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
|
// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include "dense_multi_matmul_function.h"
#include <vespa/vespalib/objects/objectvisitor.h>
#include <vespa/eval/eval/value.h>
#include <vespa/eval/eval/operation.h>
#include <cassert>
#include <cblas.h>
namespace vespalib::eval {
using namespace tensor_function;
using namespace operation;
using Dim = ValueType::Dimension;
using DimList = std::vector<Dim>;
namespace {
void my_cblas_double_multi_matmul_op(InterpretedFunction::State &state, uint64_t param) {
using CT = double;
const DenseMultiMatMulFunction &self = unwrap_param<DenseMultiMatMulFunction>(param);
size_t lhs_block_size = self.lhs_size() * self.common_size();
size_t rhs_block_size = self.rhs_size() * self.common_size();
size_t dst_block_size = self.lhs_size() * self.rhs_size();
size_t num_blocks = self.matmul_cnt();
const CT *lhs = state.peek(1).cells().typify<CT>().cbegin();
const CT *rhs = state.peek(0).cells().typify<CT>().cbegin();
auto dst_cells = state.stash.create_array<CT>(dst_block_size * num_blocks);
CT *dst = dst_cells.begin();
for (size_t i = 0; i < num_blocks; ++i, lhs += lhs_block_size, rhs += rhs_block_size, dst += dst_block_size) {
cblas_dgemm(CblasRowMajor, self.lhs_common_inner() ? CblasNoTrans : CblasTrans, self.rhs_common_inner() ? CblasTrans : CblasNoTrans,
self.lhs_size(), self.rhs_size(), self.common_size(), 1.0,
lhs, self.lhs_common_inner() ? self.common_size() : self.lhs_size(),
rhs, self.rhs_common_inner() ? self.common_size() : self.rhs_size(),
0.0, dst, self.rhs_size());
}
state.pop_pop_push(state.stash.create<DenseValueView>(self.result_type(), TypedCells(dst_cells)));
}
void my_cblas_float_multi_matmul_op(InterpretedFunction::State &state, uint64_t param) {
using CT = float;
const DenseMultiMatMulFunction &self = unwrap_param<DenseMultiMatMulFunction>(param);
size_t lhs_block_size = self.lhs_size() * self.common_size();
size_t rhs_block_size = self.rhs_size() * self.common_size();
size_t dst_block_size = self.lhs_size() * self.rhs_size();
size_t num_blocks = self.matmul_cnt();
const CT *lhs = state.peek(1).cells().typify<CT>().cbegin();
const CT *rhs = state.peek(0).cells().typify<CT>().cbegin();
auto dst_cells = state.stash.create_array<CT>(dst_block_size * num_blocks);
CT *dst = dst_cells.begin();
for (size_t i = 0; i < num_blocks; ++i, lhs += lhs_block_size, rhs += rhs_block_size, dst += dst_block_size) {
cblas_sgemm(CblasRowMajor, self.lhs_common_inner() ? CblasNoTrans : CblasTrans, self.rhs_common_inner() ? CblasTrans : CblasNoTrans,
self.lhs_size(), self.rhs_size(), self.common_size(), 1.0,
lhs, self.lhs_common_inner() ? self.common_size() : self.lhs_size(),
rhs, self.rhs_common_inner() ? self.common_size() : self.rhs_size(),
0.0, dst, self.rhs_size());
}
state.pop_pop_push(state.stash.create<DenseValueView>(self.result_type(), TypedCells(dst_cells)));
}
InterpretedFunction::op_function my_select(CellType cell_type) {
if (cell_type == CellType::DOUBLE) {
return my_cblas_double_multi_matmul_op;
}
if (cell_type == CellType::FLOAT) {
return my_cblas_float_multi_matmul_op;
}
abort();
}
struct CommonDim {
bool valid;
bool inner;
CommonDim(const DimList &list, const vespalib::string &dim)
: valid(true), inner(false)
{
if (list[list.size() - 1].name == dim) {
inner = true;
} else if (list[list.size() - 2].name != dim) {
valid = false;
}
}
const Dim &get(const DimList &dims) const {
return dims[dims.size() - (inner ? 1 : 2)];
}
const Dim &inv(const DimList &dims) const {
return dims[dims.size() - (inner ? 2 : 1)];
}
};
// Currently, non-matmul dimensions are required to be identical
// (after trivial dimensions are ignored). This restriction is added
// to reduce complexity and might be removed in the future if/when a
// relevant use-case arises.
struct DimPrefix {
bool valid;
size_t size;
DimPrefix(const DimList &a, const DimList &b)
: valid(true), size(1)
{
if (a.size() == b.size()) {
for (size_t i = 0; i < (a.size() - 2); ++i) {
if (a[i] == b[i]) {
size *= a[i].size;
} else {
valid = false;
}
}
} else {
valid = false;
}
}
};
bool check_input_type(const ValueType &type, const DimList &relevant) {
return (type.is_dense() &&
(relevant.size() >= 2) &&
((type.cell_type() == CellType::FLOAT) || (type.cell_type() == CellType::DOUBLE)));
}
bool is_multi_matmul(const ValueType &a, const ValueType &b, const vespalib::string &reduce_dim) {
auto dims_a = a.nontrivial_indexed_dimensions();
auto dims_b = b.nontrivial_indexed_dimensions();
if (check_input_type(a, dims_a) && check_input_type(b, dims_b) && (a.cell_type() == b.cell_type())) {
CommonDim cd_a(dims_a, reduce_dim);
CommonDim cd_b(dims_b, reduce_dim);
DimPrefix prefix(dims_a, dims_b);
return (cd_a.valid && cd_b.valid && prefix.valid &&
(b.dimension_index(cd_a.inv(dims_a).name) == Dim::npos) &&
(a.dimension_index(cd_b.inv(dims_b).name) == Dim::npos));
}
return false;
}
const TensorFunction &create_multi_matmul(const TensorFunction &a, const TensorFunction &b,
const vespalib::string &reduce_dim, const ValueType &result_type, Stash &stash)
{
auto dims_a = a.result_type().nontrivial_indexed_dimensions();
auto dims_b = b.result_type().nontrivial_indexed_dimensions();
CommonDim cd_a(dims_a, reduce_dim);
CommonDim cd_b(dims_b, reduce_dim);
DimPrefix prefix(dims_a, dims_b);
size_t a_size = cd_a.inv(dims_a).size;
size_t b_size = cd_b.inv(dims_b).size;
size_t common_size = cd_a.get(dims_a).size;
bool a_is_lhs = (cd_a.inv(dims_a).name < cd_b.inv(dims_b).name);
if (a_is_lhs) {
return stash.create<DenseMultiMatMulFunction>(result_type, a, b, a_size, common_size, b_size, prefix.size, cd_a.inner, cd_b.inner);
} else {
return stash.create<DenseMultiMatMulFunction>(result_type, b, a, b_size, common_size, a_size, prefix.size, cd_b.inner, cd_a.inner);
}
}
} // namespace <unnamed>
DenseMultiMatMulFunction::DenseMultiMatMulFunction(const ValueType &result_type,
const TensorFunction &lhs_in,
const TensorFunction &rhs_in,
size_t lhs_size,
size_t common_size,
size_t rhs_size,
size_t matmul_cnt,
bool lhs_common_inner,
bool rhs_common_inner)
: Super(result_type, lhs_in, rhs_in),
_lhs_size(lhs_size),
_common_size(common_size),
_rhs_size(rhs_size),
_matmul_cnt(matmul_cnt),
_lhs_common_inner(lhs_common_inner),
_rhs_common_inner(rhs_common_inner)
{
}
DenseMultiMatMulFunction::~DenseMultiMatMulFunction() = default;
InterpretedFunction::Instruction
DenseMultiMatMulFunction::compile_self(const ValueBuilderFactory &, Stash &) const
{
auto op = my_select(lhs().result_type().cell_type());
return InterpretedFunction::Instruction(op, wrap_param<DenseMultiMatMulFunction>(*this));
}
void
DenseMultiMatMulFunction::visit_self(vespalib::ObjectVisitor &visitor) const
{
Super::visit_self(visitor);
visitor.visitInt("lhs_size", _lhs_size);
visitor.visitInt("common_size", _common_size);
visitor.visitInt("rhs_size", _rhs_size);
visitor.visitInt("matmul_cnt", _matmul_cnt);
visitor.visitBool("lhs_common_inner", _lhs_common_inner);
visitor.visitBool("rhs_common_inner", _rhs_common_inner);
}
const TensorFunction &
DenseMultiMatMulFunction::optimize(const TensorFunction &expr, Stash &stash)
{
auto reduce = as<Reduce>(expr);
if (reduce && (reduce->aggr() == Aggr::SUM) && (reduce->dimensions().size() == 1)) {
auto join = as<Join>(reduce->child());
if (join && (join->function() == Mul::f)) {
const TensorFunction &a = join->lhs();
const TensorFunction &b = join->rhs();
if (is_multi_matmul(a.result_type(), b.result_type(), reduce->dimensions()[0])) {
return create_multi_matmul(a, b, reduce->dimensions()[0], expr.result_type(), stash);
}
}
}
return expr;
}
} // namespace
|
