https://openmtx.com/posts/Recent content in Posts on openmtxHugoen-usThu, 02 Jul 2026 00:00:00 +0000https://openmtx.com/posts/the-other-5000/Thu, 02 Jul 2026 00:00:00 +0000https://openmtx.com/posts/the-other-5000/<p>It started with an eBay seller I&rsquo;d bought from before, and a number: 5000.</p> <p>I&rsquo;ve got two refurbished Dell Precision 5820 towers, both bought cheap — the kind of cheap that happens when a company refreshes its workstations and the recycler just wants them gone. One came with an RTX 2080. The other, luckier, came with a Quadro RTX 5000.</p> <h2 id="the-good-5000">The good 5000<a class="hanchor" href="#the-good-5000" aria-label="Link to this section">#</a></h2> <p>That Quadro RTX 5000 had been quietly carrying me. Sixteen gigs of GDDR6, Turing silicon, the good stuff. I&rsquo;d been running <code>qwen3.6-27b</code> on it at a 3-bit quant — squeezed in tight — and it wasn&rsquo;t half bad. Decoding at 15–20 tokens a second, fast enough to keep me in the loop. I pointed it at some genuinely nasty bugs in a project I was wrestling with, and it earned its keep: it caught things I&rsquo;d stopped being able to see because I&rsquo;d been staring too long.</p>https://openmtx.com/posts/my-strix-halo/Sat, 27 Jun 2026 00:00:00 +0000https://openmtx.com/posts/my-strix-halo/<p>It started, like a few of my worst ideas, with an end-of-year email from HP.</p> <p>A Z2 Mini G1a — AMD&rsquo;s Strix Halo in a small quiet box — on promotion. The 128GB configuration landed around $2,000. I did the only math that mattered at the time: 128GB of LPDDR5X was roughly $800 on its own at the end of 2025, so the CPU, the NPU, and the whole rest of the machine were basically free. What did I know.</p>