Tensor-parallel DeepSeek V4 Flash inference on dual AMD Strix Halo over OdinLink USB4/TB5 RDMA or Mellanox RoCE v2.
Like wkljohn/ds4-strix-halo-tp-odinlink, 0 likes 0 Bookmark wkljohn/ds4-strix-halo-tp-odinlink, 0 bookmarks 0 A deterministic inference engine for DeepSeek-V4.1-Flash 1017 and Qwen3.8-Flash-Next on a single AMD Strix Halo 128 GB
Like mrsirg97-rgb/libosmos, 0 likes 0 Bookmark mrsirg97-rgb/libosmos, 0 bookmarks 0 DeepSeek-V4.1-Flash-1017 EXL3 2.9bpw on 2x NVIDIA DGX Spark (GB10) with TensorFold: 2.1-2.3x the vLLM kit's decode speed.
Like Elevatormusic/DeepSeek-V4.1-Flash-1017-EXL3-TensorFold-2x-DGX-Spark, 0 likes 0 Bookmark Elevatormusic/DeepSeek-V4.1-Flash-1017-EXL3-TensorFold-2x-DGX-Spark, 0 bookmarks 0 llama.cpp modified to work with Strix Halo + dGPU as an accelerator, Dual Strix Halo Machines, or both
Like sixvolts/llama-halo-hybrid, 0 likes 0 Bookmark sixvolts/llama-halo-hybrid, 0 bookmarks 0 LLM inference engine written from scratch in HIP for AMD Strix Halo (gfx1151) and Qwen3.8-Flash-Next - ~50 tokens/s out to 492k context
Like shawnshekari/strixite, 0 likes 0 Bookmark shawnshekari/strixite, 0 bookmarks 0 Rust LLM inference engine tuned for AMD Strix Halo (gfx1151): a fork of Strata with ggml/llama.cpp-derived HIP kernels. Five models on one 128 GB machine.
Like nsbradley88/talus-public, 0 likes 0 Bookmark nsbradley88/talus-public, 0 bookmarks 0 DeepSeek-V4.1-Flash on 2x NVIDIA DGX Spark with the TensorFold engine: 1.8-1.9x code, 2.5x multi-stream, 1.7-1.95x prefill vs the vLLM kit, exact speculative decoding. Work in progress.
Like jayleaton/deepseek-v41-tensorfold-spark, 0 likes 0 Bookmark jayleaton/deepseek-v41-tensorfold-spark, 0 bookmarks 0 DeepSeek-V4.1-Flash-1017 EXL3 on 2x DGX Sparks with TensorFold
Like MiaAI-Lab/DeepSeek-v4.1-Flash-1017-2x-DGX-Sparks, 0 likes 0 Bookmark MiaAI-Lab/DeepSeek-v4.1-Flash-1017-2x-DGX-Sparks, 0 bookmarks 0 
