1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
|
/* Copyright 2017 The TensorFlow Authors. All Rights Reserved.
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.
==============================================================================*/
// XLA-specific Concat Ops.
#include <limits>
#include <vector>
#include "tensorflow/compiler/tf2xla/type_util.h"
#include "tensorflow/compiler/tf2xla/xla_helpers.h"
#include "tensorflow/compiler/tf2xla/xla_op_kernel.h"
#include "tensorflow/compiler/tf2xla/xla_op_registry.h"
#include "tensorflow/compiler/xla/client/xla_builder.h"
#include "tensorflow/compiler/xla/literal_util.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/framework/register_types.h"
#include "tensorflow/core/framework/tensor.h"
#include "tensorflow/core/framework/tensor_types.h"
#include "tensorflow/core/framework/types.h"
#include "tensorflow/core/kernels/bounds_check.h"
#include "tensorflow/core/kernels/concat_lib.h"
#include "tensorflow/core/lib/core/status.h"
#include "tensorflow/core/platform/types.h"
namespace tensorflow {
namespace {
// Used to determine the number of Tensors allowed in a Concat op to prevent
// going over the max gpu parameter memory size. This is an issue because concat
// is variadic and can have an unlimited number of arguments when called.
// Concat ops with more Tensors than this will be split into multiple concat
// ops.
//
// TODO(b/112613927): Remove the logic here and put it properly in an HLO pass
// along with boxing large numbers of parameters.
constexpr int64 kMaxConcatArgsPerOp = 500;
// --------------------------------------------------------------------------
class ConcatBaseOp : public XlaOpKernel {
public:
ConcatBaseOp(OpKernelConstruction* c, int axis_index)
: XlaOpKernel(c), axis_index_(axis_index) {}
void Compile(XlaOpKernelContext* ctx) override {
const TensorShape concat_dim_tensor_shape = ctx->InputShape(axis_index_);
OP_REQUIRES(
ctx, IsLegacyScalar(concat_dim_tensor_shape),
errors::InvalidArgument(
"Concat dim tensor should be a scalar integer, but got shape ",
concat_dim_tensor_shape.DebugString()));
xla::Literal literal;
OP_REQUIRES_OK(ctx, ctx->ConstantInput(axis_index_, &literal));
// TODO(annarev): add a helper to support int64 input.
const int32 concat_dim = literal.Get<int>({});
std::vector<xla::XlaOp> values;
std::vector<TensorShape> shapes;
OP_REQUIRES_OK(ctx, ctx->InputList("values", &values, &shapes));
const int N = values.size();
const int input_dims = shapes[0].dims();
const TensorShape& input_shape = shapes[0];
int32 axis = concat_dim < 0 ? concat_dim + input_dims : concat_dim;
OP_REQUIRES(ctx,
(0 <= axis && axis < input_dims) ||
(allow_legacy_scalars() && concat_dim == 0),
errors::InvalidArgument(
"ConcatOp : Expected concatenating dimensions in the range "
"[",
-input_dims, ", ", input_dims, "), but got ", concat_dim));
// Make a vector holding the XlaOp for each of the inputs that has non-zero
// elements.
std::vector<xla::XlaOp> input_data;
std::vector<xla::XlaOp> partial_concats;
int output_concat_dim = 0;
const bool input_is_scalar = IsLegacyScalar(input_shape);
for (int i = 0; i < N; ++i) {
xla::XlaOp handle = values[i];
const TensorShape& in_shape = shapes[i];
const bool in_is_scalar = IsLegacyScalar(in_shape);
OP_REQUIRES(
ctx,
in_shape.dims() == input_dims || (input_is_scalar && in_is_scalar),
errors::InvalidArgument(
"ConcatOp : Ranks of all input tensors should match: shape[0] = ",
input_shape.DebugString(), " vs. shape[", i,
"] = ", in_shape.DebugString()));
if (in_shape.dims() == 0) {
// Inputs that come in as scalars must be reshaped to 1-vectors.
input_data.push_back(xla::Reshape(handle, {1}));
} else {
input_data.push_back(handle);
}
output_concat_dim += in_shape.dims() > 0 ? in_shape.dim_size(axis) : 1;
// Concat is associative, so it can be split into many operations when too
// many arguments are in a single op. This is a temporary workaround for
// b/112613927 where too many parameters in an XlaLaunchOp later result in
// too many parameters to a single GPU kernel.
if (i && i % kMaxConcatArgsPerOp == 0) {
partial_concats.push_back(
xla::ConcatInDim(ctx->builder(), input_data, axis));
input_data.clear();
}
}
// Add any inputs that have not been put into another concat yet.
partial_concats.insert(partial_concats.end(), input_data.begin(),
input_data.end());
VLOG(1) << "Concat dim " << concat_dim << " equivalent to " << axis;
// Don't add an additional "identity" concatenate for better readibility of
// IR.
if (partial_concats.size() == 1) {
ctx->SetOutput(0, partial_concats.front());
} else {
ctx->SetOutput(0,
xla::ConcatInDim(ctx->builder(), partial_concats, axis));
}
}
private:
int axis_index_;
};
class ConcatOp : public ConcatBaseOp {
public:
explicit ConcatOp(OpKernelConstruction* c)
: ConcatBaseOp(c, /* axis_index */ 0) {}
};
// ConcatV2 operation is the same as Concat except 'concat_dim'
// is the last input instead of the first and renamed to 'axis'.
class ConcatV2Op : public ConcatBaseOp {
public:
explicit ConcatV2Op(OpKernelConstruction* c)
: ConcatBaseOp(c, /* axis_index */ c->num_inputs() - 1) {}
};
REGISTER_XLA_OP(Name("Concat").CompileTimeConstInput("concat_dim"), ConcatOp);
REGISTER_XLA_OP(Name("ConcatV2")
.TypeConstraint("Tidx", DT_INT32)
.CompileTimeConstInput("axis"),
ConcatV2Op);
class ConcatOffsetOp : public XlaOpKernel {
public:
explicit ConcatOffsetOp(OpKernelConstruction* ctx) : XlaOpKernel(ctx) {}
void Compile(XlaOpKernelContext* ctx) override {
const TensorShape concat_dim_shape = ctx->InputShape(0);
OP_REQUIRES(
ctx, IsLegacyScalar(concat_dim_shape),
errors::InvalidArgument(
"Concat dim tensor should be a scalar integer, but got shape ",
concat_dim_shape.DebugString()));
for (int i = 1; i < ctx->num_inputs(); ++i) {
OP_REQUIRES(ctx, TensorShapeUtils::IsVector(ctx->InputShape(i)),
errors::InvalidArgument("input ", i,
" should be a vector, but got shape ",
ctx->InputShape(i).DebugString()));
}
// Suppose a Concat() op needs to Concatenate N tensors, each of
// which has the same number of dimensions. Their shapes match
// except the concat dimension.
//
// E.g., say, we want to concatenate 3 tensors in the 2nd
// dimension, and their shapes are:
//
// [2, 2, 5, 7]
// [2, 3, 5, 7]
// [2, 4, 5, 7]
//
// Here, N=3, cdim=1, dims=4. The concatenated tensor has shape
// [2,9,5,7]. We will compute the cumulative sum along the 2nd
// dimension to figure out each input's offset in the concatenated
// output:
// [0, 0, 0, 0]
// [0, 2, 0, 0]
// [0, 5, 0, 0]
const int32 N = ctx->num_inputs() - 1;
const TensorShape inp0_shape = ctx->InputShape(1);
xla::Literal inp0_literal;
OP_REQUIRES_OK(ctx, ctx->ConstantInput(1, &inp0_literal));
const int64 dims = inp0_shape.num_elements();
xla::Literal concat_dim_literal;
OP_REQUIRES_OK(ctx, ctx->ConstantInput(0, &concat_dim_literal));
const int64 cdim = concat_dim_literal.Get<int>({});
VLOG(1) << "ConcatOffset " << cdim << "," << dims;
int32 axis = cdim < 0 ? cdim + dims : cdim;
OP_REQUIRES(ctx, FastBoundsCheck(axis, dims),
errors::InvalidArgument("Concat dim is out of range: ", axis,
" vs. ", dims));
int32 offset = 0;
for (int i = 0; i < N; ++i) {
const TensorShape inp_shape = ctx->InputShape(1 + i);
OP_REQUIRES(ctx, dims == inp_shape.num_elements(),
errors::InvalidArgument("input ", i, " should contain ", dims,
" elements, but got ",
inp_shape.num_elements()));
xla::Literal inp_literal;
OP_REQUIRES_OK(ctx, ctx->ConstantInput(1 + i, &inp_literal));
Tensor out_constant(DT_INT32, TensorShape({dims}));
auto out_vec = out_constant.vec<int32>();
for (int64 j = 0; j < dims; ++j) {
if (j == axis) {
out_vec(j) = offset;
offset += inp_literal.Get<int>({j});
} else {
const int32 inp0_element = inp0_literal.Get<int>({j});
const int32 inp_element = inp_literal.Get<int>({j});
OP_REQUIRES(ctx, (inp0_element == inp_element),
errors::InvalidArgument("input[", i, ",", j,
"] mismatch: ", inp0_element,
" vs. ", inp_element));
out_vec(j) = 0;
}
}
ctx->SetConstantOutput(i, out_constant);
}
}
};
REGISTER_XLA_OP(Name("ConcatOffset")
.CompileTimeConstInput("concat_dim")
.CompileTimeConstInput("shape"),
ConcatOffsetOp);
} // namespace
} // namespace tensorflow
|
