Any plans for releasing Startware for OMAP-L137 or C6747 DSP?

Thank you for your interest in StarterWare.  Unfortunately, there are currently no plans to release StarterWare for OMAP-L137 or C6747.

Sigh... According to the TI Sales Reps, the C6747 is supposed to be a simple migration step up from the C6726 chip. On paper it looks that way, but in reality it is not.

First you buy an OMAP-L137 kit, and then find out that you have to upgrade to CCS v3.3 Release 12 and make other corresponding software changes to your existing development setup. Then you discover that projects are no longer built the same way as earlier CCS v3.3 projects because the tools that you just installed have transitioned over to the Eclipsed based model.  Likewise you discover that there is no Chip Support Library for the C6747 as there was for the C6726 chip.  Next you do some more research into using the DSP BIOS, only to find out that the McASP driver doesn't support simultaneous transmit and receive transfers over the same McASP port.

Finally I catch a glimmer of hope that there is something called StarterWare that lets you control the SOC features of the C674x chip family without requiring the DSP BIOS. Now I found out that the StarterWare is only for the C6748 chip.  Just Great!

I started with a working C6726 based project that used two McASP ports (with simultaneous Tx/Rx on each port).  I needed a few more MIPs to add some additional features, and now I find out that the C6747 chip is a migration path over a cliff.

Hi Joseph

Sorry to hear that ramp from C6726 to C6747 has been challenging. Using CCSv5 gets you to the latest/greatest tools even though I do understand the challenges on the learning cycle that comes with it. 

If you do like what you see in Starterware (which also is based on CCSv5 based examples) , and your key issue is to get an example up and running using McASP and EDMA etc.,  I do think you can use the Starterware targetted and C6748.

C6748 and C6747 has a lot of commonality, and even though C6747 has more number of McASPs and serializer (which I believe would be one reason you are using C6747 instead of C6748). The McASP0 memory map and base address, EDMA0 memory map and base address, McASP interrupt # to DSP Interrupt controller and McASP events # to EDMA are identical on both these devices, so you might see a lot of software compatibility and re-use. 

There will obviously be some challenges as the package is targetted to work out of the box for the C6748/OMAPL138 based EVMs , however I do think the crux of the code can be re-used. I know this is still not the ideal solution, but I did want to highlight the commonality between the devices w.r.t to McASP/EDMA/memory maps etc.

Joseph Zolyak said:

Next you do some more research into using the DSP BIOS, only to find out that the McASP driver doesn't support simultaneous transmit and receive transfers over the same McASP port.

Can you also elaborate on the above, so that we can run this by the BIOS PSP team also.

Regards

Mukul 

I have skimmed through the StarterWare example code and include files, and modifying it for re-use with the C6747 chip isn't as simple as increasing the number of McASP instances from 1 to 2, and adding the addresses for the memory mapped registers that control McASP #1.  Take for instance the header file edma_event.h which is included by the example program mcaspPlayBk.c.  The EDMA event enumeration is going to be different for the C6747 than it is for the C6748. Who knows how many more files like this that I am going to have to examine and tweek, and I am not an expert on either DSP chip.

As for the McASP issue in DSP BIOS, go to the document located at:

C:\Program Files\Texas Instruments\pspdrivers_01_20_00\packages\ti\pspiom\mcasp\docs\BIOSPSP_McASP_Driver_Design.pdf

On page 11 it states:

     Note: The driver does not support two concurrent channels performing the same
     operation (transmit/receive) i.e. any given channel has to operate in either the transmit or
     receive mode but not both simultaneously. The IDriver allocates two channels only if they
     perform different I/O operations.

Hi all,

I also miss the Starterware package for OMAP-L137. May I ask why there are no plans to release the software for L137 family? I faced to the same issues than Joseph so I'd like to receive a kind of feedback in this regard.

Regards,
Gaston

Hi Gaston

Sorry to hear that the lack of Starterware on OMAPL137/C6747/AM17xx is impacting your development too. At this point the decision to not have a software release on L137 family is purely driven by resource/bandwidth/customer demand considerations. The Starterware team is focussed on improving the OMAPL138 offering, release Starterware for devices after OMAPL1x like AM335x and upcoming roadmap devices. 

We continue to investigate alternatives to see how we can extend support for OMAPL137 and family, but at this point there is nothing formalized or concrete, so Joe's original response stands true.

Regards

Mukul 

Is the StarterWare for L137/C6747 still not on TI's roadmap?

Is there a way to even pay Texas instruments to port the mcaspPlayBk example to omapL137

I am cetain the hardware is equivalent and should work fine,....

If there is someone who can do it for payment....

I would really appreciate if you can refer me to him,

Best regards

Yehuda

yehudayehudai@yahoo.com

Hi Yehuda

We will forward these post to a few 3P to see if they can offer software services on your request.

Regards

Mukul

Yehuda,

A TI third party will contact you shortly to follow up on this topic. Please check your email.

best regards,

David Zhou