Why it matters
Tensor parallel is key for large layer models. Understanding shapes intra-node parallelism.
Advertisement
The architecture
Column parallel: split matmul weights by column. All-gather output.
Row parallel: split by row. All-reduce output.
Advertisement
How it works end to end
Combined pattern in transformer: column parallel Q, K, V projections + row parallel output projection.
Communication: all-reduce after each block. NVLink bandwidth critical.
Within node only: cross-node too slow.