flashinfer.cake_gdn_nondet_train_kcp.cake_gdn_nondet_train_kcp_comm_bytes

flashinfer.cake_gdn_nondet_train_kcp.cake_gdn_nondet_train_kcp_comm_bytes(ctx: Any, *, H_qk: int, HV: int, C: int = 10240, save_transfer: bool = True, boundary_mode: str = 'saved') → dict

Bytes this rank moves per collective (c1, c2, b1, b2, halo_fwd, halo_bwd) and per boundary, counted per crossing fragment; C is the convolution channel count of the halo rows.

Parameters:
  • ctx – This rank’s context from cake_gdn_nondet_train_kcp_build_context().

  • H_qk (int) – Number of query and key heads.

  • HV (int) – Number of value heads.

  • C (int, default=10240) – Convolution channel count used to size halo traffic.

  • **kw – Backend options for this query: save_transfer and boundary_mode.