{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiebis6idvpmf5fqraguzqujpzh5ug75gmwlhfmvz23wneorzapvou",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mlsyxxadown2"
  },
  "path": "/t/practical-match-for-128gb-strix-halo-with-2x3090s-inference-for-coding/175977#post_3",
  "publishedAt": "2026-05-14T13:35:28.000Z",
  "site": "https://discuss.huggingface.co",
  "textContent": "Thanks! I’ve read all that:) Just like I have read tons of text like yours on the subject this from diferent silicon minds, and they all do not have similar opinions.\n\nRight now my Halo is on windows/ollama, I guees I’ll go Ubuntu/Llama.cpp and try to get the best speeds and learn how to benchmark, than I’ll rent a server with 2x3090 and compare results.",
  "title": "Practical match for 128Gb Strix Halo with 2x3090s? (inference for coding)"
}