6415249693
MoE call chain from ds_vllm (vllm-project/vllm latest): ex_engine/moe/ — 20 files, 8736 lines - modular_kernel.py (1630 lines) — base classes for modular MoE - experts/fused_batched_moe.py (972 lines) — NaiveBatchedExperts - prepare_finalize/batched.py (171 lines) — token grouping by expert - topk_weight_and_reduce.py (176 lines) — scatter-add finalize - fused_moe.py (1740 lines) — main fused_moe dispatch - config.py (1407 lines) — FusedMoEQuantConfig - activation.py, utils.py, layer.py, etc. xllm layer code (jd-opensource/xllm): ex_engine/xllm_layers/ — 39 files, 5859 lines - ilu/fused_moe.cpp (797 lines) — production ixformer 7-step MoE pipeline - ilu/attention.cpp (189 lines) — paged_attention + flash_attn bridge - npu_torch/qwen3_gated_delta_net_base.cpp (576 lines) — GDN reference - common/rms_norm.cpp, rotary_embedding.cpp, activation.cpp, dense_mlp.cpp xllm ILU kernels — synced 10 files to upstream (diffs from prior edits) These are reference implementations, NOT hand-written. Source repos: vllm-project/vllm, jd-opensource/xllm
63 lines
3.0 KiB
C++
63 lines
3.0 KiB
C++
/* Copyright 2025 The xLLM Authors. All Rights Reserved.
|
|
|
|
Licensed under the Apache License, Version 2.0 (the "License");
|
|
you may not use this file except in compliance with the License.
|
|
You may obtain a copy of the License at
|
|
|
|
https://github.com/jd-opensource/xllm/blob/main/LICENSE
|
|
|
|
Unless required by applicable law or agreed to in writing, software
|
|
distributed under the License is distributed on an "AS IS" BASIS,
|
|
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
See the License for the specific language governing permissions and
|
|
limitations under the License.
|
|
==============================================================================*/
|
|
#pragma once
|
|
namespace xllm::kernel::ilu {
|
|
#undef check_tensor_contiguous
|
|
#define check_tensor_contiguous(x, type) \
|
|
TORCH_CHECK(x.scalar_type() == type); \
|
|
TORCH_CHECK(x.is_cuda()); \
|
|
TORCH_CHECK(x.is_contiguous());
|
|
|
|
#undef check_tensor_half_bf_float
|
|
#define check_tensor_half_bf_float(x) \
|
|
TORCH_CHECK(x.scalar_type() == at::ScalarType::Half || \
|
|
x.scalar_type() == at::ScalarType::Float || \
|
|
x.scalar_type() == at::ScalarType::BFloat16); \
|
|
TORCH_CHECK(x.is_cuda());
|
|
|
|
// from torchCheckMsgImpl
|
|
inline const char* ixformer_check_msg_impl(const char* msg) { return msg; }
|
|
// // If there is just 1 user-provided C-string argument, use it.
|
|
|
|
#define IXFORMER_CHECK_MSG(cond, type, ...) \
|
|
(ixformer_check_msg_impl( \
|
|
"Expected " #cond \
|
|
" to be true, but got false. " \
|
|
"(Could this error message be improved? If so, " \
|
|
"please report an enhancement request to ixformer.)", \
|
|
##__VA_ARGS__))
|
|
|
|
#define IXFORMER_CHECK(cond, ...) \
|
|
{ \
|
|
if (!(cond)) { \
|
|
std::cerr << __FILE__ << " (" << __LINE__ << ")" \
|
|
<< "-" << __FUNCTION__ << " : " \
|
|
<< IXFORMER_CHECK_MSG(cond, "", ##__VA_ARGS__) << std::endl; \
|
|
throw std::runtime_error("IXFORMER_CHECK ERROR"); \
|
|
} \
|
|
}
|
|
|
|
#undef CUINFER_CHECK
|
|
#define CUINFER_CHECK(func) \
|
|
do { \
|
|
cuinferStatus_t status = (func); \
|
|
if (status != CUINFER_STATUS_SUCCESS) { \
|
|
std::cerr << "Error in file " << __FILE__ << " on line " << __LINE__ \
|
|
<< ": " << cuinferGetErrorString(status) << std::endl; \
|
|
throw std::runtime_error("CUINFER_CHECK ERROR"); \
|
|
} \
|
|
} while (0)
|
|
|
|
} // namespace xllm::kernel::ilu
|