Comment on Show HN: Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/sparentComments−rzzzt11dOptane? (Too soon?)−0x45711dOptane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading.
Comments
Optane? (Too soon?)
Optane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading.