提交 42645ff7 编写于 作者: Y Yibing Liu

Compute target index on gpu

上级 6ee22c4f
...@@ -30,7 +30,7 @@ class ArgsortOp : public framework::OperatorWithKernel { ...@@ -30,7 +30,7 @@ class ArgsortOp : public framework::OperatorWithKernel {
"Output(Indices) of ArgsortOp should not be null."); "Output(Indices) of ArgsortOp should not be null.");
auto in_dims = ctx->GetInputDim("X"); auto in_dims = ctx->GetInputDim("X");
int axis = static_cast<int>(ctx->Attrs().Get<int>("axis")); int axis = ctx->Attrs().Get<int>("axis");
auto num_dims = in_dims.size(); auto num_dims = in_dims.size();
PADDLE_ENFORCE(axis < num_dims, PADDLE_ENFORCE(axis < num_dims,
......
...@@ -26,6 +26,42 @@ namespace operators { ...@@ -26,6 +26,42 @@ namespace operators {
using Tensor = framework::Tensor; using Tensor = framework::Tensor;
using platform::PADDLE_CUDA_NUM_THREADS; using platform::PADDLE_CUDA_NUM_THREADS;
__global__ void ComputeTargetIdx(const int64_t* in_dims, int dims_size,
int axis, int64_t n, int64_t* trg_idx,
int64_t* med_ids) {
int64_t index = threadIdx.x + blockDim.x * blockIdx.x;
if (index < n) {
int64_t* shape_out_axis = new int64_t[dims_size - 1];
int64_t* dims_out_axis = new int64_t[dims_size - 1];
int64_t tmp = index;
int64_t pos_in_axis = 0;
int64_t i = dims_size - 2;
int64_t dim_axis = 0;
for (int64_t j = dims_size - 1; j >= 0; --j) {
int64_t dim = in_dims[j];
if (j != axis) {
shape_out_axis[i] = tmp % dim;
dims_out_axis[i] = dim;
i--;
} else {
dim_axis = dim;
pos_in_axis = tmp % dim_axis;
}
tmp /= dim;
}
int64_t group = (dims_size > 1) ? shape_out_axis[0] : 0;
for (int64_t j = 0; j < dims_size - 2; ++j) {
group = group * dims_out_axis[j + 1] + shape_out_axis[j + 1];
}
int64_t traget_idx = group * dim_axis + pos_in_axis;
trg_idx[index] = traget_idx;
med_ids[traget_idx] = pos_in_axis;
delete[] shape_out_axis;
delete[] dims_out_axis;
}
}
template <typename T> template <typename T>
__global__ void PermuteInData(const T* in, const int64_t* trg_idx, int64_t n, __global__ void PermuteInData(const T* in, const int64_t* trg_idx, int64_t n,
T* med_out) { T* med_out) {
...@@ -76,50 +112,27 @@ class ArgsortOpCUDAKernel : public framework::OpKernel<T> { ...@@ -76,50 +112,27 @@ class ArgsortOpCUDAKernel : public framework::OpKernel<T> {
int64_t numel = input->numel(); int64_t numel = input->numel();
int64_t groups = numel / in_dims[axis]; int64_t groups = numel / in_dims[axis];
// Mediate tensor for sorting std::vector<int64_t> in_dims_vec = vectorize(in_dims);
Tensor mediate_output; thrust::device_vector<int64_t> in_dims_dev(in_dims_vec.begin(),
in_dims_vec.end());
int64_t* in_dims_data = thrust::raw_pointer_cast(in_dims_dev.data());
// Mediate tensor for sorting data and indices
Tensor mediate_output, mediate_indices;
T* med_out_data = T* med_out_data =
mediate_output.mutable_data<T>(input->dims(), ctx.GetPlace()); mediate_output.mutable_data<T>(input->dims(), ctx.GetPlace());
int64_t* med_ids_data =
// The target index of each elemement in mediate tensor mediate_indices.mutable_data<int64_t>(in_dims, ctx.GetPlace());
std::vector<int64_t> target_idx(numel, 0); // Target index of each element along the given axis in the mediate tensors
// To record the index along the given axis for the data in mediate tensor Tensor trg_idx_t;
std::vector<int64_t> mediate_indices(numel, 0); int64_t* trg_idx = trg_idx_t.mutable_data<int64_t>(in_dims, ctx.GetPlace());
std::vector<int64_t> in_dims_out_axis = vectorize(in_dims);
in_dims_out_axis.erase(in_dims_out_axis.begin() + axis);
for (int64_t index = 0; index < numel; ++index) {
int64_t tmp = index;
int64_t pos_in_axis = 0;
std::vector<int64_t> shape;
for (int64_t j = in_dims.size() - 1; j >= 0; --j) {
if (j != axis) {
shape.push_back(tmp % in_dims[j]);
} else {
pos_in_axis = tmp % in_dims[j];
}
tmp /= in_dims[j];
}
std::reverse(shape.begin(), shape.end());
int64_t group = (shape.size() > 0) ? shape[0] : 0;
for (size_t j = 0; j < shape.size() - 1; ++j) {
group = group * in_dims_out_axis[j + 1] + shape[j + 1];
}
target_idx[index] = group * in_dims[axis] + pos_in_axis;
mediate_indices[target_idx[index]] = pos_in_axis;
}
thrust::device_vector<int64_t> med_ids_dev(mediate_indices.begin(),
mediate_indices.end());
int64_t* med_ids_data = thrust::raw_pointer_cast(med_ids_dev.data());
thrust::device_vector<int64_t> trg_idx_dev(target_idx.begin(),
target_idx.end());
int64_t* trg_idx = thrust::raw_pointer_cast(trg_idx_dev.data());
auto stream = reinterpret_cast<const platform::CUDADeviceContext&>( auto stream = reinterpret_cast<const platform::CUDADeviceContext&>(
ctx.device_context()) ctx.device_context())
.stream(); .stream();
auto num_threads = PADDLE_CUDA_NUM_THREADS; int num_threads = PADDLE_CUDA_NUM_THREADS;
ComputeTargetIdx<<<(numel - 1) / num_threads + 1, num_threads, 0, stream>>>(
in_dims_data, in_dims.size(), axis, numel, trg_idx, med_ids_data);
PermuteInData<<<(numel - 1) / num_threads + 1, num_threads, 0, stream>>>( PermuteInData<<<(numel - 1) / num_threads + 1, num_threads, 0, stream>>>(
in_data, trg_idx, numel, med_out_data); in_data, trg_idx, numel, med_out_data);
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册