For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Mojo struct

Struct_ep_dispatch_async_mxfp4

struct Struct_ep_dispatch_async_mxfp4

Registers the ep.dispatch_async.mxfp4 graph op with the graph compiler.

Implemented traits

AnyType, Deinitable, Movable

Methods

execute

static def execute[input_dtype: DType, dispatch_dtype: DType, dispatch_scale_dtype: DType, hidden_size: Int, top_k: Int, n_experts: Int, max_token_per_rank: Int, n_gpus_per_node: Int, n_nodes: Int, //, target: StringSpan[ImmStaticOrigin], *, MX_FORMAT: StringSpan[ImmStaticOrigin] = StringSpan("auto")](atomic_counters: ManagedTensorSlice[IOSpec[_, _].MutableInput, static_spec=atomic_counters.static_spec], input_tokens: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input_tokens.static_spec], topk_ids: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=topk_ids.static_spec], send_ptrs: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=send_ptrs.static_spec], recv_ptrs: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=recv_ptrs.static_spec], recv_count_ptrs: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=recv_count_ptrs.static_spec], context: DeviceContext)

Execute the Expert Parallelism async dispatch kernel. Tokens are transferred in MXFP4 format with per-token even-mode scales packed alongside the FP4 quants in the send buffer.

Parameters:

• ​input_dtype (DType): DType of the input tokens before dispatch (inferred).
• ​dispatch_dtype (DType): DType used for the quantized token payload during dispatch (inferred).
• ​dispatch_scale_dtype (DType): DType of the per-token even-mode scales accompanying the dispatched tokens (inferred).
• ​hidden_size (Int): Size of the model's hidden dimension (inferred).
• ​top_k (Int): Number of experts each token is routed to (inferred).
• ​n_experts (Int): Total number of experts across all GPUs (inferred).
• ​max_token_per_rank (Int): Maximum number of tokens per GPU (inferred).
• ​n_gpus_per_node (Int): Number of GPUs per node (inferred).
• ​n_nodes (Int): Number of physical nodes (inferred).
• ​target (StringSpan[ImmStaticOrigin]): Compile-time device target.
• ​MX_FORMAT (StringSpan[ImmStaticOrigin]): Name of the MX element format, or auto to take it from the dtype. See MXFormat.from_name. Names the MX encoding of the wire payload. It cannot be inferred from dispatch_dtype: MXFP4 and MXFP6 both travel as DType.uint8 and differ only in bits per element. The default auto falls back to FP4 for a uint8 payload and FP8 E4M3 otherwise.

Args:

• ​atomic_counters (ManagedTensorSlice[IOSpec[_, _].MutableInput, static_spec=atomic_counters.static_spec]): Atomic counters coordinating work across thread blocks during the dispatch phase.
• ​input_tokens (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input_tokens.static_spec]): Input tokens to dispatch to experts. Shape [num_tokens, hidden_size].
• ​topk_ids (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=topk_ids.static_spec]): Top-k expert IDs per token. Shape [num_tokens, top_k].
• ​send_ptrs (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=send_ptrs.static_spec]): Send buffer pointers for the dispatch phase.
• ​recv_ptrs (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=recv_ptrs.static_spec]): Receive buffer pointers for the dispatch phase.
• ​recv_count_ptrs (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=recv_count_ptrs.static_spec]): Receive count buffer pointers tracking tokens received per expert.
• ​context (DeviceContext): GPU device context for the current device.