You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: _tutos/03-mpsoccodegen.md
+2Lines changed: 2 additions & 0 deletions
Original file line number
Diff line number
Diff line change
@@ -22,6 +22,8 @@ Prerequisite:
22
22
In addition to the default requirements (see [Requirements for Running Tutorial Generated Code](/tutos/intro/#requirements-for-running-tutorial-generated-code)), download the following files:
* A [TMS320C6678 Evaluation Module (EVM)](http://www.ti.com/tool/tmdsevm6678)
26
28
*[Code Composer Studio (CCS)](http://processors.wiki.ti.com/index.php/Download_CCS) v5.2.1 with [MCSDK 2.x](http://preesm.insa-rennes.fr/website/index.php?id=ccs-setup)
Copy file name to clipboardExpand all lines: _tutos/04-softwarepipeline.md
+12-8Lines changed: 12 additions & 8 deletions
Original file line number
Diff line number
Diff line change
@@ -10,22 +10,26 @@ The following topics are covered in this tutorial:
10
10
11
11
* Software Pipelining of an Application for Throughput Optimization
12
12
13
-
Prerequisite: [Tutorial Introduction](index.php?id=tutorial-introduction), [Parallelize an Application on a Multicore CPU](index.php?id=parallelize-an-application-on-a-multicore-cpu)
13
+
Prerequisite:
14
+
*[Tutorial Introduction](/tutos/intro)
15
+
*[Parallelize an Application on a Multicore CPU](/tutos/parasobel)
14
16
15
17
16
18
###### Tutorial created the 10.14.2013 by [K. Desnos](mailto:kdesnos@insa-rennes.fr)
17
19
18
-
1\. Project setup
19
-
-----------------
20
+
## Project setup
20
21
21
-
The starting point of this tutorial is the Preesm project obtained as a result of the [Parallelize an Application on a Multicore CPU](index.php?id=parallelize-an-application-on-a-multicore-cpu) tutorial. The project resulting from this tutorial is available [\[here\]](data/uploads/tutorial_zips/tutorial1_result.zip). The external libraries, the YUV sequence and the generated C code are not included in this archive. Explanation on how to setup these external elements, compile and run the project are available [\[here\]](index.php?id=parallelize-an-application-on-a-multicore-cpu).
22
+
In addition to the default requirements (see [Requirements for Running Tutorial Generated Code](/tutos/intro/#requirements-for-running-tutorial-generated-code)), download the following files:
Software pipelining (also called blocking or block scheduling) is a popular technique used to increase the throughput of an application. The throughput of an application is usually limited by the latency of its critical path. In Synchronous Dataflow (SDF) graphs, the critical path is the chain of actors whose sum of actor execution times is the largest. In the Sobel application, the critical path is composed of the following actors: (read\_YUV)-(Split)-(Sobel\_x)-(Merge)-(display). The basic idea behind the software pipelining technique is to break the critical path into shortest parts by introducing "delays" in the path. The introduction of delays relaxes the dependency constraints between actors and makes possible the concurrent execution of the different stages of the pipelined critical path. More information on software pipelining can be found in [\[1\]](#ref).
30
+
### What is Software Pipelining?
31
+
32
+
Software pipelining (also called blocking or block scheduling) is a popular technique used to increase the throughput of an application. The throughput of an application is usually limited by the latency of its critical path. In Synchronous Dataflow (SDF) graphs, the critical path is the chain of actors whose sum of actor execution times is the largest. In the Sobel application, the critical path is composed of the following actors: ```(read_YUV)-(Split)-(Sobel_x)-(Merge)-(display)```. The basic idea behind the software pipelining technique is to break the critical path into shortest parts by introducing "delays" in the path. The introduction of delays relaxes the dependency constraints between actors and makes possible the concurrent execution of the different stages of the pipelined critical path. More information on software pipelining can be found in [\[1\]](#references).
0 commit comments