Paper runs distributed LLM inference over 10 km of multi-core fiber, no simulation
jwt0625 · x · 2026-08-31
jwt0625 shares Xu2026, "Convergence of LLM Parallelism and Spatial Parallelism: Distributed LLM Inference Over Multi-Core Fiber," which combines LLM parallelism with spatial parallelism to run inference across multi-core fiber links.
The commenter gives credit for actually running the experiment over 10 km of physical fiber instead of relying on simulations.
More from Infra
- Benchmarking Qwen3.8 BF16 across RTX 5090 and Mac via RPC — stargate425 · 2026-08-31
- Hugging Face used open weights to defend; frontier APIs failed — AlexTensor · 2026-08-31
- NCCL+MIG Support Arrives: Emulate Multi-Node 3D Parallelism on a Single GPU — StasBekman · 2026-08-31
- Inference Engineering Learning Path: From Basics to TensorRT-LLM — HowDevelop · 2026-08-31
- Hanshu Tech unveils uHBM and uLPU inference architecture — 新智元 · 2026-08-31
- Single Model Replaces Stack: 61% Cost Cut, Peak Accuracy — DynamicWebPaige · 2026-08-31