]></description><pubDate>Sun, 05 Apr 2026 18:28:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=47652385</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47652385</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47652385</guid></item><item><title><![CDATA[Inference Arena – new benchmark of local inference and training]]></title><description><![CDATA[
<p>Article URL: <a href="http://kvark.github.io/ai/performance/2026/04/04/inference-arena.html">http://kvark.github.io/ai/performance/2026/04/04/inference-arena.html</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47650440">https://news.ycombinator.com/item?id=47650440</a></p>
<p>Points: 4</p>
<p># Comments: 1</p>
]]></description><pubDate>Sun, 05 Apr 2026 15:30:22 +0000</pubDate><link>http://kvark.github.io/ai/performance/2026/04/04/inference-arena.html</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47650440</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47650440</guid></item><item><title><![CDATA[Inferena – benchmarking inference of popular models on consumer hardware]]></title><description><![CDATA[
<p>Article URL: <a href="http://inferena.tech/">http://inferena.tech/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47645390">https://news.ycombinator.com/item?id=47645390</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Sun, 05 Apr 2026 01:48:36 +0000</pubDate><link>http://inferena.tech/</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47645390</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47645390</guid></item><item><title><![CDATA[New comment by kvark in "Get free Claude max 20x for open-source maintainers"]]></title><description><![CDATA[
<p>Sadly the form doesn't even show up in Firefox. If they want to appeal to computer nerds, gotta anticipate not everyone will be on Blink engine.</p>
]]></description><pubDate>Sat, 28 Feb 2026 05:53:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=47190990</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=47190990</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47190990</guid></item><item><title><![CDATA[New comment by kvark in "Simplifying Vulkan one subsystem at a time"]]></title><description><![CDATA[
<p>The main problem with Vulkan isn't the programming model or the lack of features. These are tackled by Khronos. The problem is with coverage and update distribution. It's all over the place! If you develop general purpose software (like Zed), you can't assume that even the basic things like dynamic rendering are supported uniformly. There are always weird systems with old drivers (looking at Ubuntu 22 LTS), hardware vendors abandoning and forcefully deprecating the working hardware, and of course driver bugs...
So, by the time I'm going to be able to rely on the new shiny descriptor heap/buffer features, I'll have more gray hair and other things on the horizon.</p>
]]></description><pubDate>Tue, 10 Feb 2026 16:03:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=46961720</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=46961720</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46961720</guid></item><item><title><![CDATA[New comment by kvark in "The Burrows-Wheeler Transform"]]></title><description><![CDATA[
<p>I've spent years playing with BWT and suffix sorting at school (some work can be found be names of archon and dark). It's a beautiful domain to learn!<p>Now I'm wondering: in the era of software 2.0, everything is figured out by AI. What are the chances AI would discover this algorithm at all?</p>
]]></description><pubDate>Fri, 10 Oct 2025 04:26:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=45535365</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=45535365</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45535365</guid></item><item><title><![CDATA[New comment by kvark in "Why Blender Changing to Vulkan Is Groundbreaking [video]"]]></title><description><![CDATA[
<p>I'm not employed by Zed, so I can't answer these questions. But the product itself is great!</p>
]]></description><pubDate>Tue, 03 Jun 2025 17:10:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=44172265</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=44172265</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44172265</guid></item><item><title><![CDATA[New comment by kvark in "Why Blender Changing to Vulkan Is Groundbreaking [video]"]]></title><description><![CDATA[
<p>Sure. Maybe that's the forcing function for them to nail down all the kinks. Example: <a href="https://github.com/kvark/blade/issues/205">https://github.com/kvark/blade/issues/205</a></p>
]]></description><pubDate>Sun, 01 Jun 2025 23:35:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=44154580</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=44154580</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44154580</guid></item><item><title><![CDATA[New comment by kvark in "Why Blender Changing to Vulkan Is Groundbreaking [video]"]]></title><description><![CDATA[
<p>Shipping Vulkan in production on Linux is a challenge. Chrome was dealing with it for a while. Recently, with Zed ported to Vulkan, we saw the variety of half-broken platforms and user configurations.<p>I'd recommend Blended to not close the door on OpenGL and instead keeping it as a compatibility fallback.</p>
]]></description><pubDate>Sun, 01 Jun 2025 22:55:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=44154352</link><dc:creator>kvark</dc:creator><comments>https://news.ycombinator.com/item?id=44154352</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44154352</guid></item></channel></rss>