implementation detail · filed under inference & serving
Tensor parallelism (degree 4)
Uses tensor parallelism with degree four; the evidence describes serving configurations rather than a distinct algorithm.
Also called Four-way tensor parallelism for serving, tensor-parallel-size 4, Tensor parallelism with degree four, TP4.
- sources
- 2
- model
- 1
- lab adopt it
- 1
- strongest
- default
How sources treat it
One count per evidence span, weakest treatment to strongest.
optional 1default 1
Documented in
Evidence
2 spans quoted from the sources, strongest treatment first.
vllm serve Qwen/Qwen3.8-Flash-Next-FP8 \ --tensor-parallel-size 4
defaultinference servingin Qwen3.8-Flash-NextvLLM recipe authors
vllm serve Qwen/Qwen3.8-Flash-Next --port 8000 --tensor-parallel-size 4
optionalinference servingin Qwen3.8-Flash-NextQwen
Filed alongside
Other methods under inference & serving :: serving parallelism.
Attention data parallelismDeepEPPrefill-decode disaggregationEncoder-Prefill-Decode disaggregationTensor parallelism (degree 8)Topology-aware NVLink domain placementZero-copy fused token permutation and unpermutationData-parallel vision encodingExpert parallelismFused reduce-scatter/all-gather collectivesIdentical cache-layout pinning across prefill and decode poolsLow-precision MoE combineRound-robin routing for prefill-decode disaggregationTensor parallelism for MoE layersTensorRT-LLM all-reduce backendToken migration for balanced expert placement