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

Remove use of device_idx in dh::LaunchN. (#7063)

Browse Source 
It's an unused parameter, removing it can make the CI log more readable.
This commit is contained in:
Jiaming Yuan
2021-06-29 11:37:26 +08:00
committed by GitHub
parent dd4db347f3
commit 1c8fdf2218
25 changed files with 105 additions and 107 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/tree/gpu_hist/row_partitioner.cu
View File

@@ -96,7 +96,7 @@ void RowPartitioner::SortPosition(common::Span<bst_node_t> position,
void Reset(int device_idx, common::Span<RowPartitioner::RowIndexT> ridx,
common::Span<bst_node_t> position) {
CHECK_EQ(ridx.size(), position.size());
dh::LaunchN(device_idx, ridx.size(), [=] __device__(size_t idx) {
dh::LaunchN(ridx.size(), [=] __device__(size_t idx) {
ridx[idx] = idx;
position[idx] = 0;
});
@@ -131,7 +131,7 @@ common::Span<const RowPartitioner::RowIndexT> RowPartitioner::GetRows(
// Return empty span here as a valid result
// Will error if we try to construct a span from a pointer with size 0
if (segment.Size() == 0) {
return common::Span<const RowPartitioner::RowIndexT>();
return {};
}
return ridx_.CurrentSpan().subspan(segment.begin, segment.Size());
}
@@ -180,7 +180,7 @@ void RowPartitioner::SortPositionAndCopy(const Segment& segment,
const auto d_position_other = position_.Other() + segment.begin;
const auto d_ridx_current = ridx_.Current() + segment.begin;
const auto d_ridx_other = ridx_.Other() + segment.begin;
dh::LaunchN(device_idx_, segment.Size(), stream, [=] __device__(size_t idx) {
dh::LaunchN(segment.Size(), stream, [=] __device__(size_t idx) {
d_position_current[idx] = d_position_other[idx];
d_ridx_current[idx] = d_ridx_other[idx];
});

4
src/tree/gpu_hist/row_partitioner.cuh
View File

@@ -120,7 +120,7 @@ class RowPartitioner {
int64_t* d_left_count = left_counts_.data().get() + nidx;
// Launch 1 thread for each row
dh::LaunchN<1, 128>(device_idx_, segment.Size(), [=] __device__(size_t idx) {
dh::LaunchN<1, 128>(segment.Size(), [=] __device__(size_t idx) {
// LaunchN starts from zero, so we restore the row index by adding segment.begin
idx += segment.begin;
RowIndexT ridx = d_ridx[idx];
@@ -160,7 +160,7 @@ class RowPartitioner {
void FinalisePosition(FinalisePositionOpT op) {
auto d_position = position_.Current();
const auto d_ridx = ridx_.Current();
dh::LaunchN(device_idx_, position_.Size(), [=] __device__(size_t idx) {
dh::LaunchN(position_.Size(), [=] __device__(size_t idx) {
auto position = d_position[idx];
RowIndexT ridx = d_ridx[idx];
bst_node_t new_position = op(ridx, position);

14
src/tree/updater_gpu_hist.cu
View File

@@ -94,8 +94,8 @@ class DeviceHistogram {
void Reset() {
auto d_data = data_.data().get();
dh::LaunchN(device_id_, data_.size(),
[=] __device__(size_t idx) { d_data[idx] = 0.0f; });
dh::LaunchN(data_.size(),
[=] __device__(size_t idx) { d_data[idx] = 0.0f; });
nidx_map_.clear();
}
bool HistogramExists(int nidx) const {
@@ -130,7 +130,7 @@ class DeviceHistogram {
}
// Zero recycled memory
auto d_data = data_.data().get() + nidx_map_[nidx];
dh::LaunchN(device_id_, n_bins_ * 2,
dh::LaunchN(n_bins_ * 2,
[=] __device__(size_t idx) { d_data[idx] = 0.0f; });
} else {
// Append new node histogram
@@ -367,7 +367,7 @@ struct GPUHistMakerDevice {
dh::TemporaryArray<GPUExpandEntry> entries(2);
auto evaluator = tree_evaluator.GetEvaluator<GPUTrainingParam>();
auto d_entries = entries.data().get();
dh::LaunchN(device_id, 2, [=] __device__(size_t idx) {
dh::LaunchN(2, [=] __device__(size_t idx) {
auto split = d_splits_out[idx];
auto nidx = idx == 0 ? left_nidx : right_nidx;
@@ -402,7 +402,7 @@ struct GPUHistMakerDevice {
auto d_node_hist_histogram = hist.GetNodeHistogram(nidx_histogram);
auto d_node_hist_subtraction = hist.GetNodeHistogram(nidx_subtraction);
dh::LaunchN(device_id, page->Cuts().TotalBins(), [=] __device__(size_t idx) {
dh::LaunchN(page->Cuts().TotalBins(), [=] __device__(size_t idx) {
d_node_hist_subtraction[idx] =
d_node_hist_parent[idx] - d_node_hist_histogram[idx];
});
@@ -545,7 +545,7 @@ struct GPUHistMakerDevice {
auto d_node_sum_gradients = device_node_sum_gradients.data().get();
auto evaluator = tree_evaluator.GetEvaluator<GPUTrainingParam>();
dh::LaunchN(device_id, d_ridx.size(), [=] __device__(int local_idx) {
dh::LaunchN(d_ridx.size(), [=] __device__(int local_idx) {
int pos = d_position[local_idx];
bst_float weight = evaluator.CalcWeight(
pos, param_d, GradStats{d_node_sum_gradients[pos]});
@@ -676,7 +676,7 @@ struct GPUHistMakerDevice {
auto evaluator = tree_evaluator.GetEvaluator<GPUTrainingParam>();
GPUTrainingParam gpu_param(param);
auto depth = p_tree->GetDepth(kRootNIdx);
dh::LaunchN(device_id, 1, [=] __device__(size_t idx) {
dh::LaunchN(1, [=] __device__(size_t idx) {
float left_weight = evaluator.CalcWeight(kRootNIdx, gpu_param,
GradStats{split.left_sum});
float right_weight = evaluator.CalcWeight(