Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

enable rocm, fix row_partitioner.cuh

Browse Source
This commit is contained in:
amdsc21 2023-03-08 06:22:25 +01:00
parent 0fc1f640a9
commit 270c7b4802
1 changed files with 37 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
src/tree/gpu_hist/row_partitioner.cuh
View File

@ -116,7 +116,13 @@ template <typename RowIndexT, typename OpT, typename OpDataT>
void SortPositionBatch(common::Span<const PerNodeData<OpDataT>> d_batch_info,
common::Span<RowIndexT> ridx, common::Span<RowIndexT> ridx_tmp,
common::Span<bst_uint> d_counts, std::size_t total_rows, OpT op,
dh::device_vector<int8_t>* tmp, cudaStream_t stream) {
dh::device_vector<int8_t>* tmp,
#if defined(XGBOOST_USE_HIP)
hipStream_t stream
#else
cudaStream_t stream
#endif
) {
dh::LDGIterator<PerNodeData<OpDataT>> batch_info_itr(d_batch_info.data());
WriteResultsFunctor<OpDataT> write_results{batch_info_itr, ridx.data(), ridx_tmp.data(),
d_counts.data()};
@ -221,7 +227,12 @@ class RowPartitioner {
dh::device_vector<int8_t> tmp_;
dh::PinnedMemory pinned_;
dh::PinnedMemory pinned2_;
#if defined(XGBOOST_USE_HIP)
hipStream_t stream_;
#else
cudaStream_t stream_;
#endif
public:
RowPartitioner(int device_idx, size_t num_rows);
@ -276,9 +287,16 @@ class RowPartitioner {
h_batch_info[i] = {ridx_segments_.at(nidx.at(i)).segment, op_data.at(i)};
total_rows += ridx_segments_.at(nidx.at(i)).segment.Size();
}
#if defined(XGBOOST_USE_HIP)
dh::safe_cuda(hipMemcpyAsync(d_batch_info.data().get(), h_batch_info.data(),
h_batch_info.size() * sizeof(PerNodeData<OpDataT>),
hipMemcpyDefault, stream_));
#else
dh::safe_cuda(cudaMemcpyAsync(d_batch_info.data().get(), h_batch_info.data(),
h_batch_info.size() * sizeof(PerNodeData<OpDataT>),
cudaMemcpyDefault, stream_));
#endif
// Temporary arrays
auto h_counts = pinned_.GetSpan<bst_uint>(nidx.size(), 0);
@ -288,11 +306,22 @@ class RowPartitioner {
SortPositionBatch<RowIndexT, UpdatePositionOpT, OpDataT>(
dh::ToSpan(d_batch_info), dh::ToSpan(ridx_), dh::ToSpan(ridx_tmp_), dh::ToSpan(d_counts),
total_rows, op, &tmp_, stream_);
#if defined(XGBOOST_USE_HIP)
dh::safe_cuda(hipMemcpyAsync(h_counts.data(), d_counts.data().get(), h_counts.size_bytes(),
hipMemcpyDefault, stream_));
#else
dh::safe_cuda(cudaMemcpyAsync(h_counts.data(), d_counts.data().get(), h_counts.size_bytes(),
cudaMemcpyDefault, stream_));
#endif
// TODO(Rory): this synchronisation hurts performance a lot
// Future optimisation should find a way to skip this
#if defined(XGBOOST_USE_HIP)
dh::safe_cuda(hipStreamSynchronize(stream_));
#else
dh::safe_cuda(cudaStreamSynchronize(stream_));
#endif
// Update segments
for (size_t i = 0; i < nidx.size(); i++) {
@ -325,9 +354,16 @@ class RowPartitioner {
template <typename FinalisePositionOpT>
void FinalisePosition(common::Span<bst_node_t> d_out_position, FinalisePositionOpT op) {
dh::TemporaryArray<NodePositionInfo> d_node_info_storage(ridx_segments_.size());
#if defined(XGBOOST_USE_HIP)
dh::safe_cuda(hipMemcpyAsync(d_node_info_storage.data().get(), ridx_segments_.data(),
sizeof(NodePositionInfo) * ridx_segments_.size(),
hipMemcpyDefault, stream_));
#else
dh::safe_cuda(cudaMemcpyAsync(d_node_info_storage.data().get(), ridx_segments_.data(),
sizeof(NodePositionInfo) * ridx_segments_.size(),
cudaMemcpyDefault, stream_));
#endif
constexpr int kBlockSize = 512;
const int kItemsThread = 8;