<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: ashvardanian</title><link>https://news.ycombinator.com/user?id=ashvardanian</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Thu, 02 Jul 2026 22:56:00 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=ashvardanian" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by ashvardanian in "Claude Sonnet 5"]]></title><description><![CDATA[
<p>Got really excited for this model and asked my Opus planners in 3 pretty different projects to use Sonnets instead of Opus subagents to help me experiment on HPC kernels faster. Not one of them ended up writing a single line of code... Sonnets just kept spinning, wasting tokens. Can't remember the last time it happened with Opus in my codebases. Reverting back.</p>
]]></description><pubDate>Tue, 30 Jun 2026 23:50:48 +0000</pubDate><link>https://news.ycombinator.com/item?id=48740737</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=48740737</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48740737</guid></item><item><title><![CDATA[KinetIQ Ascend: Toward 100% Reliable Manipulation and Superhuman Speed]]></title><description><![CDATA[
<p>Article URL: <a href="https://thehumanoid.ai/technology/kinetiq-ascend/">https://thehumanoid.ai/technology/kinetiq-ascend/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=48686654">https://news.ycombinator.com/item?id=48686654</a></p>
<p>Points: 5</p>
<p># Comments: 2</p>
]]></description><pubDate>Fri, 26 Jun 2026 13:51:45 +0000</pubDate><link>https://thehumanoid.ai/technology/kinetiq-ascend/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=48686654</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48686654</guid></item><item><title><![CDATA[Nvidia CUDA Python 1.0 and CUDA 13.3 Release]]></title><description><![CDATA[
<p>Article URL: <a href="https://developer.nvidia.com/blog/nvidia-cuda-13-3-enhances-gpu-development-with-tile-programming-in-c-compiler-autotuning-and-python-updates/">https://developer.nvidia.com/blog/nvidia-cuda-13-3-enhances-gpu-development-with-tile-programming-in-c-compiler-autotuning-and-python-updates/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=48468120">https://news.ycombinator.com/item?id=48468120</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 09 Jun 2026 21:38:00 +0000</pubDate><link>https://developer.nvidia.com/blog/nvidia-cuda-13-3-enhances-gpu-development-with-tile-programming-in-c-compiler-autotuning-and-python-updates/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=48468120</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48468120</guid></item><item><title><![CDATA[New comment by ashvardanian in "Cloudflare Flagship"]]></title><description><![CDATA[
<p>I really like the speed at which Cloudflare is executing toward becoming a critical infrastructure player with all of those new product offerings. That said, not everything needs to be serverless. Their Gen 13 hardware looks impressive, and it’s a pity you can’t rent it by the hour like AWS EC2 Metal instances.</p>
]]></description><pubDate>Wed, 27 May 2026 14:42:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=48295153</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=48295153</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48295153</guid></item><item><title><![CDATA[Laguna XS.2 and Laguna M.1 by Poolside]]></title><description><![CDATA[
<p>Article URL: <a href="https://poolside.ai/blog/introducing-laguna-xs2-m1">https://poolside.ai/blog/introducing-laguna-xs2-m1</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47936628">https://news.ycombinator.com/item?id=47936628</a></p>
<p>Points: 5</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 28 Apr 2026 16:25:02 +0000</pubDate><link>https://poolside.ai/blog/introducing-laguna-xs2-m1</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47936628</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47936628</guid></item><item><title><![CDATA[FP8 Search and KV-Caching in USearch]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.unum.cloud/blog/float8">https://www.unum.cloud/blog/float8</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47837721">https://news.ycombinator.com/item?id=47837721</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 20 Apr 2026 17:31:19 +0000</pubDate><link>https://www.unum.cloud/blog/float8</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47837721</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47837721</guid></item><item><title><![CDATA[Escaping the Fork: How Meta Modernized WebRTC Across 50 Use Cases]]></title><description><![CDATA[
<p>Article URL: <a href="https://engineering.fb.com/2026/04/09/developer-tools/escaping-the-fork-how-meta-modernized-webrtc-across-50-use-cases/">https://engineering.fb.com/2026/04/09/developer-tools/escaping-the-fork-how-meta-modernized-webrtc-across-50-use-cases/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47783478">https://news.ycombinator.com/item?id=47783478</a></p>
<p>Points: 3</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 15 Apr 2026 18:49:11 +0000</pubDate><link>https://engineering.fb.com/2026/04/09/developer-tools/escaping-the-fork-how-meta-modernized-webrtc-across-50-use-cases/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47783478</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47783478</guid></item><item><title><![CDATA[Porting Go's io package to C]]></title><description><![CDATA[
<p>Article URL: <a href="https://antonz.org/porting-go-io/">https://antonz.org/porting-go-io/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47520573">https://news.ycombinator.com/item?id=47520573</a></p>
<p>Points: 6</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 25 Mar 2026 17:33:26 +0000</pubDate><link>https://antonz.org/porting-go-io/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47520573</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47520573</guid></item><item><title><![CDATA[Schema as the Core of Reliability in AI Memory]]></title><description><![CDATA[
<p>Article URL: <a href="https://xmemory.ai/schema-as-the-core-of-reliability/">https://xmemory.ai/schema-as-the-core-of-reliability/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47492775">https://news.ycombinator.com/item?id=47492775</a></p>
<p>Points: 6</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 23 Mar 2026 17:48:30 +0000</pubDate><link>https://xmemory.ai/schema-as-the-core-of-reliability/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47492775</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47492775</guid></item><item><title><![CDATA[New comment by ashvardanian in "NumKong: 2'000 Mixed Precision Kernels for All"]]></title><description><![CDATA[
<p>I'm not aware of that, but it would likely be a great application area for SME!</p>
]]></description><pubDate>Sat, 21 Mar 2026 17:51:06 +0000</pubDate><link>https://news.ycombinator.com/item?id=47469400</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47469400</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47469400</guid></item><item><title><![CDATA[New comment by ashvardanian in "NumKong: 2'000 Mixed Precision Kernels for All"]]></title><description><![CDATA[
<p>The README was written by a human. I’ve used models extensively to refine the content, but never accepted more than a couple of lines of edits at a time.</p>
]]></description><pubDate>Sat, 21 Mar 2026 05:56:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=47464359</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47464359</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47464359</guid></item><item><title><![CDATA[NumKong: 2'000 Mixed Precision Kernels for All]]></title><description><![CDATA[
<p>Article URL: <a href="https://ashvardanian.com/posts/numkong/">https://ashvardanian.com/posts/numkong/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47459447">https://news.ycombinator.com/item?id=47459447</a></p>
<p>Points: 47</p>
<p># Comments: 6</p>
]]></description><pubDate>Fri, 20 Mar 2026 19:29:39 +0000</pubDate><link>https://ashvardanian.com/posts/numkong/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47459447</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47459447</guid></item><item><title><![CDATA[The State of Allocators in 2026]]></title><description><![CDATA[
<p>Article URL: <a href="https://cetra3.github.io/blog/state-of-allocators-2026/">https://cetra3.github.io/blog/state-of-allocators-2026/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47396806">https://news.ycombinator.com/item?id=47396806</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 16 Mar 2026 09:32:10 +0000</pubDate><link>https://cetra3.github.io/blog/state-of-allocators-2026/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47396806</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47396806</guid></item><item><title><![CDATA[New comment by ashvardanian in "What makes Intel Optane stand out (2023)"]]></title><description><![CDATA[
<p>I don't have the inside scoop on Intel's current mess, but they definitely have a habit of killing off their coolest projects.</p>
]]></description><pubDate>Sun, 15 Mar 2026 15:46:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=47388532</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47388532</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47388532</guid></item><item><title><![CDATA[New comment by ashvardanian in "Meta's Race to Scale AI Chips for Billions: Four Chips in Two Years"]]></title><description><![CDATA[
<p>Would it be accurate to say that Meta currently produces more RISC-V chips than other vendors? The specs for those chips look extremely interesting and seem much more programmable than Google's TPUs. It would be cool to see Meta making them available to third parties.</p>
]]></description><pubDate>Wed, 11 Mar 2026 15:37:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=47337005</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47337005</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47337005</guid></item><item><title><![CDATA[Recraft V4]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.recraft.ai/blog/introducing-recraft-v4-design-taste-meets-image-generation">https://www.recraft.ai/blog/introducing-recraft-v4-design-taste-meets-image-generation</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47059810">https://news.ycombinator.com/item?id=47059810</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 18 Feb 2026 11:13:44 +0000</pubDate><link>https://www.recraft.ai/blog/introducing-recraft-v4-design-taste-meets-image-generation</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47059810</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47059810</guid></item><item><title><![CDATA[New comment by ashvardanian in "Nebius to buy AI agent search company Tavily for 275M"]]></title><description><![CDATA[
<p><a href="https://www.bloomberg.com/news/articles/2026-02-10/nebius-agrees-to-buy-ai-agent-search-company-tavily-for-275-million" rel="nofollow">https://www.bloomberg.com/news/articles/2026-02-10/nebius-ag...</a><p><a href="https://www.tavily.com/blog/tavily-is-joining-nebius" rel="nofollow">https://www.tavily.com/blog/tavily-is-joining-nebius</a></p>
]]></description><pubDate>Sun, 15 Feb 2026 14:58:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=47024217</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47024217</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47024217</guid></item><item><title><![CDATA[Nebius to buy AI agent search company Tavily for 275M]]></title><description><![CDATA[
<p>Article URL: <a href="https://nebius.com/newsroom/nebius-announces-agreement-to-acquire-tavily-to-add-agentic-search-to-its-ai-cloud-platform">https://nebius.com/newsroom/nebius-announces-agreement-to-acquire-tavily-to-add-agentic-search-to-its-ai-cloud-platform</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47024216">https://news.ycombinator.com/item?id=47024216</a></p>
<p>Points: 2</p>
<p># Comments: 1</p>
]]></description><pubDate>Sun, 15 Feb 2026 14:58:56 +0000</pubDate><link>https://nebius.com/newsroom/nebius-announces-agreement-to-acquire-tavily-to-add-agentic-search-to-its-ai-cloud-platform</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47024216</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47024216</guid></item><item><title><![CDATA[New comment by ashvardanian in "Zvec: A lightweight, fast, in-process vector database"]]></title><description><![CDATA[
<p>8K QPS is probably quite trivial on their setup and a 10M dataset. I rarely use comparably small instances & datasets in my benchmarks, but on 100M-1B datasets on a larger dual-socket server, 100K QPS was easily achievable in 2023: <a href="https://www.unum.cloud/blog/2023-11-07-scaling-vector-search-with-intel" rel="nofollow">https://www.unum.cloud/blog/2023-11-07-scaling-vector-search...</a> ;)<p>Typically, the recipe is to keep the hot parts of the data structure in SRAM in CPU caches and a lot of SIMD. At the time of those measurements, USearch used ~100 custom kernels for different data types, similarity metrics, and hardware platforms. The upcoming release of the underlying SimSIMD micro-kernels project will push this number beyond 1000. So we should be able to squeeze a lot more performance later this year.</p>
]]></description><pubDate>Sun, 15 Feb 2026 00:44:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=47020007</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47020007</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47020007</guid></item><item><title><![CDATA[Running Async WebAssembly on Seastar's Reactor]]></title><description><![CDATA[
<p>Article URL: <a href="https://rockwotj.com/blog/async-wasm-in-seastar/">https://rockwotj.com/blog/async-wasm-in-seastar/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47015044">https://news.ycombinator.com/item?id=47015044</a></p>
<p>Points: 3</p>
<p># Comments: 0</p>
]]></description><pubDate>Sat, 14 Feb 2026 15:04:08 +0000</pubDate><link>https://rockwotj.com/blog/async-wasm-in-seastar/</link><dc:creator>ashvardanian</dc:creator><comments>https://news.ycombinator.com/item?id=47015044</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47015044</guid></item></channel></rss>