New OMAP-L1x Platform

Last week, we had a discussion on the Beagle IRC about an upcoming new platform that is being released by TI this summer that will be ideally suited for applications in such arenas as digital audio, industrial monitoring and low power medical. For those who missed this, I just wanted to follow up on this list in case you’re interested in finding out more.

This new Experimenter Board will be based on a power-optimized OMAP-L1x chip that consists of an ARM9 core (as opposed to the Beagle’s ARM Cortex A8) as well as C674x core (which combines floating and fixed point operations and functionality in the same device). The board will be released by Logic (nee LogicPD) for approximately $169.

Peripheral support will include USB 2.0, SATA, MMC/SD and Ethernet. The ARM OS provided will be a DaVinci Git release of the Linux kernel and the DSP will include the DSP/BIOS RTOS, including new features such as advanced power management and a FAT file system for flash devices.

I’d like ask for any input from the Beagle community on the level of interest in developing software for this kind of platform as well as find out what kind of concerns and care-abouts you might have with the release of such a device.

Thanks in advance,

Dave

Friedland, David wrote:

Last week, we had a discussion on the Beagle IRC about an upcoming new platform that is being released by TI this summer that will be ideally suited for applications in such arenas as digital audio, industrial monitoring and low power medical. For those who missed this, I just wanted to follow up on this list in case you're interested in finding out more.
       This new Experimenter Board will be based on a power-optimized OMAP-L1x chip that consists of an ARM9 core (as opposed to the Beagle's ARM Cortex A8) as well as C674x core (which combines floating and fixed point operations and functionality in the same device). The board will be released by Logic (nee LogicPD) for approximately $169.
       Peripheral support will include USB 2.0, SATA, MMC/SD and Ethernet. The ARM OS provided will be a DaVinci Git release of the Linux kernel and the DSP will include the DSP/BIOS RTOS, including new features such as advanced power management and a FAT file system for flash devices.
       I'd like ask for any input from the Beagle community on the level of interest in developing software for this kind of platform as well as find out what kind of concerns and care-abouts you might have with the release of such a device.
       Thanks in advance,
       Dave

Sounds pretty interesting to me. Here are my thoughts:

1) Available for large quantity orders?

2) LCD is mentioned as a feature of the processor, is this available on the experimenter and to what capacity?

3) Is the experimenter also going to be "community support" style that we're used to on BeagleBoard, or is this going to be more of a corporate/traditional support method?

Thanks!

-Adam

Adam,

1) The Experimenter board would be sourced through Logic. I'm sure they would be delighted to accommodate large quantity orders.

2) The OMAP device has an integrated LCD controller, but there is no LCD display on the Experimenter board itself. A UI daughterboard is also being planned for release at the same time and will include graphical and textual LCD displays on it as well as enabling other I/O such as video and NOR/NAND flash. But of course, there will be an extra cost for acquiring the UI board.

3) That's the idea -- the lower cost Experimenter Board will be oriented around the community support model.

Dave

Last week, we had a discussion on the Beagle IRC about an upcoming new
platform that is being released by TI this summer that will be ideally
suited for applications in such arenas as digital audio, industrial
monitoring and low power medical. For those who missed this, I just wanted
to follow up on this list in case you're interested in finding out more.

This new Experimenter Board will be based on a power-optimized OMAP-L1x chip
that consists of an ARM9 core (as opposed to the Beagle's ARM Cortex A8) as
well as C674x core (which combines floating and fixed point operations and
functionality in the same device). The board will be released by Logic (nee
LogicPD) for approximately $169.

Peripheral support will include USB 2.0, SATA, MMC/SD and Ethernet. The ARM
OS provided will be a DaVinci Git release of the Linux kernel and the DSP
will include the DSP/BIOS RTOS, including new features such as advanced
power management and a FAT file system for flash devices.

I'd like ask for any input from the Beagle community on the level of
interest in developing software for this kind of platform as well as find
out what kind of concerns and care-abouts you might have with the release of
such a device.

I'm hoping Angstrom comes up on it easily. Having support in Davinci
git will be a big improvement over taking patches from random old
kernels and tryin gto get them to work with current kernels :slight_smile:

My interests lie in SDR implementation. With this part, I will be very
dependent on accessing the DSP effectively. We need some light weight,
easy to understand way to move data in/out of the dsp and run a few
signal processing algorithms in the DSP.

codec engine/dsplink are not helping me :frowning:

Philip

It doesn't seem to have been registered at http://www.arm.linux.org.uk/developer/machines/, which is needed if you want to get linux support and u-boot support upstream.

FWIW, initial DA830 support is going into davinci git this week by the looks of it :slight_smile:

regards,

Koen

Friedland, David wrote:

Adam,

1) The Experimenter board would be sourced through Logic. I'm sure they would be delighted to accommodate large quantity orders.

2) The OMAP device has an integrated LCD controller, but there is no LCD display on the Experimenter board itself. A UI daughterboard is also being planned for release at the same time and will include graphical and textual LCD displays on it as well as enabling other I/O such as video and NOR/NAND flash. But of course, there will be an extra cost for acquiring the UI board.

3) That's the idea -- the lower cost Experimenter Board will be oriented around the community support model.

Dave

Hi Dave,

Thanks so much for taking the time to answer these questions. I'm very impressed with the work you folks at TI do and I look forward to playing with these new boards.

-Adam

Thanks for your comments Phillip.

I agree that the Linux software strategy for this board should be a more straightforward one. The plan is to release the Arago (OE/Angstrom overlay) software for this Experimenter board, so I believe it will be close to what you are asking for.

The included software will also definitely include Link and Codec Engine. However, if there are other solutions to DSP/ARM I/O that anyone wants to contribute to the broader community, I think that would be great. Please note that the Codec Engine module is now covered under a BSD license and all of its source code is available to users, so someone could fork this and develop something different if they were motivated to do so. Also, I believe that Link ships with some examples that implement light-weight message passing without the use of Codec Engine. I'm sure that if someone wanted to trade off all of the flexibility that Codec Engine offers (e.g., multi-OS support, cache management, address translation, asynchronous calls, etc), they could come up with something simpler. But hopefully the continuing maturation of the software will also make it easier to use.

Dave

Any comments relative to competing platforms and how this
compares? ???MHz floating-point C674x DSP with SATA, an ARM9, and USB/
Ethernet/SD for $169. Does that meet the expectations of what you'd
want for some interesting projects?

Any comments relative to competing platforms and how this
compares? ???MHz floating-point C674x DSP with SATA, an ARM9, and USB/
Ethernet/SD for $169. Does that meet the expectations of what you'd
want for some interesting projects?

Hi Jason,

I will definitely go ahead any buy one, although I find it somewhat pricy
compared to BeagleBoard. I know it's the BeagleBoard which is cheap, but I
would really like to see this new L1x board at a price point around
119$-129$...

Reason for this:
1) L137 compared to OMAP3530 is 22$ in difference according to www.ti.com
2) Assuming it's made in a 1mm BGA technology the PCB price will as well be
~3-5$ cheaper due to no reason for blind micro-VIAs - Might be it can even
be made on a 4-layer board?

It might be that the price is OK, but currently there are a lot of unknowns
(at least to me) like:
- What's the expected amount of memory & flash?
- Which kind of audio & video-interfaces will the new board present?
- What about the Power solution? Will it be TWL-based or discrete?
- Will JTAG and a serial port be available
- Will it contain any kind of expansion connector like the Beagle?

Does any official documentation (which you can share) already exist on this?

Will there eventually be an OMAP35xx+ with SATA and MAC as well? Keeping the
performance of OMAP3, but with addition of extra IP's - It would be a nice
next step from OMAP36xx :-).

One additional idea would be to upgrade the memory interface to DDR2, which
would be able to double external memory access performance, which is really
a limiting factor in case of multiple DMA transfers together with the
Cortex-A8 and C64+ running code from memory :slight_smile:

Best regards - Looking forward to receiving your feedback
  Søren :slight_smile:

> Last week, we had a discussion on the Beagle IRC about an upcoming new
> platform that is being released by TI this summer that will be ideally
> suited for applications in such arenas as digital audio, industrial
> monitoring and low power medical. For those who missed this, I just
> wanted to follow up on this list in case you're interested in finding out
> more.
>
> This new Experimenter Board will be based on a power-optimized OMAP-L1x
> chip that consists of an ARM9 core (as opposed to the Beagle's ARM Cortex
> A8) as well as C674x core (which combines floating and fixed point
> operations and functionality in the same device). The board will be
> released by Logic (nee LogicPD) for approximately $169.
>
> Peripheral support will include USB 2.0, SATA, MMC/SD and Ethernet. The
> ARM OS provided will be a DaVinci Git release of the Linux kernel and the
> DSP will include the DSP/BIOS RTOS, including new features such as
> advanced power management and a FAT file system for flash devices.
>
> I'd like ask for any input from the Beagle community on the level of
> interest in developing software for this kind of platform as well as find
> out what kind of concerns and care-abouts you might have with the release
> of such a device.
>
> Thanks in advance,
>
> Dave

Any comments relative to competing platforms and how this
compares? ???MHz floating-point C674x DSP with SATA, an ARM9, and USB/
Ethernet/SD for $169. Does that meet the expectations of what you'd
want for some interesting projects?

I noticed the datasheet posted for the L137 isn't very specific on the blocks
but is there any info on how similar the equivalent blocks are with the
OMAP3's? Things I am wondering about are -

Is the normal 2D frame buffer register set going to be the same as the regular
OMAPs?
Is the USB 2.0 block a MUSB block? From the description it looks like it.
Are things like the SPI and audio blocks the same? I noticed they did not use
the terms MCSPI or MCBSP. In fact the audio is described as a MCASP.

What about things like clock gating for power management? Is this similar to
what is on the OMAP3/2?

If this is different, is this chip more software compatible with the DaVinci
line? Just trying to see how much sharing of drivers is there with the other
chips.

Last week, we had a discussion on the Beagle IRC about an upcoming new
platform that is being released by TI this summer that will be ideally
suited for applications in such arenas as digital audio, industrial
monitoring and low power medical. For those who missed this, I just
wanted to follow up on this list in case you're interested in finding out
more.

This new Experimenter Board will be based on a power-optimized OMAP-L1x
chip that consists of an ARM9 core (as opposed to the Beagle's ARM Cortex
A8) as well as C674x core (which combines floating and fixed point
operations and functionality in the same device). The board will be
released by Logic (nee LogicPD) for approximately $169.

Peripheral support will include USB 2.0, SATA, MMC/SD and Ethernet. The
ARM OS provided will be a DaVinci Git release of the Linux kernel and the
DSP will include the DSP/BIOS RTOS, including new features such as
advanced power management and a FAT file system for flash devices.

I'd like ask for any input from the Beagle community on the level of
interest in developing software for this kind of platform as well as find
out what kind of concerns and care-abouts you might have with the release
of such a device.

Thanks in advance,

Dave

Any comments relative to competing platforms and how this
compares? ???MHz floating-point C674x DSP with SATA, an ARM9, and USB/
Ethernet/SD for $169. Does that meet the expectations of what you'd
want for some interesting projects?

I noticed the datasheet posted for the L137 isn't very specific on the blocks
but is there any info on how similar the equivalent blocks are with the
OMAP3's? Things I am wondering about are -

internally it's akin to a davinci, hence the development in linux-davinci git.

Is the normal 2D frame buffer register set going to be the same as the regular
OMAPs?

no idea

Is the USB 2.0 block a MUSB block?

The DA830 patches seem to confirm that.

disclaimer: this is my interpretation of the kernel patches posted to the l-d-o ml and I'm not a coder

regards,

Koen

I think it's pretty interesting, especially for applications where the
OMAP3530's display subsystem isn't required. It should be much easier
to put onto a board than say the OMAP3525 and lead to a lower cost
solution for embedded applications. The built-in Ethernet interface is
nice too. If this board allows the OMAP-L1x chip to benefit from the
community software development effort in a similar way to the
Beagleboard, it would seem like a great idea. Also, if the available
software environment makes it easy for almost anyone to use the DSP
for custom software (offloading the ARM) then it would seem even more
attractive.

Richard.

Søren,

Thanks for the feedback and questions. I'll try to answer as many as
I can up front:

- What's the expected amount of memory & flash?
[DF] The Experimenter Board is configured with 64 MB of mDDR memory.
It also has 8 MB of SPI flash.

- Which kind of audio & video-interfaces will the new board present?
[DF] The Experimenter board will include an AIC3106 audio stereo
codec.

- What about the Power solution? Will it be TWL-based or discrete?
[DF] Neither. It will use another (new) integrated solution called
the TPS6507x. This single chip power management IC is designed
specifically for portable applications and includes a battery charger
for advanced batteries, 3 step-down converters, 10 bit A/D converter
and touch screen interface.

- Will JTAG and a serial port be available?
[DF] The board will include an RS-232 connector and both 14-pin and 20-
pin JTAG connectors. The Experimenter board will support external
emulation through these JTAG connectors and will also include an
integrated XDS-100 on board emulation solution for those people who
don't have an external emulator product.

- Will it contain any kind of expansion connector like the Beagle?
[DF] Yes. And some daughter boards are already being designed to use
with the Experimenter board, including a UI card that supports LCDs
and video as well as NAND/NOR flash.

I apologize for giving you details of the board in such an informal
piece-meal way. The purpose was to give people in the Beagleboard
community an early heads-up this was coming and get a sense of the
level of interest in such a platform. There will be more formal
material available soon and I'll let people on this list know when its
available.

Thanks,
Dave

Hi, David,

Thanks for quick answer. Sounds like a ok/cool peripheral set...
- Then it's only the price which is a bit too high (compared to Beagle :slight_smile:

As I understand this, the main idea is that the SPI flash is only for
booting and initialization of the SATA, which gives access to all the
storage space you need. Will it be possible to upgrade/extend the mDDR
memory through the expansion connector or is this fixed at 64MB?

I'm looking forward to seeing some official material - When can we expect
the new chip/board to arrive?

Best regards and thanks - Enjoy your weekend
  Søren

Hi David,

Count me as another very interested buyer.

I would like to repeat Jason's request for comment on the floating
point performance of the two OMAP versions (NEON vs DSP C674x).

What will be the maximum bandwidth of for any of the interfaces on the
expansion connector? My intent is to use an OMAP computer in
industrial control applications by interfacing it with an FPGA.

Thank you,
Eugene

Hi David,

I'm very interested in buying one too even though I agree with Soren
that the price point is a little bit high.

Regards,
Matthieu.

In regards to floating point performance, the OMAP-L137's DSP can
handle 6 FP SP operations per clock cycle sustained, so in general I
would expect it to be significantly faster than the floating point
capability on the Cortex A8 even though there is a disparity in the
raw clock frequencies, in particular if you need full IEEE compliance
meaning you would be using the VFP Lite and not actually the NEON. I
haven't actually come across any floating point benchmarks that can
actually use the NEON due to the need for IEEE compliant floating
point types.

Søren,

The OMAP chip and the DDR memory are configured on a small daughter
card (called a "SOM board") that plugs in to the baseboard. This was
done to allow further flexibility. So to change the amount of mDDR on
the board, you would need a different SOM. There is a plan for TI to
sell a higher priced version of this board that include the UI board
and 128 MB of mDDR (instead of 64 MB). If you didn't need the UI
board, I'm not sure at this point whether Logic would offer the
upgraded SOM as a separate part number.

The board is slated to be released before end of Q3 (and hopefully
well before that). I'll post a more exact schedule when I get it.

Dave

BTW, with regard to the bandwidth spec of the expansion connector,
this is still pending. However, there is a UPP peripheral and this
would be a very good interface for an FPGA solution. Not sure yet
what the max clock rate will be on the actual board though. We should
be able to find that out soon.

So, the Omap L-137 supports mDDR ? In some places, it says that it
also supports DDR2. Can you confirm ?