hip_redop.h

Go to the documentation of this file.

/*
 * Copyright 2025 Stanford University, NVIDIA Corporation
 * SPDX-License-Identifier: Apache-2.0
 *
 * 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
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * 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.
 */
 
#ifndef REALM_HIP_REDOP_H
#define REALM_HIP_REDOP_H
 
#include "realm/realm_config.h"
 
#if defined(__CUDACC__) || defined(__HIPCC__)
#ifdef REALM_USE_HIP
#include <hip/hip_runtime.h>
#endif
#endif
 
namespace Realm {
 
  namespace Hip {
 
#if defined(__CUDACC__) || defined(__HIPCC__)
    // the ability to add CUDA kernels to a reduction op is only available
    //  when using a compiler that understands CUDA
    namespace ReductionKernels {
      template <typename REDOP, bool EXCL>
      __global__ void apply_hip_kernel(uintptr_t lhs_base, uintptr_t lhs_stride,
                                       uintptr_t rhs_base, uintptr_t rhs_stride,
                                       size_t count, REDOP redop)
      {
        size_t tid = blockIdx.x * blockDim.x + threadIdx.x;
        for(size_t idx = tid; tid < count; tid += blockDim.x * gridDim.x)
          redop.template apply_hip<EXCL>(
              *reinterpret_cast<typename REDOP::LHS *>(lhs_base + idx * lhs_stride),
              *reinterpret_cast<const typename REDOP::RHS *>(rhs_base +
                                                             idx * rhs_stride));
      }
 
      template <typename REDOP, bool EXCL>
      __global__ void fold_hip_kernel(uintptr_t rhs1_base, uintptr_t rhs1_stride,
                                      uintptr_t rhs2_base, uintptr_t rhs2_stride,
                                      size_t count, REDOP redop)
      {
        size_t tid = blockIdx.x * blockDim.x + threadIdx.x;
        for(size_t idx = tid; tid < count; tid += blockDim.x * gridDim.x)
          redop.template fold_hip<EXCL>(
              *reinterpret_cast<typename REDOP::RHS *>(rhs1_base + idx * rhs1_stride),
              *reinterpret_cast<const typename REDOP::RHS *>(rhs2_base +
                                                             idx * rhs2_stride));
      }
    }; // namespace ReductionKernels
 
    // this helper adds the appropriate kernels for REDOP to a ReductionOpUntyped,
    //  although the latter is templated to work around circular include deps
    template <typename REDOP, typename T /*= ReductionOpUntyped*/>
    void add_hip_redop_kernels(T *redop)
    {
      // store the host proxy function pointer, as it's the same for all
      //  devices - translation to actual cudaFunction_t's happens later
      redop->hip_apply_excl_fn =
          reinterpret_cast<void *>(&ReductionKernels::apply_hip_kernel<REDOP, true>);
      redop->hip_apply_nonexcl_fn =
          reinterpret_cast<void *>(&ReductionKernels::apply_hip_kernel<REDOP, false>);
      redop->hip_fold_excl_fn =
          reinterpret_cast<void *>(&ReductionKernels::fold_hip_kernel<REDOP, true>);
      redop->hip_fold_nonexcl_fn =
          reinterpret_cast<void *>(&ReductionKernels::fold_hip_kernel<REDOP, false>);
    }
#endif
 
  }; // namespace Hip
 
}; // namespace Realm
 
#endif