Bootstrap¶
StreamPU comes with simple tests to validate its behavior. The later are
always a good way to bootstrap when you want to write your first code with the
DSEL. The source codes of the following tests are located in the
tests/bootstrap folder. Each file corresponds to an executable test.
Simple Chains¶
At some point we have to start somewhere :-). The following graphs are a very
simple chains made from increment/incrementf tasks that simply perform
"\(+1\)" on the data. Each time there are 6 Incrementer (\(t_{[2:7]}\)) so the
final expected values in the Finalizer (\(t_8\)) are equal to the values from
the Initializer (\(t_1\)) "\(+6\)".
increment tasks that have one
input socket and one output socket.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-simple-chain [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-u, --subseq Enable subsequence in the executed sequence [false]
-v, --verbose Enable verbose mode [false]
-h, --help This help [false]
incrementf tasks that have
only one forward socket.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-simple-chain-fwd [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [true]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-u, --subseq Enable subsequence in the executed sequence [false]
-v, --verbose Enable verbose mode [false]
-h, --help This help [false]
increment
and incrementf tasks.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-simple-chain-hybrid [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-v, --verbose Enable verbose mode [false]
-h, --help This help [false]
Looping¶
The looping tests are here to introduce and to validate how to implement simple
control flow. Each of the graph below execute one or multiple loops and in the
innermost loop 6 increment tasks are executed. Switcher modules are used to
create two different paths, one for the loop and the other for the exit (the
Finalizer task here). The output socket of the iterate task (from the
Iterator module) is bound to the input socket 1 of the commute task. The
iterate task controls if the commute should execute the 6 increment tasks or
the end the stream by executing the finalize task.
Each time there are 6 Incrementer in the innermost loop so the final expected
values in the Finalizer are equal to the values from the Initializer (\(t_1\))
"\(+(i \times j\times 6)\)" (\(j = 1\) in the for loop and do-while loop
examples).
for-loop where the condition is evaluated
first in \(t_3\) (a basic loop counter) and the the body of the loop is
executed (\(t_{[5:10]}\)) after. Note that this for-loop example can easily
be extended to a more generic while-loop if the \(t_3\) task is replaced
by an other task that depends on an input data socket.
In the command line, -i 10 indicates that the loop is repeated 10 times.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-for-loop [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-i, --n-loop Number of iterations to perform in the loop [10]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-h, --help This help [false]
do-while loop where the condition \(t_9\) is
evaluated after the body of the loop (\(t_{[3:8]}\)).
In the command line, -i 10 indicates that the loop is repeated 10 times.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-do-while-loop [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-i, --n-loop Number of iterations to perform in the loop [9]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-h, --help This help [false]
Implementation of 2 nested for-loops. -j 2 controls the number of times
the innermost loops is repeated and -i 5 controls the outermost loop.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-nested-loops [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-i, --n-loop-out Number of iterations to perform in the outer loop [5]
-j, --n-loop-in Number of iterations to perform in the inner loop [2]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-h, --help This help [false]
Switch-case¶
The following test implements a switch-case pattern based on increment
tasks (\(t_{[4;9]}\)). Depending on the control task (\(t_2\)), one of the three
different paths (or cases in the switch-case pattern) will be executed. The
first path (or case 0) is composed by \(t_4\), \(t_5\) and \(t_6\) tasks, the second
path (or case 1) is composed by \(t_7\) and \(t_8\) tasks while the
third path (or case 2) is only composed by the \(t_9\) task.
The final expected values in the Finalizer depends on the selected path. If
the first path is chosen (\(t_{[4:6]}\)), "\(+3\)" is added to the values from the
Initializer (\(t_1\)). If the second path is chosen (\(t_{[7:8]}\)), "\(+2\)" is
added to the values from the Initializer (\(t_1\)). Finally, if the last path is
chosen (\(t_{9}\)), "\(+1\)" is added to the values from the Initializer (\(t_1\)).
The -y option indicates that the Controller is cyclic: for the first stream
the first path is selected, for the second stream the second path is taken, for
the third stream the third path is taken, for the fourth stream the first path
is taken and so on...
It is also possible to have a fixed path for all the streams with the -a N
option (with N the path id).
Note
You cannot use -y and -a parameters at the same time, they are
exclusive.
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-exclusive-paths [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-e, --n-exec Number of sequence executions [100000]
-a, --path Path to take in the switch (0, 1 or 2) [0]
-o, --dot-filepath Path to dot output file [empty]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-y, --cyclic-path Enable cyclic selection of the path (with this `--path` is ignored) [false]
-h, --help This help [false]
Simple Pipeline¶
This test is an implementation of a 3 stages pipeline. The first stage
\(S1 = t_1\) reads data from a file. Thus, the generate task (\(t_1\)) is
intrinsically sequential (= executes on a single thread). The second stage
\(S2 = t_{[2:7]}\) just copies the data from the stage \(S_1\) through relay
tasks. The relay tasks can be replicated and ran over multiple threads.
Finally, the last stage \(S3 = t_8\) is a Sink that writes the data on the
file system. This operation is also intrinsically sequential.
The purpose of this test is to validate that buffer exchanges between the pipeline stages are working well. At the end, the code checks that the input file and the output file are the same.
Note
The second output socket of the generate task (\(t_1\)) is bound to the
second input socket of the send_count task (\(t_8\)). If the read data is
not a multiple of the stream size, then the stream is padded with 0. But,
these zeros are NOT written on the file system by the send_count task.
The second output socket of the generate task (\(t_1\)) contains the number
of bits that have been read from the input file. Then, the send_count task
can avoid to write the padding zeros (if any).
Command Line Arguments
The following verbatim is a copy-paste from the -h stdout:
usage: ./bin/test-simple-pipeline [options]
-t, --n-threads Number of threads to run in parallel [10]
-f, --n-inter-frames Number of frames to process in one task [1]
-s, --sleep-time Sleep time duration in one task (microseconds) [5]
-d, --data-length Size of data to process in one task (in bytes) [2048]
-u, --buffer-size Size of the buffer between the different stages of the pipeline [2048]
-o, --dot-filepath Path to dot output file [empty]
-i, --in-filepath Path to the input file (used to generate bits of the chain) [empty]
-j, --out-filepath Path to the output file (written at the end of the chain) ["file.out"]
-c, --copy-mode Enable to copy data in sequence (performance will be reduced) [false]
-b, --step-by-step Enable step-by-step sequence execution (performance will be reduced) [false]
-p, --print-stats Enable to print per task statistics (performance will be reduced) [false]
-g, --debug Enable task debug mode (print socket data) [false]
-q, --force-sequence Force sequence instead of pipeline [false]
-w, --active-waiting Enable active waiting in the pipeline synchronizations [false]
-h, --help This help [false]