# SPDX-License-Identifier: Apache-2.0 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project from abc import ABC, abstractmethod from collections.abc import Callable from dataclasses import dataclass from enum import Enum from math import prod from typing import final import torch import vllm.envs as envs from vllm.model_executor.layers.fused_moe.config import FusedMoEQuantConfig from vllm.model_executor.layers.fused_moe.utils import ( _resize_cache, count_expert_num_tokens, disable_inplace, ) from vllm.utils.math_utils import cdiv from vllm.v1.worker.ubatching import ( dbo_current_ubatch_id, dbo_enabled, dbo_maybe_run_recv_hook, dbo_register_recv_hook, dbo_yield, ) # # This file defines a set of base classes used to make MoE kernels more modular. # The goal is to be able to utilize different communication mechanisms with # any fused MoE kernel without needing to have combinatoric implementations. # # The fused moe kernels are broken down into the following components: # # [Router] → [Quantize-Dispatch] → [Permute-Experts-Unpermute] → [Combine] # # Each component will be independent of (but may inform) the others except for # [Quantize-Dispatch] and `[Combine] (see below). The components can then be # mixed and matched with so that DP+EP can be supported easily for multiple # MoE kernel implementations. # # The following main classes are defined: # * FusedMoEPrepareAndFinalize - an abstract base class for preparation of MoE # inputs (e.g. quantization, distribution) and finalization of Moe outputs. # The prepare method must take care of any needed quantization and the # finalize method, informed by the FusedMoEPermuteExpertsUnpermute method, # may apply weights and/or do the final reduction of the output. # * FusedMoEPermuteExpertsUnpermute - an abstract base class for the main fused # MoE operation, i.e matmul + act_mul + optionally quant + matmul. # Some FusedMoEPermuteExpertsUnpermute implementations may choose to do # the weight application and/or reduction. The class communicates this # to [Finalize] via a TopKWeightAndReduce object. # * FusedMoEModularKernel - an interface class that combines a # FusedMoEPrepareAndFinalize and a FusedMoEPermuteExpertsUnpermute to # provide the standard fused MoE kernel interface. # * TopKWeightAndReduce - A TopKWeightAndReduce implementation chosen # by the FusedMoEPermuteExpertsUnpermute implementation that is passed # on to [Finalize]. # # [Quantize-Prepare] and [Finalize] functionality are bundled into a single # class `FusedMoEPrepareAndFinalize` since they could use collective # communication mechanisms that need to be consistent. # class FusedMoEActivationFormat(Enum): """ The standard activation format (num_tokens, hidden dim). """ Standard = ("standard",) """ The batched experts format (num experts, max tokens per expert, hidden dim) """ BatchedExperts = ("batched_experts",) @dataclass class ExpertTokensMetadata: """ Metadata regarding expert-token routing. """ expert_num_tokens: torch.Tensor expert_num_tokens_cpu: torch.Tensor | None @staticmethod def make_from_list( expert_num_tokens_list: list[int], device: str ) -> "ExpertTokensMetadata": expert_num_tokens_cpu = torch.tensor( expert_num_tokens_list, device="cpu", dtype=torch.int32 ) return ExpertTokensMetadata( expert_num_tokens=expert_num_tokens_cpu.to(device, non_blocking=True), expert_num_tokens_cpu=expert_num_tokens_cpu, ) class TopKWeightAndReduce(ABC): """ An abstract base class for weight application and reduction implementations. """ @abstractmethod def apply( self, output: torch.Tensor | None, fused_expert_output: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, apply_router_weight_on_input: bool, ) -> torch.Tensor: """ Apply topk_weights to the fused_experts_outputs and/or reduce. If an output tensor is not passed, it will be created in the function. """ raise NotImplementedError # # PrepareResultType is a tuple of: # - quantized + dispatched a. # - quantized + dispatched a1_scales. # - Optional ExpertTokensMetadata containing gpu/cpu tensors # as big as the number of local experts with the information about the # number of tokens assigned to each local expert. # - Optional dispatched expert topk IDs # - Optional dispatched expert topk weight # # See `prepare` method below. # PrepareResultType = tuple[ torch.Tensor, torch.Tensor | None, ExpertTokensMetadata | None, torch.Tensor | None, torch.Tensor | None, ] ReceiverType = Callable[[], PrepareResultType] # TODO: pass FusedMoEParallelConfig in as ctor parameter? class FusedMoEPrepareAndFinalize(ABC): """ An abstract base class for the [Quantize-Prepare] and [Finalize] steps described above. """ def post_init_setup(self, fused_experts: "FusedMoEPermuteExpertsUnpermute"): """ Initialize FusedMoEPrepareAndFinalize settings that depend on FusedMoEPermuteExpertsUnpermute experts object. The FusedMoEPrepareAndFinalize implementations that have such dependencies may choose to override this function. """ return @abstractmethod def prepare( self, a1: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, num_experts: int, expert_map: torch.Tensor | None, apply_router_weight_on_input: bool, quant_config: FusedMoEQuantConfig, ) -> PrepareResultType: """ Perform any quantization (and/or) dispatching needed for this kernel. - a1: The (unquantized) input to the MoE layer. - topk_ids: The topk ids. - topk_weights: The topk weights. - num_experts: The total number of experts in the global expert space. - expert_map: A tensor mapping expert indices from the global expert space to the local expert space of the expert parallel shard. - apply_router_weight_on_input: When True, apply the weights to the activations, before quantization + dispatching. - quant_config: Quantization info provided by the fused experts. Returns a tuple of: - quantized + dispatched a. - Optional quantized + dispatched a1_scales. - Optional ExpertTokensMetadata containing gpu/cpu tensors as big as the number of local experts with the information about the number of tokens assigned to each local expert. - Optional dispatched expert topk IDs - Optional dispatched expert topk weight """ raise NotImplementedError def supports_async(self) -> bool: """ Indicates whether or not this class implements prepare_async and finalize_async. """ return False def prepare_async( self, a1: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, num_experts: int, expert_map: torch.Tensor | None, apply_router_weight_on_input: bool, quant_config: FusedMoEQuantConfig, ) -> tuple[Callable, ReceiverType] | ReceiverType: """ Perform any quantization (and/or) dispatching needed for this kernel but do not wait for results from other workers. - a1: The (unquantized) input to the MoE layer. - a1_scale: Optional scales for a1 - a2_scale: Optional scales for the second MoE gemm. Required to make sure the quantization is consistent for both gemms. - topk_ids: The topk ids. - topk_weights: The topk weights. - num_experts: The total number of experts in the global expert space. - expert_map: A tensor mapping expert indices from the global expert space to the local expert space of the expert parallel shard. - apply_router_weight_on_input: When True, apply the weights to the activations, before quantization + dispatching. Returns a callback or a hook callback pair that when invoked waits for results from other workers and has the same return signature as `prepare`, if a hook is returned this is more lightweight check that the recv is complete without doing extra work (used by DBO, will be refactored in the very near future) e.g. ret = obj.prepare_async(...) if isinstance(ret, tuple): hook, receiver = ret hook() if hook is not None: a, a_scales, expert_meta, topk_ids, topk_weights = receiver() is equivalent to: a, a_scales, expert_meta, topk_ids, topk_weights = obj.prepare(...) """ raise NotImplementedError @abstractmethod def finalize( self, output: torch.Tensor, fused_expert_output: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, apply_router_weight_on_input: bool, weight_and_reduce_impl: TopKWeightAndReduce, ) -> None: """ Perform any combine plus apply weights and perform a reduction on the fused experts output. - output: The output tensor, written in place. Must be (M, K) shape. - fused_expert_output: The unweighted, unreduced output of the fused experts, it will have (M, topk, K) shape. - topk_weights: The weights to be applied to the fused_experts_output. - topk_ids: The topk_ids. - apply_router_weight_on_input: When False, apply the weights to fused_expert_output. - weight_and_reduce_impl: An optional TopKWeightAndReduce implementation. """ raise NotImplementedError def finalize_async( self, output: torch.Tensor, fused_expert_output: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, apply_router_weight_on_input: bool, weight_and_reduce_impl: TopKWeightAndReduce, ) -> tuple[Callable, Callable] | Callable: """ Perform any combine plus apply weights and perform a reduction on the fused experts output but do not wait for results from other workers. - output: The output tensor, written in place. Must be (M, K) shape. - fused_expert_output: The unweighted, unreduced output of the fused experts, it will have (M, topk, K) shape. - topk_weights: The weights to be applied to the fused_experts_output. - topk_ids: The topk_ids. - apply_router_weight_on_input: When False, apply the weights to fused_expert_output. - weight_and_reduce_impl: An optional TopKWeightAndReduce implementation. Returns a callback or a hook callback pair that when invoked waits for results from other workers and has the same return signature as `finalize`, if a hook is returned this is more lightweight check that the recv is complete without doing extra work (used by DBO, will be refactored in the very near future) ret = obj.finalize_async(output, ...) ... output not valid yet ... if isinstance(ret, tuple): hook, receiver = ret hook() receiver() ... output valid here ... is equivalent to: obj.finalize(output, ...) """ raise NotImplementedError @property @abstractmethod def activation_format(self) -> FusedMoEActivationFormat: """ A property indicating the output format of the activations for the 'prepare' method. """ raise NotImplementedError @abstractmethod def topk_indices_dtype(self) -> torch.dtype | None: """ The PrepareFinalize All2All implementations generally constrain the dtype of the topk_ids they support. This function returns the required topk indices dtype so it can be respected. Return None if there are no such restrictions. """ raise NotImplementedError @abstractmethod def max_num_tokens_per_rank(self) -> int | None: """ Some PrepareFinalize All2All implementations are batched. Meaning, they can process only as set of tokens at a time. This function returns the batch size i.e the maximum number of tokens the implementation can process at a time. Return None if there are no such restrictions. """ raise NotImplementedError @abstractmethod def num_dispatchers(self) -> int: raise NotImplementedError @abstractmethod def output_is_reduced(self) -> bool: """ Indicates whether or not the output of finalize is reduced across all ranks. """ raise NotImplementedError # TODO: add supported activations method (return string) class FusedMoEPermuteExpertsUnpermute(ABC): """ An abstract base class for the [Permute-Experts-Unpermute] step described above. """ def __init__( self, quant_config: FusedMoEQuantConfig, ): """ quant_config: Quantization parameters for this experts instance. """ self.quant_config = quant_config @property @abstractmethod def activation_formats( self, ) -> tuple[FusedMoEActivationFormat, FusedMoEActivationFormat]: """ A property which is a tuple of the input and output activation formats for the 'apply' method. """ raise NotImplementedError def moe_problem_size( self, a1: torch.Tensor, w1: torch.Tensor, w2: torch.Tensor, topk_ids: torch.Tensor, ) -> tuple[int, int, int, int, int]: """ Extract the MoE problem size from the given tensor arguments: - a: The hidden states, input to the MoE layer. - w1: The first set of expert weights. - w2: The second set of expert weights. - topk_ids: The topk ids. Note: extracting the problem shape from the weight and activation tensors is not obvious. It needs to be done this way specifically due to subtle issues with particular kernels, e.g. the int4 kernels divide the trailing dimension by two, so it's not "correct" to extract N or K from the trailing dimension of w1 or w2. Similarly, some kernels transpose the weights, so this needs to be kept in mind. Note: This implementation covers most cases. However, if experts require a specialized implementation, like MarlinExperts, they are free to override this function. """ assert w1.dim() == 3 and w2.dim() == 3 E, N, _ = w1.size() K = a1.size(-1) if a1.dim() == 2: # Make sure we are using the correct a1 (pre-permute). assert topk_ids.size(0) == a1.size(0), f"{topk_ids.size(0)} != {a1.size(0)}" M = a1.size(0) else: assert a1.dim() == 3 assert a1.size(0) == E, f"{a1.size(0)} == {E}" M = a1.size(1) # This is max_num_tokens assert topk_ids.dim() == 2 topk = topk_ids.size(1) return E, M, N, K, topk # # Various helpers for accessing quantization parameters from the # quant_config. # @property def quant_dtype(self) -> torch.dtype | None: return self.quant_config.quant_dtype @property def block_shape(self) -> list[int] | None: return self.quant_config.block_shape @property def per_act_token_quant(self) -> bool: return self.quant_config.per_act_token_quant @property def per_out_ch_quant(self) -> bool: return self.quant_config.per_out_ch_quant @property def a1_scale(self) -> torch.Tensor | None: return self.quant_config.a1_scale @property def a2_scale(self) -> torch.Tensor | None: return self.quant_config.a2_scale @property def a1_gscale(self) -> torch.Tensor | None: return self.quant_config.a1_gscale @property def a2_gscale(self) -> torch.Tensor | None: return self.quant_config.a2_gscale @property def w1_scale(self) -> torch.Tensor | None: return self.quant_config.w1_scale @property def w2_scale(self) -> torch.Tensor | None: return self.quant_config.w2_scale @property def w1_zp(self) -> torch.Tensor | None: return self.quant_config.w1_zp @property def w2_zp(self) -> torch.Tensor | None: return self.quant_config.w2_zp @property def w1_bias(self) -> torch.Tensor | None: return self.quant_config.w1_bias @property def w2_bias(self) -> torch.Tensor | None: return self.quant_config.w2_bias @property def g1_alphas(self) -> torch.Tensor | None: return self.quant_config.g1_alphas @property def g2_alphas(self) -> torch.Tensor | None: return self.quant_config.g2_alphas # TODO (bnell): make this return a CHUNK_SIZE or None instead? @abstractmethod def supports_chunking(self) -> bool: """ A flag indicating whether or not this class supports activation chunking. """ raise NotImplementedError @abstractmethod def supports_expert_map(self) -> bool: """ A flag indicating whether or not this class supports expert maps """ raise NotImplementedError def supports_packed_ue8m0_act_scales(self) -> bool: """ A flag indicating whether or not this class can process packed ue8m0 activation scales. """ return False def workspace_dtype(self, act_dtype: torch.dtype) -> torch.dtype: """ Workspace type: The dtype to use for the workspace tensors. """ return act_dtype @abstractmethod def workspace_shapes( self, M: int, N: int, K: int, topk: int, global_num_experts: int, local_num_experts: int, expert_tokens_meta: ExpertTokensMetadata | None, ) -> tuple[tuple[int, ...], tuple[int, ...], tuple[int, ...]]: """ Compute the shapes for the temporary and final outputs of the two gemms and activation in the fused expert function. Since the gemms are independent, the workspace for the first gemm can be shared with the workspace for the last gemm. Inputs: - M: number of tokens. - N: Row (or column) dimension of expert weights. - K: hidden dimension - topk: The number of top-k experts to select. - global_num_experts: global number of experts. - local_num_experts: local number of experts due to DP/EP. - expert_tokens_meta: number of tokens per expert metadata for batched format. Returns a tuple of: - workspace13 shape tuple: must be large enough to hold the result of either expert gemm. - workspace2 shape tuple: must be large enough to hold the result of the activation function. - output shape tuple: must be exact size of the final gemm output. - Note: workspace shapes can be 0 if the workspace is not needed. But in order for activation chunking to work, the first dimension of each tuple must be the number of tokens when the shape is not 0. """ raise NotImplementedError def activation( self, activation: str, output: torch.Tensor, input: torch.Tensor ) -> None: assert output.size(-1) * 2 == input.size(-1) if activation == "silu": torch.ops._C.silu_and_mul(output, input) elif activation == "gelu": torch.ops._C.gelu_and_mul(output, input) elif activation == "swigluoai": # alpha = 1.702, limit = 7.0 torch.ops._C.swigluoai_and_mul(output, input) else: raise ValueError(f"Unsupported FusedMoe activation: {activation}") def enable_chunking(self): return ( envs.VLLM_ENABLE_FUSED_MOE_ACTIVATION_CHUNKING and self.supports_chunking() ) def finalize_weight_and_reduce_impl(self) -> TopKWeightAndReduce: raise NotImplementedError @abstractmethod def apply( self, output: torch.Tensor, hidden_states: torch.Tensor, w1: torch.Tensor, w2: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, activation: str, global_num_experts: int, expert_map: torch.Tensor | None, a1q_scale: torch.Tensor | None, a2_scale: torch.Tensor | None, workspace13: torch.Tensor, workspace2: torch.Tensor, expert_tokens_meta: ExpertTokensMetadata | None, apply_router_weight_on_input: bool, ) -> None: """ This function computes the intermediate result of a Mixture of Experts (MoE) layer using two sets of weights, w1 and w2. Parameters: - output: (torch.Tensor): The unweighted, unreduced output tensor. - hidden_states: (torch.Tensor): The (quantized) input tensor to the MoE layer. - w1 (torch.Tensor): The first set of expert weights. - w2 (torch.Tensor): The second set of expert weights. - topk_weights: A map of row to expert weights. Some implementations choose to do weight application. - topk_ids (torch.Tensor): A map of row to expert id. - activation (str): The activation function to apply after the first MoE layer. - global_num_experts (int): The total number of experts in the global expert space. - expert_map (Optional[torch.Tensor]): A tensor mapping expert indices from the global expert space to the local expert space of the expert parallel shard. - a1q_scale (Optional[torch.Tensor]): Optional quantized scale to be used for a1. Result of quantization from prepare/finalize and not from the FusedMoEQuantConfig. - workspace13 (torch.Tensor): A scratch tensor used for gemm outputs must be large enough to hold output of either MoE gemm. - workspace2 (torch.Tensor): A scratch tensor used for the activation function. - expert_tokens_meta (Optional[ExpertTokensMetadata]) - An optional ExpertTokensMetadata object containing gpu/cpu tensors as big as the number of local experts with the information about the number of tokens assigned to each local expert. - apply_router_weight_on_input: True if router weights are already applied on the input. This is relevant if the implementation chooses to do weight application. """ raise NotImplementedError def _slice_scales( scales: torch.Tensor | None, start: int, end: int ) -> torch.Tensor | None: if scales is not None: if scales.numel() == 1: return scales else: return scales[start:end] return None class SharedResizableBuffer: def __init__(self): self.buffer = None def get( self, shape: tuple[int, ...], device: torch.device, dtype: torch.dtype ) -> torch.Tensor: assert shape != () shape_numel = prod(shape) if ( self.buffer is None or self.buffer.numel() < shape_numel or self.buffer.device != device or self.buffer.dtype != dtype ): self.buffer = torch.empty(shape_numel, device=device, dtype=dtype) return self.buffer[:shape_numel].view(*shape) @final class FusedMoEModularKernel(torch.nn.Module): """ This class combines a FusedMoEPrepareAndFinalize instance and a FusedMoEPermuteExpertsUnpermute to provide an interface that is compatible with the `fused_experts` function in fused_moe.py. It takes care of managing any required scratch space. Note: Instances of this class should only be used for a single model layer due to any layer specific state that may be used by the component objects. """ class SharedBuffers: def __init__(self) -> None: self.fused_out = SharedResizableBuffer() self.workspace13 = SharedResizableBuffer() self.workspace2 = SharedResizableBuffer() # Persistent buffers that are shared across `FusedMoEModularKernel` # instances (layers), to save memory and allocattions. # # We have two sets of buffers to support dual batch overlap (DBO) where each # microbatch (ubatch) should use its own set of buffers to avoid # cross-ubatch contimination. # NOTE that memory is lazily allocated for these buffers, meaning that if # DBO isn't being used, the second SharedBuffers will be empty. shared_buffers: list[SharedBuffers] = [SharedBuffers(), SharedBuffers()] def __init__( self, prepare_finalize: FusedMoEPrepareAndFinalize, fused_experts: FusedMoEPermuteExpertsUnpermute, shared_experts: torch.nn.Module | None = None, ): super().__init__() self.prepare_finalize = prepare_finalize self.fused_experts = fused_experts self.shared_experts = shared_experts self._post_init_setup() assert ( prepare_finalize.activation_format == fused_experts.activation_formats[0] ), ( f"{prepare_finalize.__class__.__name__}." f"{prepare_finalize.activation_format} == " f"{fused_experts.__class__.__name__}." f"{fused_experts.activation_formats[0]}" ) def _post_init_setup(self): """ Resolve any leftover setup dependencies between self.prepare_finalize and self.fused_experts here. """ self.prepare_finalize.post_init_setup(self.fused_experts) def supports_expert_map(self) -> bool: """ A flag indicating whether or not this class supports expert maps. """ return self.fused_experts.supports_expert_map() def output_is_reduced(self) -> bool: """ Indicates whether or not the output of fused MoE kernel is reduced across all ranks. """ return self.prepare_finalize.output_is_reduced() def _chunk_info(self, M: int) -> tuple[int, int]: """ Compute number of chunks and chunk size for given M. If chunking is not supported, set the CHUNK_SIZE to M so we get num_chunks == 1. Take max(M, 1) to avoid divide by zero. If there are no tokens to process, the number of chunks will be zero. """ CHUNK_SIZE = max( 1, ( M if not self.fused_experts.supports_chunking() else min(M, envs.VLLM_FUSED_MOE_CHUNK_SIZE) ), ) num_chunks = cdiv(M, CHUNK_SIZE) # If there are no tokens, then there should be no loop iterations. assert M > 0 or num_chunks == 0 return num_chunks, CHUNK_SIZE def _allocate_buffers( self, out_dtype: torch.dtype, device: torch.device, M_chunk: int, M_full: int, N: int, K: int, top_k: int, global_num_experts: int, local_num_experts: int, expert_tokens_meta: ExpertTokensMetadata | None, ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """ Allocate temporary and output buffers for the fused experts op. Inputs: - out_dtype: output type of workspace and output tensors. - device: the device of the workspace and output tensors. See `workspace_shapes` for a description of the remainder of arguments. Returns a tuple of (workspace13, workspace2, output) tensors. """ assert M_full > 0 and M_chunk > 0 num_chunks, _ = self._chunk_info(M_full) # select per-ubatch buffers to avoid cross-ubatch reuse under DBO ubatch_idx = dbo_current_ubatch_id() buffers = self.shared_buffers[ubatch_idx] workspace_dtype = self.fused_experts.workspace_dtype(out_dtype) # Get intermediate workspace shapes based off the chunked M size. workspace13_shape, workspace2_shape, _ = self.fused_experts.workspace_shapes( M_chunk, N, K, top_k, global_num_experts, local_num_experts, expert_tokens_meta, ) # Get final output shape based on the full M size. _, _, fused_out_shape = self.fused_experts.workspace_shapes( M_full, N, K, top_k, global_num_experts, local_num_experts, expert_tokens_meta, ) # We can reuse the memory between cache1 and cache3 because by the # time we need cache3, we're done with cache1. workspace13 = buffers.workspace13.get( workspace13_shape, device=device, dtype=workspace_dtype ) workspace2 = buffers.workspace2.get( workspace2_shape, device=device, dtype=workspace_dtype ) # Construct the entire output that can then be processed in chunks. # Reuse workspace13 for the output in the non-chunked case as long # as it is large enough. This will not always be the case for standard # format experts and with experts that have empty workspaces. if num_chunks == 1 and prod(workspace13_shape) >= prod(fused_out_shape): fused_out = _resize_cache(workspace13, fused_out_shape) else: fused_out = buffers.fused_out.get( fused_out_shape, device=device, dtype=out_dtype ) return workspace13, workspace2, fused_out @staticmethod def _slice_output_tensor( fused_out: torch.Tensor, chunk_idx: int, num_chunks: int, CHUNK_SIZE: int, M: int, ) -> torch.Tensor: if num_chunks == 1: return fused_out assert fused_out.size(0) % M == 0, f"fused_out shape {fused_out.shape} vs M {M}" factor = fused_out.size(0) // M out_chunk_size = CHUNK_SIZE * factor s = chunk_idx * out_chunk_size e = min(s + out_chunk_size, fused_out.size(0)) return fused_out[s:e] @staticmethod def _slice_expert_tokens_metadata( num_chunks: int, full_expert_tokens_meta: ExpertTokensMetadata | None, chunk_topk_ids: torch.Tensor, local_num_experts: int, expert_map: torch.Tensor | None, ) -> ExpertTokensMetadata | None: if num_chunks == 1 or full_expert_tokens_meta is None: return full_expert_tokens_meta # The existing expert_num_tokens is for the entire a1q # input. Chunking forces recomputation of the number # of tokens assigned to each expert. c_expert_num_tokens = count_expert_num_tokens( chunk_topk_ids, local_num_experts, expert_map ) c_expert_num_tokens_cpu = None need_expert_num_tokens_cpu = ( full_expert_tokens_meta.expert_num_tokens_cpu is not None ) if need_expert_num_tokens_cpu: # This is blocking as some implementations need the count # on the CPU to determine appropriate input/out fused-moe # buffers c_expert_num_tokens_cpu = c_expert_num_tokens.to("cpu", non_blocking=False) return ExpertTokensMetadata( expert_num_tokens=c_expert_num_tokens, expert_num_tokens_cpu=c_expert_num_tokens_cpu, ) def _prepare( self, hidden_states: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, global_num_experts: int, expert_map: torch.Tensor | None, apply_router_weight_on_input: bool, ) -> tuple[ torch.Tensor, torch.Tensor | None, ExpertTokensMetadata | None, torch.Tensor, torch.Tensor, ]: """ The _prepare method is a wrapper around self.prepare_finalize.prepare that handles DBO and async. """ if not self.prepare_finalize.supports_async(): # We shouldn't be running an a2a kernel that doesn't # support async prepare/finalize # TODO(lucas): enable in follow-up assert not dbo_enabled() ( a1q, a1q_scale, expert_tokens_meta, _expert_topk_ids, _expert_topk_weights, ) = self.prepare_finalize.prepare( hidden_states, topk_weights, topk_ids, global_num_experts, expert_map, apply_router_weight_on_input, self.fused_experts.quant_config, ) else: # Overlap shared expert compute with all2all dispatch. dbo_maybe_run_recv_hook() prepare_ret = self.prepare_finalize.prepare_async( hidden_states, topk_weights, topk_ids, global_num_experts, expert_map, apply_router_weight_on_input, self.fused_experts.quant_config, ) # TODO(lucas): refactor this in the alternative schedules followup # currently unpack if we have hook + receiver pair or just # receiver (see finalize_async docstring) hook, receiver = ( prepare_ret if isinstance(prepare_ret, tuple) else (None, prepare_ret) ) if hook is not None: if dbo_enabled(): # If DBO is being used, register the hook with the ubatch # context and call it in dbo_maybe_run_recv_hook instead of # passing it to the receiver. dbo_register_recv_hook(hook) dbo_yield() else: hook() ( a1q, a1q_scale, expert_tokens_meta, _expert_topk_ids, _expert_topk_weights, ) = receiver() # Maybe prepare gathered topk_ids and topk_weights from other EP ranks. topk_ids = topk_ids if _expert_topk_ids is None else _expert_topk_ids topk_weights = ( topk_weights if _expert_topk_weights is None else _expert_topk_weights ) return a1q, a1q_scale, expert_tokens_meta, topk_ids, topk_weights def _fused_experts( self, in_dtype: torch.dtype, a1q: torch.Tensor, a1q_scale: torch.Tensor | None, w1: torch.Tensor, w2: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, activation: str, global_num_experts: int, local_num_experts: int, expert_map: torch.Tensor | None, apply_router_weight_on_input: bool, expert_tokens_meta: ExpertTokensMetadata | None, ) -> torch.Tensor: _, M_full, N, K, top_k = self.fused_experts.moe_problem_size( a1q, w1, w2, topk_ids ) num_chunks, CHUNK_SIZE = self._chunk_info(M_full) def input_chunk_range(chunk_idx: int) -> tuple[int, int]: if num_chunks == 1: # Use a1q.size(0) here since batched format does not # keep M in the first dimension. return 0, a1q.size(0) else: s = chunk_idx * CHUNK_SIZE e = min(s + CHUNK_SIZE, M_full) return s, e # This happens when none of the tokens from the all2all reach this # EP rank. Also, note that this is only relevant for CUDAGraph # incompatible all2all kernels like the DeepEP high-throughput # kernels. CUDAGraph compatible all2all kernels like the pplx # kernels and the DeepEP low-latency kernels are always batched # and can never run into the tensor.numel() == 0 case. if M_full == 0: assert num_chunks == 0 workspace13 = None workspace2 = None fused_out = torch.empty_like(a1q, dtype=in_dtype) else: assert num_chunks > 0 workspace13, workspace2, fused_out = self._allocate_buffers( in_dtype, a1q.device, CHUNK_SIZE, M_full, N, K, top_k, global_num_experts, local_num_experts, expert_tokens_meta, ) for chunk_idx in range(num_chunks): s, e = input_chunk_range(chunk_idx) c_expert_tokens_meta = self._slice_expert_tokens_metadata( num_chunks, expert_tokens_meta, topk_ids[s:e], local_num_experts, expert_map, ) c_fused_out = self._slice_output_tensor( fused_out, chunk_idx, num_chunks, CHUNK_SIZE, M_full ) self.fused_experts.apply( output=c_fused_out, hidden_states=a1q[s:e], w1=w1, w2=w2, topk_weights=topk_weights[s:e], topk_ids=topk_ids[s:e], activation=activation, global_num_experts=global_num_experts, expert_map=expert_map, a1q_scale=_slice_scales(a1q_scale, s, e), a2_scale=_slice_scales(self.fused_experts.a2_scale, s, e), workspace13=workspace13, workspace2=workspace2, expert_tokens_meta=c_expert_tokens_meta, apply_router_weight_on_input=apply_router_weight_on_input, ) return fused_out def _finalize( self, output: torch.Tensor, fused_out: torch.Tensor, hidden_states: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, apply_router_weight_on_input: bool, ) -> torch.Tensor | tuple[torch.Tensor, torch.Tensor]: """ The _finalize method is a wrapper around self.prepare_finalize.finalize that handles DBO, async and shared expert overlap. """ shared_output: torch.Tensor | None = None if not self.prepare_finalize.supports_async(): assert not dbo_enabled() self.prepare_finalize.finalize( output, fused_out, topk_weights, topk_ids, apply_router_weight_on_input, self.fused_experts.finalize_weight_and_reduce_impl(), ) if self.shared_experts is not None: shared_output = self.shared_experts(hidden_states) else: finalize_ret = self.prepare_finalize.finalize_async( output, fused_out, topk_weights, topk_ids, apply_router_weight_on_input, self.fused_experts.finalize_weight_and_reduce_impl(), ) if self.shared_experts is not None: shared_output = self.shared_experts(hidden_states) # TODO(lucas): refactor this in the alternative schedules followup # currently unpack if we have hook + receiver pair or just # receiver (see finalize_async docstring) hook, receiver = ( finalize_ret if isinstance(finalize_ret, tuple) else (None, finalize_ret) ) if hook is not None: if dbo_enabled(): # If DBO is being used, register the hook with the ubatch # context and call it in dbo_maybe_run_recv_hook instead of # passing it to the receiver. dbo_register_recv_hook(hook) dbo_yield() else: hook() receiver() if self.shared_experts is None: return output else: assert shared_output is not None return shared_output, output def forward( self, hidden_states: torch.Tensor, w1: torch.Tensor, w2: torch.Tensor, topk_weights: torch.Tensor, topk_ids: torch.Tensor, inplace: bool = False, activation: str = "silu", global_num_experts: int = -1, expert_map: torch.Tensor | None = None, apply_router_weight_on_input: bool = False, ) -> torch.Tensor | tuple[torch.Tensor, torch.Tensor]: """ This function computes a Mixture of Experts (MoE) layer using two sets of weights, w1 and w2, and top-k gating mechanism. Parameters: - hidden_states: (torch.Tensor): The input tensor to the MoE layer. - w1 (torch.Tensor): The first set of expert weights. - w2 (torch.Tensor): The second set of expert weights. - topk_weights (torch.Tensor): The topk weights applied at the end of the layer. - topk_ids (torch.Tensor): A map of row to expert id. - inplace (bool): If True, perform the operation in-place. Defaults to False. - activation (str): The activation function to apply after the first MoE layer. - global_num_experts (int): The total number of experts in the global expert space. - expert_map (Optional[torch.Tensor]): A tensor mapping expert indices from the global expert space to the local expert space of the expert parallel shard. - apply_router_weight_on_input (bool): When true, the topk weights are applied directly on the inputs. This is only applicable when topk is 1. Returns: - torch.Tensor: The output tensor after applying the MoE layer. """ if inplace and self.shared_experts is None and not disable_inplace(): output = hidden_states else: output = torch.zeros_like(hidden_states) local_num_experts = w1.size(0) if global_num_experts == -1: global_num_experts = local_num_experts a1q, a1q_scale, expert_tokens_meta, topk_ids, topk_weights = self._prepare( hidden_states, topk_weights, topk_ids, global_num_experts, expert_map, apply_router_weight_on_input, ) fused_out = self._fused_experts( in_dtype=hidden_states.dtype, a1q=a1q, a1q_scale=a1q_scale, w1=w1, w2=w2, topk_weights=topk_weights, topk_ids=topk_ids, activation=activation, global_num_experts=global_num_experts, local_num_experts=local_num_experts, expert_map=expert_map, apply_router_weight_on_input=apply_router_weight_on_input, expert_tokens_meta=expert_tokens_meta, ) return self._finalize( output, fused_out, hidden_states, topk_weights, topk_ids, apply_router_weight_on_input, )