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_wait_fp8
struct Struct_ep_dispatch_wait_fp8
Registers the ep.dispatch_wait.fp8 graph op with the graph compiler.
Implemented traitsβ
Methodsβ
executeβ
static def execute[hidden_size: Int, top_k: Int, n_experts: Int, max_token_per_rank: Int, n_gpus_per_node: Int, n_nodes: Int, dispatch_scale_granularity: StaticString, //, num_input_tokens: Int = Int(-1)](output_tokens: T, output_scales: T, row_offsets: T, expert_ids: T, src_info: T, atomic_counters: T, recv_ptrs: T, recv_count_ptrs: T, context: DeviceContext)
Execute the Expert Parallelism dispatch completion kernel. Received tokens are in Blockwise FP8 format.
runβ
static def run[hidden_size: Int, top_k: Int, n_experts: Int, max_token_per_rank: Int, n_gpus_per_node: Int, n_nodes: Int, dispatch_scale_granularity: StaticString, num_input_tokens: Int = Int(-1), ExpertLoadT: ExpertLoadSink = NullExpertLoad](output_tokens: T, output_scales: T, row_offsets: T, expert_ids: T, src_info: T, atomic_counters: T, recv_ptrs: T, recv_count_ptrs: T, context: DeviceContext, expert_load: ExpertLoadT)
Runs ep.dispatch_wait.fp8, reporting each local expert's routed-token count to expert_load.