Lorenzo Lutti wrote:
Stephen Berry ha scritto:
You do have to pay a ransom to MontaVista for the rights to sell
their kernel - this I believe is around $6K. I never really
understood how this worked since so much of the software was GPL'd.
This is true for the commercial MV Linux; but what about the free
version that you can check out from the GIT repository? Accidentally,
it should be the "Davinci linux open source" that theoretically gives
the name to this mailing list... :)
My understanding is that the git kernel - as it relates to DM355 support
- is not quite there yet. And by that I mean the ability to run the
encode/decode demos that are shipped with the EVM. If I'm wrong -
somebody please tell me!
SPI support is sparse. The kernel is 2.6.10 based with a zillion
patches. No cohesive low power scheme. MMC can be booted off of the
kernel if you don't need SDHC. If you want that they can be added
(and original Ti MMC drivers removed) as loadable modules - but then
you can't boot from SDHC. The VPFE/VPBE support is passable, but
needs work. Yaffs2 for NAND does not include checkpointing, so expect
long boot times with a larger filesystems.
Ok, so basically the status is exactly the same as two years ago:
there isn't any working, full-blown kernel that you can use for an
actual production device without spending several months-men to modify
the kernel, fill the voids, and so on.
The 6446 may be much further along, especially when it comes to the git
kernel, but I wouldn't know for sure. The 355 has only been out for a
year so that is not exactly true.
Basically the questions can be collapsed into a simpler one: what do
I get by just spending the 495$ for the DVEVM? Do I get enough
software code/tools to develop an actual product, or other
commercial tools are practlcally required?
[...]
You get most of what you need.
Well, actually you didn't give me this idea. :)
You asked if you get enough tools to do the job - you do. Of course
eveybody's goals are always different.
My intention was to determine if/how this platform could be already
useable for low-cost, low-volume applications (100 to 1000 pieces, at
best 10000). This basically means that three requisites need to be met:
-Availability of low-cost boards based on DaVinci;
-Low non-recurring costs (licenses, development system, etc);
-Working low-level hardware support (i.e. I don't have to spend months
writing kernel code).
I've started investigating on the first requisite, and something
actually popped up; but as for the last two requisites, it looks like
we are just in the same place as two years ago.
This isn't really a complaint, it's just an osservation; I totally
understand that products like DaVinci are mainly aimed at
zillion-pieces applications. What I don't understand is why TI sells
the DM355 board at such an aggressive price, if then you have to spend
thousand of dollars anyway to get anything working.
Cheers, Lorenzo
There are more low cost boards now then there were last year ;) But
seriously the DM355 does not have the same start up costs as the 6446 or
other true DSP based video chips. When we started looking Ti quoted us
an enormous amount of money to buy the JPEG and MPEG4 codecs for the
6446, while the 355 has no such costs. Also the production bundle for
the 355 is a lot less. For us it was a better way to go.
Steve
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
