Developer Guide

Contents

Distributed Processing

This mode assumes that data set is split in
nblocks
blocks across computation nodes.

Algorithm Parameters

The SVD algorithm in the distributed processing mode has the following parameters:
Parameter
Default Value
Description
computeStep
Not applicable
The parameter required to initialize the algorithm. Can be:
  • step1Local
    - the first step, performed on local nodes
  • step2Master
    - the second step, performed on a master node
  • step3Local
    - the final step, performed on local nodes
algorithmFPType
float
The floating-point type that the algorithm uses for intermediate computations. Can be
float
or
double
.
method
defaultDense
Performance-oriented computation method, the only method supported by the algorithm.
leftSingularMatrix
requiredInPackedForm
Specifies whether the matrix of left singular vectors is required. Can be:
  • notRequired
    - the matrix is not required
  • requiredInPackedForm
    - the matrix in the packed format is required.
rightSingularMatrix
requiredInPackedForm
Specifies whether the matrix of right singular vectors is required. Can be:
  • notRequired
    - the matrix is not required
  • requiredInPackedForm
    - the matrix in the packed format is required.
Use the three-step computation schema to compute SVD:

Step 1 - on Local Nodes

SVD Distributed Processing Workflow Step 1
In this step, SVD accepts the input described below. Pass the Input ID as a parameter to the methods that provide input for your algorithm. For more details, see Algorithms.
Input ID
Input
data
Pointer to the
n
i
x
p
numeric table that represents the
i
-th data block on the local node. The input can be an object of any class derived from
NumericTable
.
In this step, SVD calculates the results described below. Pass the Partial Result ID as a parameter to the methods that access the results of your algorithm. For more details, see Algorithms.
Partial Result ID
Result
outputOfStep1ForStep2
A collection that contains numeric tables each with the partial result to transmit to the master node for Step 2. By default, these tables are objects of the
HomogenNumericTable
class, but you can define them as objects of any class derived from
NumericTable
except
PackedSymmetricMatrix
,
PackedTriangularMatrix
, and
CSRNumericTable
.
outputOfStep1ForStep3
A collection that contains numeric tables each with the partial result to keep on the local node for Step 3. By default, these tables are objects of the
HomogenNumericTable
class, but you can define them as objects of any class derived from
NumericTable
except
PackedSymmetricMatrix
,
PackedTriangularMatrix
, and
CSRNumericTable
.

Step 2 - on Master Node

SVD Distributed Processing Workflow Step 2
In this step, SVD accepts the input from each local node described below. Pass the Input ID as a parameter to the methods that provide input for your algorithm. For more details, see Algorithms.
Input ID
Input
inputOfStep2FromStep1
A collection that contains results computed in Step 1 on local nodes (
outputOfStep1ForStep2
). The collection can contain objects of any class derived from
NumericTable
except the
PackedSymmetricMatrix
class and
PackedTriangularMatrix
class with the
lowerPackedTriangularMatrix
layout.
key
A key, a number of type
int
. Keys enable tracking the order in which partial results from Step 1 (
inputOfStep2FromStep1
) come to the master node, so that the partial results computed in Step 2 (
outputOfStep2ForStep3
) can be delivered back to local nodes in exactly the same order.
In this step, SVD calculates the results described below. Pass the Partial Result ID or Result ID as a parameter to the methods that access the results of your algorithm. For more details, see Algorithms.
Partial Result ID
Result
outputOfStep2ForStep3
A collection that contains numeric tables to be split across local nodes to compute left singular vectors. Set to
NULL
if you do not need left singular vectors. By default, these tables are objects of the
HomogenNumericTable
class, but you can define them as objects of any class derived from
NumericTable
except
PackedSymmetricMatrix
,
PackedTriangularMatrix
, and
CSRNumericTable
.
Result ID
Result
singularValues
Pointer to the 1 x
p
numeric table with singular values (the diagonal of the matrix
Σ
). By default, this result is an object of the
HomogenNumericTable
class, but you can define the result as an object of any class derived from
NumericTable
except
PackedSymmetricMatrix
,
PackedTriangularMatrix
, and
CSRNumericTable
.
rightSingularMatrix
Pointer to the
p
x
p
numeric table with right singular vectors (matrix
V
). Pass
NULL
if right singular vectors are not required. By default, this result is an object of the
HomogenNumericTable
class, but you can define the result as an object of any class derived from
NumericTable
except
PackedSymmetricMatrix
,
PackedTriangularMatrix
, and
CSRNumericTable
.

Step 3 - on Local Nodes

SVD Distributed Processing Workflow Step 3
In this step, SVD accepts the input described below. Pass the Input ID as a parameter to the methods that provide input for your algorithm. For more details, see Algorithms.
Input ID
Input
inputOfStep3FromStep1
A collection that contains results computed in Step 1 on local nodes (
outputOfStep1ForStep3
). The collection can contain objects of any class derived from
NumericTable
except the
PackedSymmetricMatrix
and
PackedTriangularMatrix
.
inputOfStep3FromStep2
A collection that contains results computed in Step 2 on local nodes (
outputOfStep2ForStep3
). The collection can contain objects of any class derived from
NumericTable
except
PackedSymmetricMatrix
and
PackedTriangularMatrix
.
In this step, SVD calculates the results described below. Pass the Result ID as a parameter to the methods that access the results of your algorithm. For more details, see Algorithms.
Result ID
Result
leftSingularMatrix
Pointer to the
n
x
p
numeric table with left singular vectors (matrix
U
). Pass
NULL
if left singular vectors are not required. By default, this result is an object of the
HomogenNumericTable
class, but you can define the result as an object of any class derived from
NumericTable
except
PackedSymmetricMatrix
,
PackedTriangularMatrix
, and
CSRNumericTable
.

Product and Performance Information

1

Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice.

Notice revision #20110804