|
Programming on Multi-Core Processors Using TBB APIs
|
Example 1.1
|
Write a TBB program to perform Simple dot operation where a vector element is multiplied with a scaler value.
|
Example 1.2
|
Write a TBB program to perform Multiplication of two array elements and putting array result in resultant array.
|
Example 1.3
|
Advance example program , to do parallel multiple reduction operation at same
time and finding min and max value with in an array.
|
Example 1.4
|
Example program , to do parallel scan operation. The example is implemented using template and a genralized version and can be used for any basic data types. It calculates prefix sum.
|
Example 1.5
|
Example program , to do parallel scan operation. The example is implemented using template and a genralized version and can be used for any basic data types. It calculates prefix sum. With auto-partitioner.
|
Example 1.6
|
Example program , to determine the minimum and maximum values in a vector (array). It also allows user to set user-defined 'Thread Affinity' value
to compare performance for different values of number-of-cores to which thread is binded.
|
|
Description of TBB Programs |
Makefile : To compile the program.
(Download source code ;
Makefile )
|
|
Example 1.1 : Simple dot operation where a vector element is multiplied with a scaler value.
(Download source code ;
VectorDotOp.cpp )
|
- Objective
Simple dot operation where a vector element is multiplied with a scaler value.
- Description
In this program we assign random value in to a array. For this we use parallel_for directive that call callback operator function that execute parally(TBB take care).
- Input
Vector length.
- Output
Dot product of vector and scaler multiplication
|
|
Example 1.2 : Multiplication of two array elements.
(Download source code ;
ArrayMult.cpp )
|
Multiplication of tow array elements and putting array result in resultant array.
- Description
This program is use parallel_for loop that call callback function operator to multiply two one dimension array(vector) and store result in the resultant array
- Input
Length of array .
- Output
resultant array
|
|
Example 1.3 : Example program, to do parallel reduction and finding Min and Max value from a collection
of numbers.
(Download source code ;
ParellReduceMinMax.cpp )
|
- Objective
Example program , to do parallel reduction operation at same
time and finding min and max value with in an array. The example is implemented using template and a genralized version and can be used for any basic
data types.
- Description
This example is implemented with parallel_reduce building block. Where every thread perform reduction operation with in its range
specified by array index. Later in join block all result is consolidated and final reduction result
is collected. This example also find Maximum and minimum value among all array elments.
- Input
Array of integer.
- Output
result of +reduction, *reduction and Max and Min element within
input array
|
|
Example 1.4 : Example program , to do parallel scan operation. The example is implemented using template and a genralized version and can be used for any basic data types. It calculates prefix sum.
(Download source code ;
prefix_sum_parallel_scan.cpp )
|
- Objective
Computation of prefix sum of a randomly generated vector
- Description
This program uses parallel_scan loop that calculates prefix sum by obtaining previous sum of values
- Input
Length of array, grain size.
- Output
Prefix sum, time taken
|
Example 1.5 : Example program , to do parallel scan operation. The example is implemented using template and a genralized version and can be used for any basic data types. It calculates prefix sum. With auto-partitioner.
(Download source code :
prefix_sum_parallel_scan_autopartitioner.cpp )
|
- Objective
Computation of prefix sum of a randomly generated vector
- Description
This program uses parallel_scan loop that calculates prefix sum by obtaining previous sum of values. It makes use of auto-partitioner. Here grain size
automatically determined by the system.
- Input
Length of array .
- Output
Prefix sum, time taken
|
|
Example 1.6 :
Example program , to determine the minimum and maximum values in a vector (array). It also allows user to set user-defined 'Thread Affinity' value
to compare performance for different values of number-of-cores to which thread is binded.
(Download source code :
TBB_find_min_max_from_vector_with_thread_aff.cpp,
Makefile,
README
)
Download source code :
tbb_find_min_max (WinRAR ZIP archive)
|
- Objective
A program to find minimum value and maximum value and their respective indexes from a randomly generated vector of user defined size.
- Description
For a user defined value of vector size, the program generates a float vector
containing random float values. The program then searchs for minimum and
maximum value and determines their respective indices.
The user can also set thread-affinity value to decide no. of processes to which
threads are bound.
- Input
Float Array Size, Intel TBB Grain Size and Thread Affinity Mask
- Output
Thread affinity value (if specified) ,
Minimum Value of the vector & index ,
Maximum Value of the vector & index ,
Time taken (seconds) ,
Time taken (micro seconds)
|
|
| |
|
