CCS/TDA3MV: Are there any useful instructions available on how to create a Vision-SDK projekt from the scratch?

Hi Yordan,

this is not really what I was looking for. I just know how to handle CCSv7 and how to setup debug sessions etc.

What I do not know, is what components have to be added and how they have to be added to a CCS projekt, to get the vision SDK running from the scratch.

Currently I'm doing the hard way and I'm analyzing the makefiles delivered together with the VSDK3.0 but I would guess and expect, that there must be an official and documented way on how to setup a new virgin VSDK-project.
Now I have started with an empty SYS/BIOS-Projekt for the two IPUs and DSPs and a simple EVE-Projekt for the EVE and I need to find out, how I can tell my empty system how to use the VSDK. That means how do I have to fill my SYS/BIOS tasks etc. Currently I'm trying this by analyzing the example code in the hope not being on the totally wrong way.

Fact is: If I have to deliver any final software to our customers, this software has to be checked in many ways and I cannot tell them as result of a review: "Sorry, I do not know, why this code part is built in and if it is really being used or if it is just unknown dead code." Doing so would make a lot of trouble and the software won't be accepted by the customers.

Regards,
Guenter

Hi Shiju,

there are a lot of reasons why I would prefer using a CCS project, especially for debugging, but this is not the cause of my origin question.
So let me try to explain, what's currently going on here:

Currently I'm going mad with the whole use of the VSDK, because it is intransparent on how it can be used.
It appears as a kind of construction, which can only be handled well by insiders because it contains some well looking stuff in form of a couple of example chains and algorithms.

My question has the intention to know, how to start with an empty VSDK projekt which I can fill up with our own software contents and ideas and not using a full project like the given examples where I have to remove code parts I do not want to have. It is difficult in this situation to find out, what can be removed and which things should be kept. I also need to know which parts of the software can be kept constant and so can be used as a set of fixed libraries (e.g. to save build time and to get a better overview about the software structure and to allow a better versioning of used software parts). If I would have such separted libraries, than it would also be easy to setup a software project with CCS, where it is very convenient to load and debug the software.

The currently delivered documentation appears well at the first view, but if someone, who never knew the system before, takes a closer look to dig deeper in details you will find out, that there are a lot of undocumented things missing, what people who are really involved in the software do not really see. And it possible to concentrate more on the own code. Currently the whole package kills lots of our creativity because we have to take care more about other things than about our own software ideas.

Currently, when I use it as a complete system, I have a very, very long time for each development iteration/cycle. And each iteration is very, very inconvenient, because I have to do a lot of steps for starting a debug session, where every mistake I would make will be punsished hard because I have to do all steps again and so on. That drives me very mad.
The other thing is, that the debugger does not really know, which of all the many pre-defines have been set at buildtime, when the code has not been built using the IDE. So it is very difficult to walk through the code and when I want to set a preakpoint I cannot see directly the valid code parts and the F3-key does not work propperly.
So it is much more difficult to setup a reliant debug session when the build has not been made with the IDE, where the debugging session can be started directly after my build.

I've had two of your collegues from TI in my office. They told me that they also do not know, how to setup an empty system (independent on using CCS or not!). They recommended to use the current system an remove all not used components. It appeared to me, that they had similar problems like I have, when trying to use the software practically and not only theoretically.

Maybe you will understand now better, what is my aim. My aim is not to use CCS for the complete system. My aim is, to split and divide the system in a way (by using parts as libraries), that I can setup it from the scratch independet on using CCS or not. I want to be able to use my own links and algorithms and want to have a perfect overview about that, what the constant part of the VSDK can do for me and what resources it will cost or how it will interfere with my own ideas.

Regards,
Günter

BTW: I just saw, that after the build of the VSDK by using make, a lot of libraries have been created and I would guess, that some of them could be used for my purposes.