{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidpuvehiqzjizatasfpa6g6fun3i75oyycstgm56mft3c6orfrmti",
    "uri": "at://did:plc:hqad6xwuzg7oqfmwylfkvqfm/app.bsky.feed.post/3mnkg4ttmdo42"
  },
  "path": "/viewtopic.php?t=32780&p=275247#p275247",
  "publishedAt": "2026-06-05T13:43:05.000Z",
  "site": "http://forum.palemoon.org",
  "textContent": "> Well, the issue is kind of that Dav1d doesn't actually work on SSE2-only CPUs, at least not if it's compiled on a machine that has AVX instructions available. The issue seems to be that it emits AVX instructions on any compiler/machine combo that can, regardless of what you want. Hypothetically, it might work if you compile it on a machine with a CPU that can't support AVX, because that would force the compiler to generate fallback instructions or something, but I have no idea.\n\nWorks fine for me on Linux on a CPU with no AVX. Binary was compiled on a CPU that did have AVX (not that that would even matter). dav1d has code to select which SIMD instructions to use at runtime. It might be an issue with how we've hooked it into our tree, but dav1d itself absolutely does not require AVX.\n\n* * *",
  "title": "Contributed 3rd Party Builds • Re: Pale Moon for Slackware Linux",
  "updatedAt": "2026-06-05T13:43:05.000Z"
}