<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: ThunderSizzle</title><link>https://news.ycombinator.com/user?id=ThunderSizzle</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sun, 14 Jun 2026 22:39:22 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=ThunderSizzle" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by ThunderSizzle in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"]]></title><description><![CDATA[
<p>An R9700 is $1350 and can get 100 TPS running Qwen3.6-35B-A3B Q5 with 130k context window (with room to spare) with a bit of fine tuning llamacpp-vulkan, but llamacpp's repository instability and lack of real versioning frustrates me.<p>In terms of electricity, if you aren't using it, even with all the vram loaded, at most your wasting about 30 watts or so.<p>Prompt processing a large uncached context is annoying, which is why I forced a lower context window, but I don't know if it's any worse in performance than the cloud models I've used.<p>There's a niceness, to me, knowing I don't have to rent it anymore. If you rent it, the terms can change regularly.</p>
]]></description><pubDate>Sat, 13 Jun 2026 16:11:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=48518616</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48518616</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48518616</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Why I'm Forced to Say Farewell: Google Management Has Lost Its Moral Compass"]]></title><description><![CDATA[
<p>Like a compliment sandwich or something.</p>
]]></description><pubDate>Fri, 12 Jun 2026 01:34:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=48498759</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48498759</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48498759</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Show HN: FablePool – pool money behind a prompt, and Fable builds it in public"]]></title><description><![CDATA[
<p>The coding isn't the hard part. It's the people and networking. Facebook's only moat is HOA boards that think private communication behind Facebook groups somehow equates to public messaging a community...<p>In other words, once people got on it, it was too late.</p>
]]></description><pubDate>Fri, 12 Jun 2026 01:28:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=48498708</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48498708</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48498708</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Uber's $1,500/month AI limit is a useful signal for AI tool pricing"]]></title><description><![CDATA[
<p>I hope this is sarcasm, or "half" my job doesn't exist or something. Or you talking about full time non-dev scrum masters?</p>
]]></description><pubDate>Wed, 03 Jun 2026 23:40:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=48391648</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48391648</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48391648</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Nvidia RTX Spark"]]></title><description><![CDATA[
<p>My 980 is currently unpowered. I have not tried to integrate them yet. It's on my to-do list, but the first time they were both powered on, the system had booting issues, and I didn't want to care at the time, since the 980 was probably going to be idle 99% of the time anyway.<p>I'll probably try to figure that problem out in about a month. Worst case is I move it to another even older desktop to replace the 9800 GTX+ inside of that one.</p>
]]></description><pubDate>Tue, 02 Jun 2026 10:52:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=48368507</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48368507</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48368507</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Nvidia RTX Spark"]]></title><description><![CDATA[
<p>GitHub Copilot. It was one of the best values around in terms of cheap LLM access since each prompt was basically 4 cents (more or less), no matter how much it would do or how many tokens it used. A simple "Proceed" prompt that was telling the agent to execute a sophisticated plan could burn a lot of time without needing any direct intervention by the user, but as of June 1st, they switched to metered billing, meaning each token in/out has a cost now.<p>It was suspected to come soon enough, but it was a nice cheap road for my small hobby stuff. When they announced the price changes, I started to explore alternatives, and with the news of Qwen3.6 35B being both and having quality, I figured it was worth a try out, and self-hosting made the most sense to me, since that meant I was free from being a forever-renter.</p>
]]></description><pubDate>Tue, 02 Jun 2026 04:00:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=48365863</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48365863</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48365863</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Nvidia RTX Spark"]]></title><description><![CDATA[
<p>I added an R9700 32GB to my 10+ year old desktop that had a 980 4GB card in it, for a grand total of $1350 or so. The payoff compared to what I was using with GHCP was 33 months, but when GHCP announced their price increase, it basically became a 3 month payoff at minimum (so yes, GHCP did a 10x price increase for non-parallel agentic workflows)<p>I can easily run Qwen3.6 35B-A3B with Q5_K_M with a 260k+ context window with some vram to spare. It easily runs probably 80tps. It took me quite a while to find the<p>Compared to GHCP Claude Sonnet 4.5 or 4.6, I have full parity. The wall clock time is faster for agentic workflows, and rule following is about on par.<p>With either, doing something kind of novel or obscure takes more hand holding compared to just generate a GUI or crud app. For example, trying to build an actual program that performs a complicated process correctly requires quite a bit of hand holding to get it to properly help.<p>Sure, it isn't Opus or something, but I think with the right harness, it probably can get close. I think most of the issues these days is the harnesses are lacking.</p>
]]></description><pubDate>Tue, 02 Jun 2026 01:28:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=48364797</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48364797</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48364797</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Why are large language models so terrible at video games?"]]></title><description><![CDATA[
<p>I wonder if you paired a few different types of AI together, an LLM agent might be good at strategizing -. E.g. building a strategy on how to handle a scenario. But, it would need to know the entire game manual basically. Then it would pass the stratrgy to a better AI in some way. But it might not be needed if the better gaming AI can just do that part too already.<p>I admit I know nothing about this though.</p>
]]></description><pubDate>Mon, 01 Jun 2026 10:20:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=48354881</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48354881</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48354881</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Is AI causing a repeat of frontend’s lost decade?"]]></title><description><![CDATA[
<p>My instructions and steering is to force the LLM agent to focus on mocking only system boundaries and outside in unit testing. I've found these two make verbose but good tests that actually prove behavior pretty well.<p>But that was my testing strategy already, but writing one of those tests could take legitimately hours with how many columns and crazy rules our area has.</p>
]]></description><pubDate>Sat, 30 May 2026 02:59:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=48332048</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48332048</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48332048</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "DuckDuckGo search saw 28% more visits after Google said people love AI mode"]]></title><description><![CDATA[
<p>There's a lot of reasons, but it's probably how Google makes most of their money - brand names end up being high value ad targets - at least 6+ years ago when I worked in that sphere of things (ecommerce, primarily). It's legit extortion.<p>I guess that's why - because extortion is immoral. Eventually, those paying the protection fees will eventually find someone better to pay or will stop paying when the protection doesn't show up.</p>
]]></description><pubDate>Thu, 28 May 2026 19:22:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=48314107</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48314107</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48314107</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "DuckDuckGo search saw 28% more visits after Google said people love AI mode"]]></title><description><![CDATA[
<p>Right, but that's because Google permits that. OP is saying Google should simply stop permitting advertising over the real answer.</p>
]]></description><pubDate>Thu, 28 May 2026 10:44:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=48307132</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48307132</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48307132</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "YAML? That's Norway Problem"]]></title><description><![CDATA[
<p>Fine. Then auto-closing tags? If you have a schema doc, you could do that. If you are writing free-form, then I'd definitely be complicated...?<p>I still see that better than YAML's indentation problem. Relying solely on indentation is a nightmare.</p>
]]></description><pubDate>Sun, 24 May 2026 00:32:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=48253067</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48253067</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48253067</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Microsoft starts canceling Claude Code licenses"]]></title><description><![CDATA[
<p>If you do manual edits, I find it best to start a new conversation. But if your instructions and documentation is good enough, the new conversations won't have any problems picking up where it needs to be.<p>Having said that, I fear what June 1st brings for copilot
 It might suddenly be very useless for me.</p>
]]></description><pubDate>Sat, 23 May 2026 11:05:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=48246658</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48246658</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48246658</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "YAML? That's Norway Problem"]]></title><description><![CDATA[
<p>It wouldn't need to be if closing tags were allowed to be unnamed. For most cases, we can tell the closing tags easily enough for simpler files:<p><pre><code>  <project>MAML</>
  <tags>
    <item>minimal</>
    <item>readable</>
  </>
  
  <!-- A simple nested object -->
  <spec>
    <version>1</>
    <author>Anton Medvedev</>
  </>

  <!-- Array of objects -->
  <examples>
    <item>
      <name>JSON</>
      <born>2001</>
    </>
    <item>
      <name>MAML</>
      <born>2025</>
    </>
  </>

  <notes>
  This is a multiline raw strings.
  Keeps formatting as-is.
  </></code></pre></p>
]]></description><pubDate>Sat, 23 May 2026 02:02:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=48243824</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48243824</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48243824</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Intuit to lay off over 3k employees to refocus on AI"]]></title><description><![CDATA[
<p>Freetaxusa isn't. I've done complicated schedule K, etc forms on it</p>
]]></description><pubDate>Thu, 21 May 2026 10:51:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=48220543</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48220543</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48220543</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Vivaldi 8.0"]]></title><description><![CDATA[
<p>But what's your argument? That phone-based extensions are more vulnerable somehow than desktop extensions?<p>If anything, wouldn't a phone extension be more sandboxed than most desktop environments?</p>
]]></description><pubDate>Thu, 21 May 2026 10:41:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=48220469</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48220469</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48220469</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Local AI needs to be the norm"]]></title><description><![CDATA[
<p>What if you split it into less complex tasks? E.g. use the model to help decompose the task into parts, then help it iterate through it.<p>Gives you more control over the outcome and more steering anyway.</p>
]]></description><pubDate>Sat, 16 May 2026 00:27:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=48155633</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48155633</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48155633</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "The US is winning the AI race where it matters most: commercialization"]]></title><description><![CDATA[
<p>A Qwen3.6-35B-A3B or whatever it's full name is, when on a 3090, can at the very least, with very little fine tuning, compete with Haiku and blows away GPT4.1 (aka, the cheap models).<p>It might keep up with Sonnet 4.5 with some tinkering.<p>But long story short: it seems to have better performance and similar quality for a payoff of a year or so compared to cloud models. In the same way you can self host faster/easier/cheaper than cloud hosting, if you are okay with the negatives.<p>I'm returning my 3090 soon for a R9700 after some more basic benchmarking, since the higher RAM should improve my observations more.</p>
]]></description><pubDate>Wed, 13 May 2026 19:39:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=48126442</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48126442</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48126442</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Cloudflare to cut about 20% of its workforce"]]></title><description><![CDATA[
<p>I've heard this before.<p>Then COVID happened and employers didn't have 2-4 months of savings built up, and ended up shuttering due to lack of money immediately.<p>Also, since post COVID, we've had hyper inflation and a locked up housing market. $150 townhomes that a $50 family salary could afford are now going for $300. And rent has gone up to match.<p>I don't think the numbers not matching is because of anyone's personal financial responsibility. It's more from the Fed's and Congress's horrible actions over the past 2 decades and their financial irresponsibility.<p>Granted, grumbling about the powers that be doesn't solve the problem, which is why I fear civil turmoil will be here very soon.</p>
]]></description><pubDate>Sat, 09 May 2026 10:47:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=48073852</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=48073852</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48073852</guid></item><item><title><![CDATA[New comment by ThunderSizzle in "Uber wants to turn its drivers into a sensor grid for self-driving companies"]]></title><description><![CDATA[
<p>Many American roads don't have lines. Residential roads, parking lots, many business driveways have limited markings.<p>Then there's roads with just the center line markers with no road should markings.<p>Then there's a whole class of roads of lines over "demarked" old lines that weren't demarked well, or lines fading that should've been painted a long time ago.<p>I'm surprised you've never seen a non-perfect road?</p>
]]></description><pubDate>Sat, 02 May 2026 19:20:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=47989512</link><dc:creator>ThunderSizzle</dc:creator><comments>https://news.ycombinator.com/item?id=47989512</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47989512</guid></item></channel></rss>