.. meta:: :description: Build metadata buffer for all_to_all_v collective from MoE routing decisions. :date-modified: 05/21/2026 .. currentmodule:: nkilib.experimental.subkernels Build All To All V Metadata Kernel API Reference ================================================ Build metadata buffer for all_to_all_v collective from MoE routing decisions. Computes per-rank send counts and displacements from expert assignments. Background ----------- The ``build_all_to_all_v_metadata`` kernel builds the metadata buffer required by the all_to_all_v collective operation from MoE routing decisions, computing per-rank send counts and displacements from expert assignments. API Reference -------------- **Source code for this kernel API can be found at**: `build_all_to_all_v_metadata.py `_ build_all_to_all_v_metadata ^^^^^^^^^^^^^^^^^^^^^^^^^^^ .. py:function:: build_all_to_all_v_metadata(expert_index: nl.ndarray, replica_group_size: int, E: int, recv_counts_known: bool = False, has_rdispls: bool = False) Build metadata buffer for all_to_all_v collective from MoE routing decisions. :param expert_index: [T, K] int32 HBM tensor indicating the K experts each token is routed to. :type expert_index: ``nl.ndarray`` :param replica_group_size: Size of replica group for all_to_all_v collective. :type replica_group_size: ``int`` :param E: Number of global experts. :type E: ``int`` :param recv_counts_known: Not currently supported; when True, metadata includes recv counts. :type recv_counts_known: ``bool`` :param has_rdispls: Not currently supported; when True, metadata includes recv displacements. :type has_rdispls: ``bool`` :return: [n_rows, replica_group_size] uint32 HBM tensor. n_rows is 4 when has_rdispls=True, 3 otherwise. Row 0: send counts, Row 1: send displacements, Row 2: recv counts (zeros), Row 3 (optional): recv displacements (zeros). :rtype: ``nl.ndarray`` **Dimensions**: * T: Number of tokens. * K: Top-K experts per token.