Google Git
Sign in
apache / tvm / bdcfa01eae3ffe8c6d39aa26d0d1e5b311d47efb / . / include / tvm / topi / transform.h
blob: 81935dd72dda6456c8a76f1aab5e16f0b678c4fc [file] [log] [blame]
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you 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.
*/
/*!
* \file topi/transform.h
* \brief Transform op constructors
*/
#ifndef TVM_TOPI_TRANSFORM_H_
#define TVM_TOPI_TRANSFORM_H_
#include <tvm/te/operation.h>
#include <tvm/tir/data_layout.h>
#include <tvm/tir/index_map.h>
#include <tvm/topi/broadcast.h>
#include <tvm/topi/detail/broadcast.h>
#include <tvm/topi/detail/constant_utils.h>
#include <tvm/topi/detail/ravel_unravel.h>
#include <tvm/topi/detail/strided_slice.h>
#include <tvm/topi/detail/tensor_utils.h>
#include <tvm/topi/tags.h>
#include <algorithm>
#include <iterator>
#include <limits>
#include <string>
#include <unordered_set>
#include <vector>
namespace tvm {
namespace topi {
using namespace tvm::te;
using namespace topi::detail;
/*!
* \brief Creates an operation to slide a window over the input x.
*
* \param x The input tensor.
* \param axis What axis the window begins sliding over. Window will be slid
* over this axis and all following axes. The axis value determines the window
* shape (and thus, the number of strides): window shape and strides must both
* be of length `data.ndim-axis`.
* \param window_shape The window shape to form over the input. Window shape
* must be of length `data.ndim-axis`.
* \param strides How to stride the window along each dimension. Strides must be
* of length `data.ndim-axis`.
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the sliding_window operation
*/
inline Tensor sliding_window(const Tensor& x, int axis, Array<Integer> window_shape,
Array<Integer> strides, std::string name = "T_sliding_window",
std::string tag = "") {
CHECK_GE(axis, 0);
auto _axis = size_t(axis);
CHECK_LT(_axis, x->shape.size()) << "axis must be a valid dimension index of x.";
CHECK_EQ(x->shape.size() - _axis, window_shape.size())
<< "There must be a window shape for every dimension of x "
<< "over which we are sliding the window.";
CHECK_EQ(strides.size(), window_shape.size()) << "Windows and strides should be the same length.";
// Compute the new shape.
Array<PrimExpr> new_shape;
// Dimensions up until `axis` remain the same.
for (size_t i = 0; i < _axis; ++i) {
new_shape.push_back(x->shape[i]);
}
// New dimensions which result from sliding the window in each dimension. One new dimension per
// window dimension.
for (size_t i = 0; i < window_shape.size(); ++i) {
// Length of the shape along this dimension.
auto dim_len = x->shape[_axis + i];
// Length of the window along this dimension.
auto window_len = window_shape[i];
// Strides along this dimension.
auto stride = strides[i];
new_shape.push_back(floordiv(dim_len - (window_len - 1) + stride - 1, stride));
}
// Dimensions comprising the window.
for (size_t i = 0; i < window_shape.size(); ++i) {
new_shape.push_back(window_shape[i]);
}
ICHECK(new_shape.size() == _axis + 2 * window_shape.size());
return compute(
new_shape,
[&](const Array<Var>& indices) {
// The index at which to index the old tensor x.
Array<PrimExpr> idx;
// Dimensions up until `axis` remain the same.
for (size_t i = 0; i < _axis; ++i) {
idx.push_back(indices[i]);
}
for (size_t i = 0; i < window_shape.size(); ++i) {
// Which window in this dimension we are indexing.
auto window_idx = indices[_axis + i];
// Which index within the window we are indexing.
auto idx_within_window = indices[_axis + window_shape.size() + i];
// Stride value for this dimension.
auto stride = strides[i];
idx.push_back(window_idx * stride + idx_within_window);
}
ICHECK(idx.size() == x->shape.size());
return x(idx);
},
name, tag);
}
/*!
* \brief Creates an operation to insert new dimensions of length 1
*
* \param x The input tensor
* \param axis The index of the first new dimension (allows negative
* indices as offsets from the last dimension)
* \param num_newaxis The number of new dimensions to insert
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the dim expansion operation
*/
inline Tensor expand_dims(const Tensor& x, int axis, int num_newaxis = 1,
std::string name = "T_expand_dims", std::string tag = kBroadcast) {
int ndim = static_cast<int>(x->shape.size());
ICHECK(-ndim - 1 <= axis && axis <= ndim)
<< "expand_dims only accepts `axis` in [-data.ndim - 1, data.ndim]"
<< ", but got axis = " << axis << ", and data.ndim = " << ndim;
ICHECK(num_newaxis >= 0) << "expand_dims only accepts `num_newaxis >= 0`"
<< ", but got num_newaxis = " << num_newaxis;
if (axis < 0) {
// Calculate offset from last dimension
axis = ndim + axis + 1;
}
Array<PrimExpr> new_shape;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) {
new_shape.push_back(x->shape[i]);
}
for (size_t i = 0; i < static_cast<size_t>(num_newaxis); ++i) {
new_shape.push_back(1);
}
for (size_t i = axis; i < x->shape.size(); ++i) {
new_shape.push_back(x->shape[i]);
}
return compute(
new_shape,
[&](const Array<Var>& indices) {
Array<PrimExpr> idx;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) {
idx.push_back(indices[i]);
}
for (size_t i = axis + num_newaxis; i < indices.size(); ++i) {
idx.push_back(indices[i]);
}
return x(idx);
},
name, tag);
}
/*!
* \brief Permute the dimensions of an array
*
* \param x The input tensor
* \param axes The indices of the permutation. If this is empty,
* the dimensions will be reversed.
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the transpose operation
*/
inline Tensor transpose(const Tensor& x, Array<Integer> axes, std::string name = "T_transpose",
std::string tag = kInjective) {
if (!axes.defined() || axes.size() == 0) {
axes = Array<Integer>();
for (int i = static_cast<int>(x->shape.size()) - 1; i >= 0; --i) {
axes.push_back(i);
}
}
Array<PrimExpr> new_shape;
for (size_t i = 0; i < axes.size(); ++i) {
int axis = static_cast<int>(axes[i]->value);
int new_axis = axis;
if (axis < 0) {
new_axis = static_cast<int>(x->shape.size()) + axis;
axes.Set(i, new_axis);
}
ICHECK((new_axis >= 0) && (new_axis < static_cast<int>(x->shape.size())))
<< "axis=" << axis << " is invalid for the " << static_cast<int>(x->shape.size())
<< "-dimensional input tensor";
for (size_t j = 0; j < axes.size(); ++j) {
if (i != j) {
ICHECK(new_axis != static_cast<int>(axes[j]->value)) << "repeated axis in transpose";
}
}
new_shape.push_back(x->shape[new_axis]);
}
return compute(
new_shape,
[&](const Array<Var>& indices) {
std::vector<PrimExpr> idx;
for (size_t i = 0; i < axes.size(); ++i) {
idx.push_back(1);
}
for (size_t i = 0; i < axes.size(); ++i) {
int axis = static_cast<int>(axes[i]->value);
idx[axis] = indices[i];
}
return x(idx);
},
name, tag);
}
/*!
* \brief Reverse the tensor for variable length slices.
* Input is first sliced along batch axis and then elements are reversed along seq axis.
*
* \param x The input tensor
* \param seq_lengths A 1D Tensor with length x.dims[batch_axis]. Optional Tensor() can be passed.
* If not defined batch axis is ignored and tensor is reversed along seq_axis.
* \param seq_axis The axis along which the elements will be reveresed
* \param batch_axis The axis along which the tensor will be sliced
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the reverse_sequence operation
*/
inline Tensor reverse_sequence(const Tensor& x, const Tensor& seq_lengths, int seq_axis = 1,
int batch_axis = 0, std::string name = "T_reverse_sequence",
std::string tag = kInjective) {
size_t src_tensor_dim = x->shape.size();
int seq_axis_inp = seq_axis;
if (seq_lengths.defined()) {
size_t seq_lengths_dim = seq_lengths->shape.size();
int batch_axis_inp = batch_axis;
if (batch_axis < 0) {
batch_axis = static_cast<int>(x->shape.size()) + batch_axis;
}
ICHECK(seq_lengths_dim == 1) << "seq_lengths should be 1D vector";
ICHECK(GetConstInt(seq_lengths->shape[0]) == GetConstInt(x->shape[batch_axis]))
<< "For reverse_sequnece seq_lengths size should match with dimension of batch axis"
<< ", but got dimension of batch_axis = " << GetConstInt(x->shape[batch_axis])
<< ", and seq_length size = " << GetConstInt(seq_lengths->shape[0]);
ICHECK((0 <= batch_axis) && (batch_axis < static_cast<int>(x->shape.size())))
<< "batch_axis=" << batch_axis_inp << " is invalid for the "
<< static_cast<int>(x->shape.size()) << "-dimensional input tensor";
}
if (seq_axis < 0) {
seq_axis = static_cast<int>(x->shape.size()) + seq_axis;
}
ICHECK((0 <= seq_axis) && (seq_axis < static_cast<int>(x->shape.size())))
<< "seq_axis=" << seq_axis_inp << " is invalid for the " << static_cast<int>(x->shape.size())
<< "-dimensional input tensor";
auto func = [&](const Array<Var>& indices) {
Array<PrimExpr> real_indices;
for (size_t i = 0; i < src_tensor_dim; ++i) {
if (i == static_cast<size_t>(seq_axis)) {
if (seq_lengths.defined()) {
auto len = seq_lengths(indices[batch_axis]);
auto idx = if_then_else(
len <= 1 || len <= indices[i], indices[i],
if_then_else(len > x->shape[i], x->shape[i] - 1 - indices[i], len - 1 - indices[i]));
real_indices.push_back(idx);
} else {
real_indices.push_back(x->shape[i] - 1 - indices[i]);
}
} else {
real_indices.push_back(indices[i]);
}
}
return x(real_indices);
};
return compute(x->shape, func, name, tag);
}
/*!
* \brief Reshape a tensor
*
* \param x The input tensor
* \param newshape The new shape
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the reshape operation
*/
inline Tensor reshape(const Tensor& x, Array<PrimExpr> newshape, std::string name = "T_reshape",
std::string tag = kInjective) {
auto x_shape = x->shape;
Array<PrimExpr> target_shape;
for (const auto& ele : newshape) {
if (ele.as<IntImmNode>()) {
target_shape.push_back(cast(DataType::Int(32), ele));
} else {
target_shape.push_back(ele);
}
}
// If either the input shape or the target shape contains a zero, return an empty tensor.
if (is_empty_shape(target_shape) || is_empty_shape(x->shape)) {
return compute(
target_shape, [&](const Array<Var>& indices) { return tvm::cast(x->dtype, 0); }, name, tag);
} else {
return compute(
target_shape,
[&](const Array<Var>& indices) {
return x(UnravelIndex(
RavelIndex(Array<PrimExpr>{indices.begin(), indices.end()}, target_shape), x_shape));
},
name, tag);
}
}
/*!
* \brief Converts a flat index or array of flat indices into a tuple of coordinate arrays
*
* \param x The input tensor having indices.
* \param shape The shape tensor
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor of coordinate arrays.
*/
inline Tensor unravel_index(const Tensor& x, const Tensor& shape, std::string name = "T_unravel",
std::string tag = kInjective) {
auto x_shape = x->shape;
auto shape_shape = shape->shape;
Array<PrimExpr> oshape;
oshape.push_back(shape_shape[0]);
if (x_shape.size() != 0) {
oshape.push_back(x_shape[0]);
}
auto func = [&](const Array<Var>& indices) {
auto i = indices[0];
std::vector<PrimExpr> indices_divs;
PrimExpr ret = 0;
PrimExpr cur_val = 0;
PrimExpr index_val = 0;
if (x_shape.size() != 0) {
index_val = x[indices[1]];
} else {
index_val = x();
}
indices_divs.push_back(index_val);
for (int v = GetConstInt(shape_shape[0]) - 1; v >= 0; --v) {
ret = tvm::if_then_else(i == v, indexmod(indices_divs.back(), shape[v]), ret);
cur_val = indexdiv(indices_divs.back(), shape[v]);
indices_divs.push_back(cur_val);
}
return ret;
};
return compute(oshape, func, name, tag);
}
/*!
* \brief Remove size 1 dimensions from the shape of a tensor.
* The removed dimensions must have a constant size of 1.
*
* \param x The input tensor
* \param axis Indices of the dimensions to remove. If this is empty,
* all entries with a constant size of 1 will be removed.
* \param atleast1d Whether the output need to be atleast1d.
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the squeeze operation
*/
inline Tensor squeeze(const Tensor& x, Array<Integer> axis, bool atleast1d = false,
std::string name = "T_squeeze", std::string tag = kInjective) {
auto ndim = x->shape.size();
std::vector<int> axis_val;
if (!axis.defined() || axis.size() == 0) {
for (size_t i = 0; i < ndim; ++i) {
if (IsConstInt(x->shape[i]) && GetConstInt(x->shape[i]) == 1) {
axis_val.push_back(static_cast<int>(i));
}
}
} else {
for (size_t i = 0; i < axis.size(); ++i) {
int64_t val = axis[i]->value;
if (val < 0) {
val += static_cast<int>(x->shape.size());
}
if (IsConstInt(x->shape[val])) {
ICHECK_EQ(GetConstInt(x->shape[val]), 1) << "Dimension " << val << " must have size 1";
}
axis_val.push_back(val);
}
}
std::unordered_set<int> axis_set(axis_val.begin(), axis_val.end());
Array<PrimExpr> out_shape;
for (size_t i = 0; i < ndim; ++i) {
if (axis_set.count(static_cast<int>(i)) == 0) {
out_shape.push_back(x->shape[i]);
}
}
if (out_shape.size() == 0 && atleast1d) {
out_shape.push_back(1);
}
return compute(
out_shape,
[&](const Array<Var>& indices) {
Array<PrimExpr> real_indices;
int flag = 0;
for (size_t i = 0; i < ndim; ++i) {
if (axis_set.count(static_cast<int>(i)) == 0) {
real_indices.push_back(indices[i - flag]);
} else {
real_indices.push_back(0);
flag += 1;
}
}
return x(real_indices);
},
name, tag);
}
/*!
* \brief Join a sequence of tensors along an existing axis
*
* \param inputs The input tensors
* \param axis The axis along which the tensors will be joined
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the concatenate operation
*/
inline Tensor concatenate(const Array<Tensor>& inputs, int axis = 0, std::string name = "T_concat",
std::string tag = kInjective) {
int ndim = static_cast<int>(inputs[0]->shape.size());
ICHECK(-ndim <= axis && axis < ndim) << "concatenate only accepts `axis` in [-ndim, ndim)"
<< ", but got axis = " << axis << ", and ndim = " << ndim;
if (axis < 0) {
axis += ndim;
}
ICHECK_LT(axis, inputs[0]->shape.size()) << "axis out of bounds";
Array<PrimExpr> axis_sizes;
for (auto t : inputs) {
axis_sizes.push_back(t->shape[axis]);
}
arith::Analyzer analyzer;
PrimExpr join_size = axis_sizes[0];
for (size_t i = 1; i < axis_sizes.size(); ++i) {
join_size += axis_sizes[i];
}
join_size = analyzer.Simplify(join_size);
Array<PrimExpr> out_shape;
for (size_t i = 0; i < inputs[0]->shape.size(); ++i) {
out_shape.push_back(i == static_cast<size_t>(axis) ? join_size : inputs[0]->shape[i]);
}
return compute(
out_shape,
[&](const Array<Var>& indices) {
auto ret = inputs[0](indices);
auto ind = indices[axis];
for (size_t i = 0; i < inputs.size() - 1; ++i) {
ind -= axis_sizes[i];
Array<PrimExpr> idx;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) {
idx.push_back(indices[i]);
}
idx.push_back(ind);
for (size_t i = axis + 1; i < indices.size(); ++i) {
idx.push_back(indices[i]);
}
ret = tvm::if_then_else(ind >= 0, inputs[i + 1](idx), ret);
}
return ret;
},
name, tag);
}
/*!
* \brief Join a sequence of tensors along a new axis.
*
* \param inputs The input tensors
* \param axis The axis along which the tensors will be stacked
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the stack operation
*/
inline Tensor stack(const Array<Tensor>& inputs, int axis = 0, std::string name = "T_stack",
std::string tag = kInjective) {
int ndim = static_cast<int>(inputs[0]->shape.size());
ICHECK(-ndim - 1 <= axis && axis <= ndim)
<< "stack only accepts `axis` in [-ndim, ndim)"
<< ", but got axis = " << axis << ", and ndim = " << ndim;
if (axis < 0) {
axis += ndim + 1;
}
ICHECK_LT(axis, inputs[0]->shape.size() + 1) << "axis out of bounds";
const int stack_size = static_cast<int>(inputs.size());
Array<PrimExpr> out_shape;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) out_shape.push_back(inputs[0]->shape[i]);
out_shape.push_back(stack_size);
for (size_t i = static_cast<size_t>(axis); i < static_cast<size_t>(ndim); ++i)
out_shape.push_back(inputs[0]->shape[i]);
return compute(
out_shape,
[&](const Array<Var>& indices) {
Array<PrimExpr> idx;
for (size_t i = 0; i < indices.size(); ++i)
if (i != static_cast<size_t>(axis)) idx.push_back(indices[i]);
auto ind = indices[axis];
auto ret = inputs[0](idx);
for (int i = 0; i < static_cast<int>(inputs.size() - 1); ++i) {
ret = tvm::if_then_else(ind == i + 1, inputs[i + 1](idx), ret);
}
return ret;
},
name, tag);
}
/*!
* \brief Split a tensor into multiple sub-tensors
*
* \param x The input tensor
* \param split_indices The indices to split the input at. This must be in ascending
* order.
* \param axis The axis to split along.
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the split operation
*/
inline Array<Tensor> split(const Tensor& x, Array<PrimExpr> split_indices, int axis,
std::string name = "T_split", std::string tag = kInjective) {
if (axis < 0) {
axis += static_cast<int>(x->shape.size());
}
ICHECK_LT(axis, x->shape.size()) << "axis out of bounds";
auto src_axis_size = x->shape[axis];
std::vector<PrimExpr> begin_ids;
begin_ids.push_back(0);
for (auto idx : split_indices) {
auto idx_node = idx.as<IntImmNode>();
auto back_node = begin_ids.back().as<IntImmNode>();
if (idx_node && back_node) {
ICHECK_GT(idx_node->value, back_node->value) << "split_indices must be sorted";
}
begin_ids.push_back(idx);
}
Array<Array<PrimExpr> > out_shapes;
for (size_t i = 0; i < begin_ids.size(); ++i) {
PrimExpr out_axis_size;
if (i == begin_ids.size() - 1) {
out_axis_size = src_axis_size - begin_ids[i];
} else {
out_axis_size = begin_ids[i + 1] - begin_ids[i];
}
Array<PrimExpr> shape;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) {
shape.push_back(x->shape[i]);
}
shape.push_back(out_axis_size);
for (size_t i = axis + 1; i < x->shape.size(); ++i) {
shape.push_back(x->shape[i]);
}
out_shapes.push_back(shape);
}
Array<Tensor> result;
for (size_t i = 0; i < begin_ids.size(); ++i) {
result.push_back(compute(
out_shapes[i],
[&](const Array<Var>& indices) {
auto begin = begin_ids[i];
Array<PrimExpr> real_indices;
for (size_t j = 0; j < static_cast<size_t>(axis); ++j) {
real_indices.push_back(indices[j]);
}
real_indices.push_back(indices[axis] + begin);
for (size_t j = axis + 1; j < indices.size(); ++j) {
real_indices.push_back(indices[j]);
}
return x(real_indices);
},
name, tag));
}
return result;
}
/*!
* \brief strided_slice of a tensor where begin/end/stride can be mixed static and dynamic
*
* \param x The input tensor
* \param begin The indices to begin with in the slicing
* \param end Indices indicating end of the slice
* \param strides Specifies the stride values, it can be negative
* in that case, the input tensor will be reversed in that particular axis
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the dynamic_strided_slice operation
*/
inline Tensor dynamic_strided_slice(const Tensor& x, const Array<PrimExpr>& begin,
const Array<PrimExpr>& end, const Array<PrimExpr>& strides,
std::string name = "T_dynamic_strided_slice",
std::string tag = kInjective) {
const size_t src_tensor_dim = x->shape.size();
ICHECK_LE(begin.size(), src_tensor_dim);
ICHECK_LE(end.size(), src_tensor_dim);
ICHECK_LE(strides.size(), src_tensor_dim);
ICHECK_EQ(begin.size(), end.size());
ICHECK_EQ(begin.size(), strides.size());
const size_t num_slice_axes = begin.size();
Array<PrimExpr> out_shape;
for (size_t i = 0; i < num_slice_axes; ++i) {
auto d = indexdiv(end[i] - begin[i], strides[i]);
if (d->IsInstance<tvm::IntImmNode>()) {
// Preserve static dimension if possible
out_shape.push_back(d);
} else {
out_shape.push_back(tvm::tir::Var("dim"));
}
}
for (size_t i = num_slice_axes; i < src_tensor_dim; ++i) {
out_shape.push_back(x->shape[i]);
}
return te::compute(
out_shape,
[&](const Array<tvm::tir::Var>& indices) {
Array<PrimExpr> real_indices;
for (size_t i = 0; i < num_slice_axes; ++i) {
real_indices.push_back(indices[i] * strides[i] + tvm::min(begin[i], x->shape[i] - 1));
}
// keep input dim
for (size_t i = num_slice_axes; i < src_tensor_dim; ++i) {
real_indices.push_back(indices[i]);
}
return x(real_indices);
},
name, tag);
}
/*!
* \brief strided_slice of a tensor with dynamic begin/end/stride
*
* \param x The input tensor
* \param begin The indices to begin with in the slicing
* \param end Indices indicating end of the slice
* \param strides Specifies the stride values, it can be negative
* in that case, the input tensor will be reversed in that particular axis
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the dynamic_strided_slice operation
*/
inline te::Tensor dynamic_strided_slice(const te::Tensor& x, const te::Tensor& begin,
const te::Tensor& end, const te::Tensor& strides,
std::string name = "T_strided_slice_dynamic",
std::string tag = topi::kInjective) {
const int64_t num_dynamic_axes = begin->shape[0].as<IntImmNode>()->value;
ICHECK_EQ(end->shape[0].as<IntImmNode>()->value, num_dynamic_axes);
ICHECK_EQ(strides->shape[0].as<IntImmNode>()->value, num_dynamic_axes);
Array<PrimExpr> begin_expr, end_expr, strides_expr;
for (int64_t i = 0; i < num_dynamic_axes; ++i) {
auto i64_ind = IntImm(DataType::Int(64), i);
begin_expr.push_back(begin(i64_ind));
end_expr.push_back(end(i64_ind));
strides_expr.push_back(strides(i64_ind));
}
return dynamic_strided_slice(x, begin_expr, end_expr, strides_expr, name, tag);
}
/*!
* \brief Calcluate the output shape of strided_slice, the entry point for Relay type relation
*
* \param ishape The input tensor shape
* \param begin The indices to begin with in the slicing
* \param end Indices indicating end of the slice
* \param strides Specifies the stride values, it can be negative
* in that case, the input tensor will be reversed in that particular axis
* \param axes Axes along which slicing is applied. When it is specified, the length of begin, end,
* strides, and axes argument must be equal
* \param slice_mode Specifies the slice mode
*
* \return The output shape of strided_slice using the arguments above
*/
inline Array<PrimExpr> StridedSliceOutputShape(
const Array<PrimExpr>& ishape, const Array<Integer>& begin, const Array<Integer>& end,
const Array<Integer>& strides, const Array<Integer>& axes, const std::string& slice_mode) {
ICHECK(axes.size() == begin.size() && axes.size() == end.size() && axes.size() == strides.size());
std::vector<int64_t> begin_vec, end_vec, strides_vec;
std::tie(begin_vec, end_vec, strides_vec) = ConvertToVec(begin, end, strides, slice_mode);
auto begin_canonicalized = StridedSliceCanonicalizeBegin(ishape, begin_vec, strides_vec, axes,
begin[0]->dtype, slice_mode);
return StridedSliceOutputShape(ishape, begin_vec, end_vec, strides_vec, axes, slice_mode,
begin_canonicalized, true);
}
/*!
* \brief strided_slice of a tensor
*
* \param x The input tensor
* \param begin The indices to begin with in the slicing
* \param end Indices indicating end of the slice
* \param strides Specifies the stride values, it can be negative
* in that case, the input tensor will be reversed in that particular axis
* \param axes Axes along which slicing is applied. When it is specified, the length of begin, end,
* strides, and axes argument must be equal
* \param slice_mode Specifies the slice mode
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the sstrided_slice operation
*/
inline Tensor strided_slice_with_axes(const Tensor& x, const Array<Integer>& begin,
const Array<Integer>& end, const Array<Integer>& strides,
const Array<Integer>& axes, std::string slice_mode = "end",
std::string name = "T_strided_slice_with_axes",
std::string tag = kInjective) {
const size_t src_tensor_dim = x->shape.size();
ICHECK(axes.size() <= src_tensor_dim);
ICHECK(axes.size() == begin.size() && axes.size() == end.size() && axes.size() == strides.size());
std::vector<int64_t> begin_vec, end_vec, strides_vec;
std::tie(begin_vec, end_vec, strides_vec) = ConvertToVec(begin, end, strides, slice_mode);
auto begin_expr = StridedSliceCanonicalizeBegin(x->shape, begin_vec, strides_vec, axes,
begin[0]->dtype, slice_mode);
auto out_shape = StridedSliceOutputShape(x->shape, begin_vec, end_vec, strides_vec, axes,
slice_mode, begin_expr);
return te::compute(
out_shape,
[&](const Array<tir::Var>& indices) {
Array<PrimExpr> real_indices;
for (size_t i = 0; i < out_shape.size(); ++i) real_indices.push_back(indices[i]);
for (size_t i = 0; i < axes.size(); ++i) {
auto stride = make_const(strides[i].dtype(), strides_vec[i]);
PrimExpr ind = indices[axes[i].IntValue()] * stride + begin_expr[i];
real_indices.Set(axes[i].IntValue(), ind);
}
return x(real_indices);
},
name, tag);
}
/*!
* \brief strided_slice of a tensor
*
* \param x The input tensor
* \param begin The indices to begin with in the slicing
* \param end Indices indicating end of the slice
* \param strides Specifies the stride values, it can be negative
* in that case, the input tensor will be reversed in that particular axis
* \param slice_mode Specifies the slice mode
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the strided_slice operation
*/
inline Tensor strided_slice(const Tensor& x, const Array<Integer>& begin, const Array<Integer>& end,
const Array<Integer>& strides, std::string slice_mode = "end",
std::string name = "T_strided_slice", std::string tag = kInjective) {
size_t src_tensor_dim = static_cast<size_t>(x->shape.size());
Array<Integer> axes;
for (size_t i = 0; i < src_tensor_dim; ++i) axes.push_back(i);
Array<Integer> begin_full(begin);
Array<Integer> end_full(end);
Array<Integer> strides_full(strides);
const IntImm one = IntImm(DataType::Int(64), 1);
const IntImm zero = IntImm(DataType::Int(64), 0);
const IntImm max_range = IntImm(DataType::Int(64), std::numeric_limits<int64_t>::max());
for (size_t i = strides.size(); i < src_tensor_dim; ++i) {
strides_full.push_back(one);
}
for (size_t i = begin.size(); i < src_tensor_dim; ++i) {
begin_full.push_back(GetConstInt(strides_full[i]) > 0 ? zero : max_range);
}
for (size_t i = end.size(); i < src_tensor_dim; ++i) {
end_full.push_back(GetConstInt(strides_full[i]) < 0 ? zero : max_range);
}
return strided_slice_with_axes(x, begin_full, end_full, strides_full, axes, slice_mode, name,
tag);
}
/*!
* \brief Split a tensor into a number of sub-tensors
*
* \param x The input tensor
* \param num_sections The number of sections to split the tensor into.
* this must be an integer factor of the size of the axis being split.
* \param axis The axis to split along.
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the split operation
*/
inline Array<Tensor> split_sections(const Tensor& x, int num_sections, int axis,
std::string name = "T_split_sections",
std::string tag = kInjective) {
if (axis < 0) {
axis += static_cast<int>(x->shape.size());
}
ICHECK_LT(axis, x->shape.size()) << "axis out of bounds";
auto src_axis_size = x->shape[axis];
ICHECK_GT(num_sections, 0) << "Slice count must be > 0";
if (auto node = src_axis_size.as<IntImmNode>()) {
ICHECK_EQ(node->value % num_sections, 0)
<< "num_sections must be an integer factor of the size of axis " << axis << " ("
<< node->value << ")";
}
Array<PrimExpr> split_indices;
auto seg_size = indexdiv(src_axis_size, num_sections);
for (int i = 0; i < num_sections; ++i) {
// region at index 0 is added by split()
if (i != 0) {
split_indices.push_back(seg_size * i);
}
}
return split(x, split_indices, axis, name, tag);
}
/*!
* \brief Take elements from an flattened input array when axis is None.
*
* \param a The source array.
* \param indices The indices of the values to extract.
* \param batch_dims The number of batch dimensions.
* \param mode The mode of the operation.
* \param name The name of the operation.
* \param mode The mode of to handle out of bound indices.
* \param tag The tag to mark the operation.
*
* \return A Tensor whose op member is the take operation
*/
inline Tensor take(const Tensor& a, const Tensor& indices, int batch_dims,
std::string mode = "clip", std::string name = "T_take",
std::string tag = kInjective) {
Array<PrimExpr> a_shape = a->shape;
Array<PrimExpr> out_shape = indices->shape;
PrimExpr a_size = 1;
for (size_t i = 0; i < a_shape.size(); ++i) {
a_size = a_size * a_shape[i];
}
if (mode == "clip") {
return compute(
out_shape,
[&](const Array<Var>& out_index) {
auto idx = tvm::min(tvm::max(0, indices(out_index)), a_size - 1);
return a(UnravelIndex(idx, a_shape));
},
name, tag);
} else if (mode == "fast") {
LOG(WARNING) << "Fast mode segfaults when there are out-of-bounds indices. "
"Make sure input indices are in bound";
return compute(
out_shape,
[&](const Array<Var>& out_index) { return a(UnravelIndex(indices(out_index), a_shape)); },
name, tag);
} else { // mode == "wrap"
return compute(
out_shape,
[&](const Array<Var>& out_index) {
auto idx = truncmod(truncmod(indices(out_index), a_size) + a_size, a_size);
return a(UnravelIndex(idx, a_shape));
},
name, tag);
}
}
/*!
* \brief Mask the out-of-boundary elements of each sequence.
*
* \param data The source array.
* \param valid_length The real length of each sequence.
* \param mask_value The masking value.
* \param axis The axis of the temporal dimension of the sequence
* \param name The name of the operation.
* \param tag The tag to mark the operation.
*
* \return A Tensor whose op member is the sequence_mask operation
*/
inline Tensor sequence_mask(const Tensor& data, const Tensor& valid_length, double mask_value,
int axis, std::string name = "T_sequence_mask",
std::string tag = kInjective) {
ICHECK(axis == 0 || axis == 1) << "axis must be either 0 or 1";
ICHECK_EQ(valid_length->shape.size(), 1) << "valid_length must have ndim=1, i.e., (batch_size,).";
auto length_dim = data->shape[axis];
auto batch_dim = data->shape[1 - axis];
Array<PrimExpr> out_shape = data->shape;
Tensor out = compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> len_index;
auto tid = out_index[axis];
auto bid = out_index[1 - axis];
len_index.push_back(bid);
PrimExpr ret =
tvm::if_then_else(tvm::cast(valid_length->dtype, tid) >= valid_length(len_index),
tvm::tir::make_const(data->dtype, mask_value), data(out_index));
return ret;
},
name, tag);
return out;
}
/*!
* \brief Take elements from an array along an axis.
*
* \param a The source array.
* \param indices The indices of the values to extract.
* \param batch_dims The number of batch dimensions. By default is 0.
* \param axis The axis over which to select values. By default,
* the flattened input array is used.
* \param mode The mode for handling out of bound indices.
* \param name The name of the operation.
* \param tag The tag to mark the operation.
*
* \return A Tensor whose op member is the take operation
*/
inline Tensor take(const Tensor& a, const Tensor& indices, int batch_dims, int axis,
std::string mode = "clip", std::string name = "T_take",
std::string tag = kInjective) {
if (axis < 0) {
axis += static_cast<int>(a->shape.size());
}
ICHECK_GE(axis, 0) << "axis out of bounds";
ICHECK_LT(axis, a->shape.size()) << "axis out of bounds";
auto axis_dim = a->shape[axis];
int indices_len = static_cast<int>(indices->shape.size());
int batch_dims_ = batch_dims;
if (batch_dims_ != 0) {
ICHECK_GE(batch_dims_, -static_cast<int>(indices->shape.size())) << "batch_dims out of bounds";
ICHECK_LE(batch_dims_, indices->shape.size()) << "batch_dims out of bounds";
if (batch_dims_ < 0) {
batch_dims_ = indices->shape.size() + batch_dims_;
}
ICHECK_LT(batch_dims_, a->shape.size()) << "batch_dims out of bounds";
ICHECK_LE(batch_dims_, axis) << "batch_dims must be less than or equal to axis";
for (int i = 0; i < batch_dims_; ++i) {
auto addr1 = a->shape[i];
auto addr2 = indices->shape[i];
auto v1 = static_cast<IntImm*>(&addr1)->get()->value;
auto v2 = static_cast<IntImm*>(&addr2)->get()->value;
ICHECK_EQ(v1, v2) << "a.shape[" << i << "] should be equal to indices.shape[" << i << "]";
}
}
// The result shape is a.shape[:axis] + indices.shape[batch_dims:] +
// a.shape[axis + 1:].
Array<PrimExpr> out_shape;
for (int i = 0; i < batch_dims_; ++i) {
out_shape.push_back(a->shape[i]);
}
for (int i = batch_dims_; i < axis; ++i) {
out_shape.push_back(a->shape[i]);
}
for (size_t i = static_cast<size_t>(batch_dims_); i < indices->shape.size(); ++i) {
out_shape.push_back(indices->shape[i]);
}
for (size_t i = axis + 1; i < a->shape.size(); ++i) {
out_shape.push_back(a->shape[i]);
}
if (mode == "clip") {
if (batch_dims_ == 0) {
return compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> indices_position;
for (size_t j = axis; j < static_cast<size_t>(axis + indices_len); ++j) {
indices_position.push_back(out_index[j]);
}
Array<PrimExpr> real_indices;
for (size_t j = 0; j < static_cast<size_t>(axis); ++j) {
real_indices.push_back(out_index[j]);
}
auto idx = tvm::min(tvm::max(0, indices(indices_position)), axis_dim - 1);
real_indices.push_back(idx);
for (size_t j = axis + indices_len; j < out_index.size(); ++j) {
real_indices.push_back(out_index[j]);
}
return a(real_indices);
},
name, tag);
} else {
return compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> indices_position;
for (size_t j = 0; j < static_cast<size_t>(batch_dims_); ++j) {
indices_position.push_back(out_index[j]);
}
for (size_t j = axis; j < static_cast<size_t>(axis + indices_len - batch_dims_); ++j) {
indices_position.push_back(out_index[j]);
}
Array<PrimExpr> real_indices;
for (size_t j = 0; j < static_cast<size_t>(axis); ++j) {
real_indices.push_back(out_index[j]);
}
auto idx = tvm::min(tvm::max(0, indices(indices_position)), axis_dim - 1);
real_indices.push_back(idx);
for (size_t j = axis + indices_len - batch_dims_; j < out_index.size(); ++j) {
real_indices.push_back(out_index[j]);
}
return a(real_indices);
},
name, tag);
}
} else if (mode == "fast") {
LOG(WARNING) << "Fast mode segfaults when there are out-of-bounds indices. "
"Make sure input indices are in bound";
return compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> indices_position;
for (size_t j = axis; j < static_cast<size_t>(axis + indices_len); ++j) {
indices_position.push_back(out_index[j]);
}
Array<PrimExpr> real_indices;
for (size_t j = 0; j < static_cast<size_t>(axis); ++j) {
real_indices.push_back(out_index[j]);
}
real_indices.push_back(indices(indices_position));
for (size_t j = axis + indices_len; j < out_index.size(); ++j) {
real_indices.push_back(out_index[j]);
}
return a(real_indices);
},
name, tag);
} else { // mode == "wrap"
return compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> indices_position;
for (size_t j = axis; j < static_cast<size_t>(axis + indices_len); ++j) {
indices_position.push_back(out_index[j]);
}
Array<PrimExpr> real_indices;
for (size_t j = 0; j < static_cast<size_t>(axis); ++j) {
real_indices.push_back(out_index[j]);
}
auto idx = truncmod(truncmod(indices(indices_position), axis_dim) + axis_dim, axis_dim);
real_indices.push_back(idx);
for (size_t j = axis + indices_len; j < out_index.size(); ++j) {
real_indices.push_back(out_index[j]);
}
return a(real_indices);
},
name, tag);
}
}
/*!
* \brief Return the elements, either from x or y, depending on the condition.
*
* \param condition The condition array.
* \param x First array to be selected.
* \param y Second array to be selected.
* \param name The name of the operation.
* \param tag The tag to mark the operation.
*
* \return A Tensor selected from x or y depending on condition.
*/
inline Tensor where(const Tensor& condition, const Tensor& x, const Tensor& y,
std::string name = "T_where", std::string tag = kBroadcast) {
ICHECK_EQ(x->dtype, y->dtype) << "x and y must have the same dtype: " << x->dtype << " vs "
<< y->dtype;
auto get_out_shape = [&]() {
auto bh1 = detail::BroadcastShape(x->shape, y->shape);
Array<PrimExpr> common_shape1(bh1.common_shape.begin(), bh1.common_shape.end());
auto bh2 = detail::BroadcastShape(condition->shape, common_shape1);
Array<PrimExpr> common_shape2(bh2.common_shape.begin(), bh2.common_shape.end());
return common_shape2;
};
auto oshape = get_out_shape();
auto c_bh = detail::BroadcastShape(condition->shape, oshape);
auto x_bh = detail::BroadcastShape(x->shape, oshape);
auto y_bh = detail::BroadcastShape(y->shape, oshape);
auto select = [&](tvm::Array<tvm::tir::Var> ovars) {
auto c = condition(InputIndexFromBroadcast(ovars, condition, c_bh.vars1, c_bh.all_vars));
auto true_val = x(InputIndexFromBroadcast(ovars, x, x_bh.vars1, x_bh.all_vars));
auto false_val = y(InputIndexFromBroadcast(ovars, y, y_bh.vars1, y_bh.all_vars));
return tvm::tir::Select(c != 0, true_val, false_val);
};
return compute(oshape, select, name, tag);
}
/*!
* \brief Creates an operation to repeat elements of an array
*
* \param x The input tensor
* \param repeats The number of repetitions for each element
* \param axis The axis along which to repeat values (allows
* negative indices as offsets from the last dimension)
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the repeat operation
*/
inline Tensor repeat(const Tensor& x, int repeats, int axis, std::string name = "T_repeat",
std::string tag = kBroadcast) {
int ndim = static_cast<int>(x->shape.size());
ICHECK(-ndim - 1 <= axis && axis <= ndim)
<< "repeat only accepts `axis` in [-data.ndim - 1, data.ndim]"
<< ", but got axis = " << axis << ", and data.ndim = " << ndim;
ICHECK(repeats >= 1) << "repeat only accepts `repeats >= 1`"
<< ", but got repeats = " << repeats;
if (axis < 0) {
// Calculate offset from last dimension
axis += ndim;
}
Array<PrimExpr> new_shape;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) {
new_shape.push_back(x->shape[i]);
}
new_shape.push_back(repeats * x->shape[axis]);
for (size_t i = axis + 1; i < x->shape.size(); ++i) {
new_shape.push_back(x->shape[i]);
}
return compute(
new_shape,
[&](const Array<Var>& indices) {
Array<PrimExpr> idx;
for (size_t i = 0; i < static_cast<size_t>(axis); ++i) {
idx.push_back(indices[i]);
}
idx.push_back(indexdiv(indices[axis], repeats));
for (size_t i = axis + 1; i < indices.size(); ++i) {
idx.push_back(indices[i]);
}
return x(idx);
},
name, tag);
}
/*!
* \brief Creates an operation to tile elements of an array
*
* \param x The input tensor
* \param reps The number of times for repeating the tensor
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the tile operation
*/
inline Tensor tile(const Tensor& x, Array<Integer> reps, std::string name = "T_tile",
std::string tag = kBroadcast) {
size_t ndim = x->shape.size();
size_t rdim = reps.size();
size_t tdim = (ndim > rdim) ? ndim : rdim;
Array<PrimExpr> data_shape;
Array<PrimExpr> reps_shape;
Array<PrimExpr> new_shape;
if (ndim == rdim) {
for (size_t i = 0; i < ndim; ++i) {
data_shape.push_back(x->shape[i]);
reps_shape.push_back(reps[i]);
}
} else if (ndim > rdim) {
for (size_t i = 0; i < ndim; ++i) data_shape.push_back(x->shape[i]);
for (size_t i = 0; i < (ndim - rdim); ++i) reps_shape.push_back(1);
for (size_t i = 0; i < rdim; ++i) reps_shape.push_back(reps[i]);
} else {
for (size_t i = 0; i < (rdim - ndim); ++i) data_shape.push_back(1);
for (size_t i = 0; i < ndim; ++i) data_shape.push_back(x->shape[i]);
for (size_t i = 0; i < rdim; ++i) reps_shape.push_back(reps[i]);
}
for (size_t i = 0; i < tdim; ++i) new_shape.push_back(data_shape[i] * reps_shape[i]);
if (is_empty_shape(new_shape)) {
return compute(
new_shape, [&](const Array<Var>& indices) { return tvm::cast(x->dtype, 0); }, name, tag);
} else {
return compute(
new_shape,
[&](const Array<Var>& indices) {
Array<PrimExpr> idx;
if (ndim >= rdim) {
for (size_t i = 0; i < ndim; ++i) idx.push_back(indexmod(indices[i], x->shape[i]));
} else {
for (size_t i = 0; i < ndim; ++i)
idx.push_back(indexmod(indices[rdim - ndim + i], x->shape[i]));
}
return x(idx);
},
name, tag);
}
}
/*!
* \brief Creates an operation to tile elements of an array
*
* \param x The input tensor
* \param new_shape The shape of the output after tiling
* \param rdim The rank of the reps, provided by caller
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the tile operation
*/
inline Tensor dyn_tile(const Tensor& x, Array<PrimExpr> new_shape, size_t rdim,
std::string name = "T_tile", std::string tag = kBroadcast) {
size_t ndim = x->shape.size();
if (is_empty_shape(new_shape)) {
return compute(
new_shape, [&](const Array<Var>& indices) { return tvm::cast(x->dtype, 0); }, name, tag);
} else {
return compute(
new_shape,
[&](const Array<Var>& indices) {
Array<PrimExpr> idx;
if (ndim >= rdim) {
for (size_t i = 0; i < ndim; ++i) {
idx.push_back(indexmod(indices[i], x->shape[i]));
}
} else {
for (size_t i = 0; i < ndim; ++i) {
idx.push_back(indexmod(indices[rdim - ndim + i], x->shape[i]));
}
}
return x(idx);
},
name, tag);
}
}
/*!
* \brief Gather values along given axis from given indices.
*
* \param data The input data to the operator.
* \param axis The axis along which to index.
* \param indices The indices of values to gather.
* \param name The name of the operation.
* \param tag The tag to mark the operation.
*
* \return A Tensor whose op member is the gather operation
*/
inline Tensor gather(const Tensor& data, int axis, const Tensor& indices,
std::string name = "T_gather", std::string tag = kInjective) {
size_t ndim_d = data->shape.size();
size_t ndim_i = indices->shape.size();
ICHECK_GE(ndim_d, 1) << "Cannot gather from a scalar.";
ICHECK_EQ(ndim_d, ndim_i);
if (axis < 0) {
axis += ndim_d;
}
ICHECK_GE(axis, 0);
ICHECK_LT(axis, ndim_d);
if (indices->shape[axis].as<IntImmNode>()) {
size_t indices_dim_i = static_cast<size_t>(GetConstInt(indices->shape[axis]));
ICHECK_GE(indices_dim_i, 1);
}
ICHECK(indices->dtype.is_int() || indices->dtype.is_uint());
Array<PrimExpr> out_shape;
for (size_t i = 0; i < ndim_i; ++i) {
out_shape.push_back(indices->shape[i]);
}
return compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> indices_position;
for (size_t i = 0; i < ndim_i; ++i) {
indices_position.push_back(out_index[i]);
}
Array<PrimExpr> real_indices;
for (size_t i = 0; i < ndim_i; ++i) {
if (i == static_cast<size_t>(axis)) {
real_indices.push_back(indices(indices_position));
} else {
real_indices.push_back(indices_position[i]);
}
}
return data(real_indices);
},
name, tag);
}
/*!
* \brief Gather elements from a n-dimension array.
*
* \param data The source array.
* \param indices The indices of the values to extract.
* \param batch_dims The number of batch dimensions.
* \param name The name of the operation.
* \param tag The tag to mark the operation.
*
* \return A Tensor whose op member is the gather_nd operation
*/
inline Tensor gather_nd(const Tensor& data, const Tensor& indices, int batch_dims = 0,
std::string name = "T_gather_nd", std::string tag = kInjective) {
size_t ndim_d = data->shape.size();
size_t ndim_i = indices->shape.size();
ICHECK_GE(ndim_i, 1) << "indices tensor must have at least 1 dimensions";
size_t indices_dim0 = static_cast<size_t>(GetConstInt(indices->shape[0]));
ICHECK_LE(indices_dim0, ndim_d) << "dim 0 of indices tensor must be no more "
<< "than dimensions of data tensor";
Array<PrimExpr> out_shape;
for (size_t i = 1; i < ndim_i; ++i) {
out_shape.push_back(indices->shape[i]);
}
for (size_t i = indices_dim0 + batch_dims; i < ndim_d; ++i) {
out_shape.push_back(data->shape[i]);
}
return compute(
out_shape,
[&](const Array<Var>& out_index) {
Array<PrimExpr> indices_position;
indices_position.push_back(0);
for (size_t i = 0; i < ndim_i - 1; ++i) {
indices_position.push_back(out_index[i]);
}
Array<PrimExpr> real_indices;
for (size_t i = 0; i < static_cast<size_t>(batch_dims); ++i) {
real_indices.push_back(out_index[i]);
}
for (size_t i = 0; i < indices_dim0; ++i) {
indices_position.Set(0, make_const(DataType::Int(32), i));
if (indices->dtype.is_int() || indices->dtype.is_uint()) {
real_indices.push_back(indices(indices_position));
} else {
real_indices.push_back(tvm::cast(tvm::DataType::Int(32), indices(indices_position)));
}
}
if (real_indices.size() == ndim_d) {
return data(real_indices);
}
for (size_t i = ndim_i - 1; i < out_index.size(); ++i) {
real_indices.push_back(out_index[i]);
}
return data(real_indices);
},
name, tag);
}
/*!
* \brief Creates an operation that calculates a matrix multiplication
* (row-major notation):
* A(i, k) * B(k, j), if trans_a == trans_b
* the usual transposed combinations, otherwise
*
* \param A The matrix A
* \param B The matrix B
* \param trans_a Is A's layout transposed?
* \param trans_b Is B's layout transposed?
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the matmul operation
*/
inline tvm::te::Tensor matmul(const tvm::te::Tensor& A, const tvm::te::Tensor& B,
bool trans_a = false, bool trans_b = false,
std::string name = "T_matmul", std::string tag = kMatMul) {
tvm::Array<tvm::PrimExpr> output_shape{A->shape[trans_a ? 1 : 0], B->shape[trans_b ? 0 : 1]};
auto k = tvm::te::reduce_axis(tvm::Range{0, A->shape[trans_a ? 0 : 1]}, "k");
auto l = [&](tvm::tir::Var i, tvm::tir::Var j) {
return tvm::sum((trans_a ? A[k][i] : A[i][k]) * (trans_b ? B[j][k] : B[k][j]), {k});
};
return tvm::te::compute(output_shape, l, name, tag);
}
/*!
* \brief A generalization of matrix multiplication to tensors.
*
* \param A The tensor A
* \param B The tensor B
* \param axes The number of the dimensions to reduce over
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor computing the result
*/
inline Tensor tensordot(const Tensor& A, const tvm::te::Tensor& B, int axes = 2,
std::string name = "T_tensordot", std::string tag = kMatMul) {
ICHECK_GE(A->shape.size(), axes);
ICHECK_GE(B->shape.size(), axes);
Array<PrimExpr> output_shape(A->shape.begin(), A->shape.end() + (-axes));
for (auto it = B->shape.begin() + axes; it != B->shape.end(); ++it) output_shape.push_back(*it);
Array<IterVar> iter_vars;
for (int i = 0; i < axes; ++i)
iter_vars.push_back(reduce_axis(Range(0, B->shape[i]), "k" + std::to_string(i)));
auto func = [&A, &B, &iter_vars, axes](const Array<Var>& input_indices) {
Array<PrimExpr> A_indices(input_indices.begin(),
input_indices.begin() + (A->shape.size() - axes));
for (auto& v : iter_vars) A_indices.push_back(v);
Array<PrimExpr> B_indices;
for (auto& v : iter_vars) B_indices.push_back(v);
auto it = input_indices.begin() + (A->shape.size() - axes);
for (; it != input_indices.end(); ++it) B_indices.push_back(*it);
// Some passes don't like reductions with empty axis, so avoid it here
if (iter_vars.empty()) {
return A(A_indices) * B(B_indices);
} else {
return sum(A(A_indices) * B(B_indices), iter_vars);
}
};
return compute(output_shape, func, name, tag);
}
/*!
* \brief A generalization of matrix multiplication to tensors.
*
* \param A The tensor A
* \param B The tensor B
* \param A_axes The indices of the dimensions of tensor A to reduce over
* \param B_axes The indices of the dimensions of tensor B to reduce over
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor computing the result
*/
inline Tensor tensordot(const Tensor& A, const tvm::te::Tensor& B, Array<PrimExpr> A_axes,
Array<PrimExpr> B_axes, std::string name = "T_tensordot",
std::string tag = kMatMul) {
ICHECK_EQ(A_axes.size(), B_axes.size());
auto A_axes_val = GetConstIntValues(A_axes, "A_axes");
auto B_axes_val = GetConstIntValues(B_axes, "B_axes");
Array<PrimExpr> output_shape;
for (unsigned i = 0; i < A->shape.size(); ++i)
if (std::find(A_axes_val.begin(), A_axes_val.end(), i) == A_axes_val.end())
output_shape.push_back(A->shape[i]);
for (unsigned i = 0; i < B->shape.size(); ++i)
if (std::find(B_axes_val.begin(), B_axes_val.end(), i) == B_axes_val.end())
output_shape.push_back(B->shape[i]);
Array<IterVar> iter_vars;
for (unsigned i = 0; i < B_axes_val.size(); ++i)
iter_vars.push_back(reduce_axis(Range(0, B->shape[B_axes_val[i]]), "k" + std::to_string(i)));
auto func = [&A, &B, &iter_vars, A_axes_val, B_axes_val](const Array<Var>& input_indices) {
int idx_input = 0;
Array<PrimExpr> A_indices;
for (unsigned i = 0; i < A->shape.size(); ++i) {
auto axes_pos = std::find(A_axes_val.begin(), A_axes_val.end(), i);
if (axes_pos == A_axes_val.end()) {
A_indices.push_back(input_indices[idx_input++]);
} else {
A_indices.push_back(iter_vars[axes_pos - A_axes_val.begin()]);
}
}
Array<PrimExpr> B_indices;
for (unsigned i = 0; i < B->shape.size(); ++i) {
auto axes_pos = std::find(B_axes_val.begin(), B_axes_val.end(), i);
if (axes_pos == B_axes_val.end()) {
B_indices.push_back(input_indices[idx_input++]);
} else {
B_indices.push_back(iter_vars[axes_pos - B_axes_val.begin()]);
}
}
return sum(A(A_indices) * B(B_indices), iter_vars);
};
return compute(output_shape, func, name, tag);
}
inline Tensor arange(const PrimExpr& start, const PrimExpr& stop, const PrimExpr& step,
DataType dtype, std::string name = "T_arange", std::string tag = kInjective) {
PrimExpr num_elem = tvm::cast(
tvm::DataType::Int(32), tvm::ceil(tvm::cast(tvm::DataType::Float(32), stop - start) / step));
Array<PrimExpr> shape;
return compute(
{num_elem},
[&](const Array<Var>& indices) { return tvm::cast(dtype, start + step * indices[0]); }, name,
tag);
}
/*!
* \brief Produce grids by expanding input over dimensions defined by other inputs
*
* \param inputs The input tensors
* \param indexing The indexing mode, either "xy" or "ij"
* \param name The name of the operation
* \param tag The tag to mark the operation
*
* \return A Tensor whose op member is the meshgrid operation
*/
inline Array<Tensor> meshgrid(const Array<Tensor>& inputs, const std::string& indexing,
std::string name = "T_meshgrid", std::string tag = kInjective) {
const bool cartesian_indexing = indexing == "xy" && inputs.size() >= 2;
Array<PrimExpr> out_shape;
for (size_t i = 0; i < inputs.size(); ++i) {
const int src_index = (cartesian_indexing && i < 2) ? 1 - i : i;
out_shape.push_back(inputs[src_index]->shape.size() == 0 ? 1 : inputs[src_index]->shape[0]);
}
Array<Tensor> result;
for (size_t i = 0; i < inputs.size(); ++i) {
result.push_back(compute(
out_shape,
[&](const Array<Var>& indices) {
const int src_index = (cartesian_indexing && i < 2) ? 1 - i : i;
auto ndim = inputs[i]->GetShape().size();
Array<PrimExpr> real_indices = {};
if (ndim > 0) {
real_indices = {indices[src_index]};
}
return inputs[i](real_indices);
},
name, tag));
}
return result;
}
/*!
* \brief Transform the layout according to \p src_layout and \p dst_layout
* \param src the source input.
* \param src_layout the source layout.
* \param dst_layout the destination layout.
* \param name output tensor name.
* \param tag output tensor tag.
* \return A tensor with shape in \p dst_layout
*/
inline Tensor layout_transform(const Tensor& src, const std::string& src_layout,
const std::string& dst_layout,
const std::string name = "T_layout_trans",
const std::string tag = kInjective) {
Layout src_layout_struct(src_layout);
Layout dst_layout_struct(dst_layout);
if (src_layout_struct.Equals(dst_layout_struct)) {
return src;
}
ICHECK(src_layout_struct.defined() && dst_layout_struct.defined())
<< "cannot convert from/to undefined layout";
auto layout_converter = tir::BijectiveLayout(src_layout_struct, dst_layout_struct);
ICHECK(layout_converter.defined())
<< "cannot convert from " << src_layout << " to " << dst_layout;
Array<PrimExpr> dst_shape = layout_converter.ForwardShape(src->shape);
return compute(
dst_shape,
[&](const Array<Var>& dst_indices) {
Array<PrimExpr> dst_indices_expr(dst_indices.begin(), dst_indices.end());
Array<PrimExpr> src_indices = layout_converter.BackwardIndex(dst_indices_expr);
PrimExpr in_range = PrimExpr(1) > PrimExpr(0); // init with dtype=bool and value=true
for (size_t i = 0; i < src.ndim(); ++i) {
in_range = in_range && (src_indices[i] < src->shape[i]);
}
return if_then_else(in_range, src(src_indices), tvm::cast(src->dtype, PrimExpr(0)));
},
name, tag);
}
/*! \brief Utility function for auto_scheduler_layout_transform */
inline void parse_auto_scheduler_layout(const String& layout, Array<PrimExpr>* shape,
std::vector<std::string>* axes) {
int32_t factor = 0;
std::string axis = "";
for (char c : std::string(layout)) {
if (c >= 'A' && c <= 'z') {
axis += c;
if (factor != 0) {
shape->push_back(factor);
factor = 0;
}
} else if (c >= '0' && c <= '9') {
factor = factor * 10 + c - '0';
if (!axis.empty()) {
axes->push_back(axis);
axis = "";
}
} else {
LOG(FATAL) << "Invalid layout " << layout;
}
}
if (!axis.empty()) {
axes->push_back(axis);
}
}
/*!
* \brief Transform the auto-scheduler generated layout according to
* \p src_layout and \p dst_layout
* \param src the source input.
* \param src_layout the source layout.
* \param dst_layout the destination layout.
* \param name output tensor name.
* \param tag output tensor tag.
* \return A tensor with shape in \p dst_layout
*/
inline Tensor auto_scheduler_layout_transform(const Tensor& src, const String& src_layout,
const String& dst_layout,
const String name = "T_auto_scheduler_layout_trans",
const String tag = kInjective) {
Array<PrimExpr> src_shape;
std::vector<std::string> src_axes;
Array<PrimExpr> dst_shape;
std::vector<std::string> dst_axes;
parse_auto_scheduler_layout(src_layout, &src_shape, &src_axes);
parse_auto_scheduler_layout(dst_layout, &dst_shape, &dst_axes);
return compute(
dst_shape,
[&](const Array<Var>& dst_indices) {
Array<PrimExpr> dst_indices_expr(dst_indices.begin(), dst_indices.end());
Array<PrimExpr> src_indices;
for (const std::string& src_axis : src_axes) {
PrimExpr src_index = 0;
CHECK_EQ(dst_indices_expr.size(), dst_axes.size());
for (size_t i = 0; i < dst_axes.size(); ++i) {
if (dst_axes[i] == src_axis) {
src_index = src_index * dst_shape[i] + dst_indices_expr[i];
}
}
src_indices.push_back(src_index);
}
return src(src_indices);
},
name, tag);
}
/*!
* \brief Transform the meta-schedule generated layout according to TIR's IndexMap
* \param src the source input.
* \param index_map The TIR IndexMap
* \param name output tensor name.
* \param tag output tensor tag.
* \return A tensor. The layout transformation method
* \note Example:
*
* For the indexing pattern below:
*
* for i in range(32):
* for j in range(64):
* load A[
* i / 16 * 4 + j / 16,
* i % 16 * 16 + j % 16,
* ]
*
* The corresponding indexing pattern in TIR is:
*
* A[i, j] => A'[i / 4, j / 16, i % 4, j % 16]
*
* which converts the pattern to:
*
* for i in range(32):
* for j in range(64):
* load A'[
* i / 16 + j / 64,
* i % 16,
* j % 64 / 16,
* j % 16,
* ]
*
* In this case, the transformation pattern is:
* A'[a, b, c, d] = A[a * 4 + c, b * 16 + d]
*/
inline Tensor meta_schedule_layout_transform(const Tensor& src, const tir::IndexMap& index_map,
const String name = "T_meta_schedule_layout_trans",
const String tag = kInjective) {
Array<Range> iter_domain;
iter_domain.reserve(src->shape.size());
for (const PrimExpr& e : src->shape) {
iter_domain.push_back(Range::FromMinExtent(make_zero(e->dtype), e));
}
Array<PrimExpr> post_transform_shape = index_map->MapShape(src->shape);
return compute(
post_transform_shape,
[src, inv = index_map.Inverse(iter_domain)](const Array<Var>& indices) -> PrimExpr {
return src(inv->MapIndices(Array<PrimExpr>{indices.begin(), indices.end()}));
},
name, tag);
}
/*!
* \brief Get the shape of input tensor.
* \param src the input tensor.
* \param dtype the type of the elements in the tensor.
* \param name output tensor name.
* \param tag output tensor tag.
* \return Tensor of input shape.
*/
inline Tensor shape(const Tensor& src, DataType dtype, const std::string name = "T_shape",
const std::string tag = kInjective) {
int ndim = static_cast<int>(src->shape.size());
Array<PrimExpr> out_shape{ndim};
return compute(
out_shape,
[&](const Array<Var>& indices) {
auto idx = indices[0];
PrimExpr ret = 0;
for (int i = 0; i < ndim; ++i) {
ret = tvm::if_then_else(idx == i, src->shape[i], ret);
}
return tvm::cast(dtype, ret);
},
name, tag);
}
/*!
* \brief Get the size of input tensor.
* \param src the input tensor.
* \param dtype the type of the elements in the tensor.
* \param name output tensor name.
* \param tag output tensor tag.
* \return Tensor of input shape.
*/
inline Tensor ndarray_size(const Tensor& src, const DataType& dtype,
const std::string& name = "ndarray_size",
const std::string& tag = kInjective) {
int ndim = static_cast<int>(src->shape.size());
Array<PrimExpr> out_ndarray_size = {};
return compute(
out_ndarray_size,
[&](const Array<Var>& indices) {
PrimExpr ret = 1;
for (int i = 0; i < ndim; ++i) {
ret *= src->shape[i];
}
return tvm::cast(dtype, ret);
},
name, tag);
}
/*!
* \brief Returns a one-hot tensor where the locations repsented by indices take value on_value,
other locations take value off_value.
* \param indices locations to set to on_value.
* \param on_value value that locations represented by indices take on.
* \param off_value value that other locations take on.
* \param depth depth of the one-hot dimension.
* \param axis axis to fill.
* \param dtype data type of the output tensor.
* \param oshape shape of the output tensor.
* \param name output tensor name.
* \param tag output tensor tag.
* \return one-hot tensor.
*/
inline Tensor one_hot(const Tensor& indices, const PrimExpr on_value, const PrimExpr off_value,
int depth, int axis, const DataType& dtype,
Array<PrimExpr> oshape = Array<PrimExpr>(),
const std::string name = "T_one_hot", const std::string tag = kInjective) {
int true_axis = (axis == -1) ? indices->shape.size() : axis;
if (oshape.size() == 0) {
int ndim = indices->shape.size() + 1;
int indices_index = 0;
for (int i = 0; i < ndim; i++) {
if (i == true_axis) {
oshape.push_back(Integer(depth));
} else {
oshape.push_back(indices->shape[indices_index++]);
}
}
}
PrimExpr on_value_cast = cast(dtype, on_value);
PrimExpr off_value_cast = cast(dtype, off_value);
return compute(
oshape,
[&](const Array<Var>& iter_vars) {
Array<Var> indices_indices;
for (size_t i = 0; i < iter_vars.size(); i++) {
if (static_cast<int>(i) == true_axis) {
continue;
}
indices_indices.push_back(iter_vars[i]);
}
auto idx = iter_vars[true_axis];
return tir::Select(indices(indices_indices) == idx, on_value_cast, off_value_cast);
},
name, tag);
}
/*!
* \brief Get a dense tensor.
* \param sparse_indices sparse_indices[i] contains sparse_values[i] will be placed.
* \param output_shape is the shape of the dense output tensor .
* \param sparse_values is a 0-D or 1-D tensor. Values for each row of sparse_indices.
* \param default_value is a 0-D tensor. Defaults to zero.
* \param name output tensor name.
* \param tag output tensor tag.
* \return Tensor of output_shape.
*/
inline Tensor sparse_to_dense(const Tensor& sparse_indices, const Array<PrimExpr>& output_shape,
const Tensor& sparse_values, const PrimExpr& default_value,
const std::string name = "T_sparse_to_dense",
const std::string tag = kInjective) {
ICHECK(sparse_indices->dtype.is_int()) << "sparse_indices only accepts integer values";
ICHECK_LE(sparse_indices->shape.size(), 3)
<< "sparse_indices tensor should be 0D, 1D, or 2D only";
ICHECK_LE(sparse_values->shape.size(), 2) << "sparse_values tensor should be 0D or 1D only";
const auto rank_sparse_indices = static_cast<int>(sparse_indices->shape.size());
Array<PrimExpr> oshape;
for (auto l : output_shape) {
oshape.push_back(l);
}
return compute(
oshape,
[&](const Array<Var>& indices) {
PrimExpr ret = default_value;
if (0 == rank_sparse_indices) {
ret = if_then_else(indices[0] == sparse_indices(), sparse_values(), ret);
} else if (1 == rank_sparse_indices) {
for (int j = 0; j < GetConstInt(sparse_indices->shape[0]); j++) {
ret = if_then_else(indices[0] == sparse_indices[j], sparse_values[j], ret);
}
} else {
for (int j = 0; j < GetConstInt(sparse_indices->shape[0]); j++) {
PrimExpr aggregate_condition;
for (int k = 0; k < GetConstInt(sparse_indices->shape[1]); k++) {
PrimExpr comparision = indices[k] == sparse_indices[j][k];
aggregate_condition = 0 == k ? comparision : aggregate_condition && comparision;
}
ret = if_then_else(aggregate_condition, sparse_values[j], ret);
}
}
return ret;
},
name, tag);
}
/*!
* \brief Returns a tensor with the diagonal of input tensor replaced with the provided diagonals.
* \param input input tensor.
* \param diagonal values to be filled in the diagonals.
* \param k1 lower limit (included) of the range of diagonals.
* \param k2 upper limit (included) of the range of diagonals.
* \param super_diag_right_align bool, true iff super-diagonal is right aligned (left-padded).
* \param sub_diag_right_align bool, true iff sub-diagonal is right aligned (left-padded).
* \param name output tensor name.
* \param tag output tensor tag.
* \return new tensor with given diagonal values.
*/
inline Tensor matrix_set_diag(const Tensor& input, const Tensor& diagonal, int k1, int k2,
bool super_diag_right_align, bool sub_diag_right_align,
const std::string name = "T_matrix_set_diag",
const std::string tag = kInjective) {
size_t ndim = input->shape.size() - 1;
bool only_one_diagonal = k1 == k2;
return compute(
input->shape,
[&](const Array<Var>& iter_vars) {
auto get_diag = [&]() {
Array<PrimExpr> diagonal_indices;
PrimExpr k, offset = 0;
for (size_t i = 0; i < ndim - 1; i++) {
diagonal_indices.push_back(iter_vars[i]);
}
if (only_one_diagonal) {
k = k1;
} else {
// Determining which diagonal/sub-diagonal/super-diagonal it is
k = iter_vars[ndim] - iter_vars[ndim - 1];
diagonal_indices.push_back(k2 - k);
// Calculating the offset in diagonal tensor for this diagonal
auto get_offset = [&](PrimExpr M, PrimExpr N) {
// offset = max_diagonal_length - diagonal_length
return diagonal->shape[diagonal->shape.size() - 1] - if_then_else(M < N, M, N);
};
offset = if_then_else(
k >= 0,
super_diag_right_align ? get_offset(input->shape[ndim] - k, input->shape[ndim - 1])
: 0,
sub_diag_right_align ? get_offset(input->shape[ndim], input->shape[ndim - 1] + k)
: 0);
}
diagonal_indices.push_back(if_then_else(k >= 0, iter_vars[ndim - 1], iter_vars[ndim]) +
offset);
return diagonal(diagonal_indices);
};
return if_then_else((PrimExpr)iter_vars[ndim] - iter_vars[ndim - 1] >= k1,
if_then_else((PrimExpr)iter_vars[ndim] - iter_vars[ndim - 1] <= k2,
get_diag(), input(iter_vars)),
input(iter_vars));
},
name, tag);
}
/*!
* \brief Numpy style advanced indexing with tensor.
* \param data is input data.
* \param indices is list of indexing tensors.
* \param name output tensor name.
* \param tag output tensor tag.
* \return Output tensor.
*/
inline Tensor adv_index(const Tensor& data, const Array<Tensor>& indices,
const std::string name = "advanced_index",
const std::string tag = kInjective) {
ICHECK_LE(indices.size(), data->shape.size()) << "too many indices for data!";
Array<PrimExpr> oshape;
Array<PrimExpr> broadcast_shape;
Array<Tensor> bindices;
broadcast_shape = indices[0]->shape;
for (size_t i = 1; i < indices.size(); ++i) {
auto bh = detail::BroadcastShape(broadcast_shape, indices[i]->shape);
broadcast_shape = Array<PrimExpr>(bh.common_shape.begin(), bh.common_shape.end());
}
if (indices.size() == 1) {
// quick path
bindices = indices;
} else {
// Do broadcast for indices
for (size_t i = 0; i < indices.size(); ++i) {
bindices.push_back(broadcast_to(indices[i], broadcast_shape));
}
}
for (const auto& dim : broadcast_shape) {
oshape.push_back(dim);
}
for (size_t i = indices.size(); i < data->shape.size(); ++i) {
oshape.push_back(data->shape[i]);
}
return compute(
oshape,
[&](const Array<Var>& iter_var) {
Array<PrimExpr> tensor_indices;
for (size_t i = 0; i < broadcast_shape.size(); ++i) {
tensor_indices.push_back(iter_var[i]);
}
Array<PrimExpr> real_indices;
for (size_t i = 0; i < bindices.size(); ++i) {
real_indices.push_back(bindices[i](tensor_indices));
}
for (size_t i = broadcast_shape.size(); i < iter_var.size(); ++i) {
real_indices.push_back(iter_var[i]);
}
return data(real_indices);
},
name, tag);
}
} // namespace topi
} // namespace tvm
#endif // TVM_TOPI_TRANSFORM_H_