f4e2264a83
Copied from upstream_ref (NOT rewritten — exact upstream code):
ixformer C++ API (the authoritative header):
include/ixformer.h — ixformer::infer namespace: topk_softmax,
moe_compute_token_index_api, moe_w16a16_group_gemm, moe_expand_input,
moe_output_reduce_sum, silu_and_mul, rms_norm, xllm_paged_attention, etc.
include/ilu_ops_api.h — xllm::kernel::ilu namespace: moe_active_topk,
moe_gen_idx, moe_expand_input, group_gemm, moe_combine_result,
batch_prefill, batch_decode, rms_norm, matmul, act_and_mul, etc.
ILU kernel wrappers (call ixformer::infer directly):
csrc/ilu_kernel_fused_moe.cpp — topk routing + gen_idx + expand + combine
csrc/ilu_kernel_group_gemm.cpp — batched expert GEMM
csrc/ilu_kernel_{activation,norm,rope,matmul,attention}.cpp
ILU layer implementations (full pipeline):
csrc/ilu_layer_fused_moe.{cpp,h} — 797 lines, the complete MoE pipeline
that competitor 168 ran as corex_moe.py
csrc/ilu_layer_attention.{cpp,h} — prefill/decode attention dispatch
CUDA MoE kernels (from xllm + ds_vllm):
csrc/moe/moe_topk_softmax_kernels.cuh — CUB BlockReduce + warp topk
csrc/moe/moe_topk_sigmoid_kernels.cuh — sigmoid scoring variant
csrc/moe/moe_topk.cuh + moe_fused_topk.cu — entry points
csrc/moe/moeTopKFuncs.cuh — TRT-LLM derived vllm-compatible topk
csrc/moe/moe_ops.h + moe_align_sum_kernels.cu — alignment kernels
Common layer headers:
csrc/common_fused_moe{,_base}.h + common_moe_fused_topk.{cpp,h}
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
|