Hello Community,
In the past I spent quite a lot of time with the OMAP3530 (We have a OMAP3530 EVM and some beagle boards here at University) .I tried a lot of things, installing different operating systems, coding some programs in C, got a feeling for the whole system, understood the booting process starting from u-boot etc...
However, the main goal is the implementation of signal-processing algorithms on the DSP side of the OMAP. Some weeks ago I got a copy of the book "OMAP and DaVinci Software for Dummies" which really gave me a good idea on how the whole system is set up in detail.
I also managed to get some signal-processing algorithms working, like the little FIR filter with the IUNIVERSAL interface.
(see http://wiki.tiprocessors.com/index.php/Getting_started_with_IUNIVERSAL)
This actually gives me the possibility to implement DSP code and get it running from the ARM side. However, developing more sophisticated algorithms seems quite a challenge because the steps I have to do are very time consuming.
Up to now I'm doing it that way:
- Get the particular DSP source file from IUNIVERSAL structure (in FIR example it is fir_ti_ialg.c) and edit the process function
(as this is the function that is called by CodecEngine...) - Edit the code with an Editor like GEdit to implement my own functions that are called from the process function
- run the makefile from IUniversal (takes approx. 1 minute)
- start the program on OMAP3530EVM
This works but is very very uncomfortable.
In our signal-processing labs at University (we are Using CodeComposerStudio and DSK6713 floating-point boards) it was much easier to develop programs. There we coded them in a comfortable IDE CCS, compiled the code within some seconds and we were able to debug the code on the chip with jtag.
To do the same with my dual-core OMAP3530 system would be brilliant! But I'm not sure how I should do that.
My first idea is the following:
Start a new project in CCS selecting C64x CPU Cycle Accurate Simulator Little Endian_0/TMS320C64+
Then I can write some functions and debug them in CodeComposerStudio (Emulation). When I think the code works fine I copy the
functions to the IUNIVERSAL ti_ialg.c again and start the building process with the IUNIVERSAL tools and copy the executables to the board. So I'm not using CCS for compilation.
Still time consuming, but at least i can use CCS for softwaredebugging.
So would you think that's a good way to develop complex programs?
Is it possible to get the whole IUNIVERSAL structure built within codecomposerstudio? But then CCS must be able to do the package creation with RTSC codec packaging tools etc right?!
And what about JTAG devices? It sounds like it is possible to write some DSP programs for the C64+ processor only (ignoring the arm side and the whole codecengine stuff) ? Then I would rather develop my code on the DSP chip. Having searched for a JTAG device I found the xds560
(http://focus.ti.com/docs/toolsw/folders/print/xds560.html ) but 3000$ are far too much we can afford for students projects at University. Are there cheaper possibilities?
I hope you understood my intentions.
I'm very very thankful for any suggestions
Maik