Gahow Wang
445e491123
third_party/vllm/ now tracked in git for direct patch management.
Based on vLLM v0.18.1 release with one patch applied:
vllm/v1/core/sched/scheduler.py:
Replace fatal assert with graceful skip when KV transfer callback
arrives for an already-aborted request during PD disaggregated serving.
Future vLLM modifications should be made directly in third_party/vllm/
and committed normally. The patches/ directory is kept as documentation
of what changed from upstream.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
19 lines
446 B
C++
19 lines
446 B
C++
#pragma once
|
|
|
|
#ifndef USE_ROCM
|
|
#include <cub/cub.cuh>
|
|
#if CUB_VERSION >= 200800
|
|
#include <cuda/std/functional>
|
|
using CubAddOp = cuda::std::plus<>;
|
|
using CubMaxOp = cuda::maximum<>;
|
|
#else // if CUB_VERSION < 200800
|
|
using CubAddOp = cub::Sum;
|
|
using CubMaxOp = cub::Max;
|
|
#endif // CUB_VERSION
|
|
#else
|
|
#include <hipcub/hipcub.hpp>
|
|
namespace cub = hipcub;
|
|
using CubAddOp = hipcub::Sum;
|
|
using CubMaxOp = hipcub::Max;
|
|
#endif // USE_ROCM
|