Any word on when this product will be available? You showed prototypes of the board in July 2013, that's a year ago. In May this year you announced that it will be for sale by the end of the month. Yet it is still "out of stock". In the mean time new Allwinner based products are announced almost weekly.
The Sitara processors have some advantages over the Allwinner SOCs, for one thing, they are vastly better documented.
Linux support for Sitara is total MESS
if you want to add board != of BBB the chances things to work are close to zero
you enable the LCD and your Ethernet stop working for example or other weird stuff happend
perhaps we are dumb, but the major problem is that nobody helps you if you choose Sitara to work with
we posted several times questions on BeagleBoard forum without reply, and we are not alone you can count how many people there ask for help for boards they made with Sitara with same result - nobody helps them
Robert Nelson who is taking care for BB Linux support now do not respond to our e-mails anymore (used to reply before when he was not so involved with BB)
So we will keep trying to make Linux support for AM3352, but we don't know when this will be complete
...and we have plenty of more interesting stuff to do, at least Allwinner have strong Linux-Sunxi community not locked with strings to particular board/vendor and you can communicate and get help to your (most of the times maybe silly) questions :)
Quote from: olimex on July 16, 2014, 05:35:14 PM
Linux support for Sitara is total MESS
[...] but the major problem is that nobody helps you if you choose Sitara to work with
Sorry to hear you are having problems with the software.
It's a bit hard to help if the hardware is not available.
Quote from: olimex on July 16, 2014, 05:35:14 PM
Linux support for Sitara is total MESS
if you want to add board != of BBB the chances things to work are close to zero
you enable the LCD and your Ethernet stop working for example or other weird stuff happend
perhaps we are dumb, but the major problem is that nobody helps you if you choose Sitara to work with
we posted several times questions on BeagleBoard forum without reply, and we are not alone you can count how many people there ask for help for boards they made with Sitara with same result - nobody helps them
Robert Nelson who is taking care for BB Linux support now do not respond to our e-mails anymore (used to reply before when he was not so involved with BB)
So we will keep trying to make Linux support for AM3352, but we don't know when this will be complete
...and we have plenty of more interesting stuff to do, at least Allwinner have strong Linux-Sunxi community not locked with strings to particular board/vendor and you can communicate and get help to your (most of the times maybe silly) questions :)
I have not worked with a TI board yet, but from what I can see all main drivers for TI are in mainline linux and TI does a lot of work keeping their stuff in mainline, which is great and which is why I was hoping to get a am3352 SOM soon to try and work with it.
Sunxi has a great community, but they just started to mainline all drivers, which will probably still take month till most parts are supported. So long everybody is stuck with a 3.4 kernel.
There are a few other vendors out there with Sitara SOMs so it should be doable.
Are the problems hardware or software related?
I am sure there are people out there who can help with this stuff. As all parts should be mainline isn´t the most hard part defining a new device tree file and the rest should work (at least in theory)?
As you have a community here as well maybe some people who are interested could help with this stuff?
It may be possible to ask for volunteers who provide details of their expertise to qualify for free boards. Or some such.
(Offhand, BB appear to be competitors so I don't get why they would help.)
John
BB probably make no money from their project, well Circuit Co may take their share for the assembly but Jason Krinder probably gets just patting on his back from the TI managment, they exist to promote TI Sitara chips
this is not new with TI, many other TI development boards are with same business model like the $4.30 MSP430 Launchpad etc
all BB massive advertising campains are backed by TI machine and lot of $$$ is put in this project which will probably never return back
so it's hard to call BB competitor, as AM3352-SOM will do the same - promote Sitara chips use and sales
my observations so far is that TI do not care to develop wider Open Source community around Sitara and they will suffer in long term from this decission.
Allwinner community is bigger, people more easily start their projects around Allwinner chips (Olimex have it's humble share in this with the OSHW designs) you can always get help on linux-sunxi mailing list no matter if you do something with development board, TV android stick or tablet on which you want to run Linux
Quote from: jlucius on July 16, 2014, 06:51:49 PM
I have not worked with a TI board yet, but from what I can see all main drivers for TI are in mainline linux and TI does a lot of work keeping their stuff in mainline, which is great and which is why I was hoping to get a am3352 SOM soon to try and work with it.
Sunxi has a great community, but they just started to mainline all drivers, which will probably still take month till most parts are supported. So long everybody is stuck with a 3.4 kernel.
There are a few other vendors out there with Sitara SOMs so it should be doable.
Are the problems hardware or software related?
I am sure there are people out there who can help with this stuff. As all parts should be mainline isn´t the most hard part defining a new device tree file and the rest should work (at least in theory)?
As you have a community here as well maybe some people who are interested could help with this stuff?
mainline? imx233 is also in "mainline" but many things do not work correctly and nobody cares to fix them for more than year, same is with am3352 in mainline is the configuration for BB and nothing else, try to enable two Ethernets for instance :)
Sunxi is stuck at 3.4 due to the decission of Allwinner to not use device tree but their own FEX this means every time they release something Sunxi community have to duplicate their work which is lot of work and silly in long term, now when Allwinner is part of Linaro things must change and they should leave FEX (which is not bad IMO ;) )
it's hard to say if something is pure hardware or software problem when it doesn't work
I suppose any chip-maker doesn't really care as long as millions are being sold. AW are doing very well on that metric. TI I suspect are doing relatively poorly. That would make some firms work to overcome the roadblocks (such as lack of working software).
The chip makers who fail to assist with software can expect a very tough financial environment. If that is TI, well it's their choice.
I'd like to see stronger competition for AW from other chip makers, but it's up to the chip makers.
John
Quote from: olimex on July 17, 2014, 11:56:37 AM
mainline? imx233 is also in "mainline" but many things do not work correctly and nobody cares to fix them for more than year, same is with am3352 in mainline is the configuration for BB and nothing else, try to enable two Ethernets for instance :)
it's hard to say if something is pure hardware or software problem when it doesn't work
OK mainline may not be error free, but at least I don´t have to deceide either to not have some functions with a 3.1x kernel or backport needed functions to 3.4. But as I said there is great work beeing done at sunxi community and it´s getting better every day.
I am no hardware guy so this might be naive, but are there no TI field engineers giving support to you, I suppose Olimex is big enough to qualify for support?
Quote from: JohnS on July 17, 2014, 12:22:40 PM
I suppose any chip-maker doesn't really care as long as millions are being sold. AW are doing very well on that metric. TI I suspect are doing relatively poorly. That would make some firms work to overcome the roadblocks (such as lack of working software).
Thats probably why TI dumped their mobile business?
http://www.reuters.com/article/2012/09/25/texasinstruments-wireless-idUSL1E8KP5FN20120925?irpc=932
What other chip vendors are there (for chips that can be bought and used by companies like olimex AND have some kind of Linux support)?
Guess that only leaves the three known vendors? Freescale, TI, AW ?
What about Nvidia, Samsung, Broadcom?
Rockchip?
Mediatek violating GPL because they don´t release source code?
Seems there are not many options.
There are also atmel ones, good linux support. Or the 'ew freescale one, which are much more aimed for industrial application. And they rock (imx6)
Snag is Atmel quite often seem to make orphans of their chips and various people have complained about Freescale in other ways. Different problems than AW but no-one seems well behaved!
John
Freescale do not do it better than TI http://www.eetimes.com/author.asp?section_id=69&doc_id=1286785
they closed their research center in Israel last year and left the imx7 development
Seems that according to you TI and Freescale are making a mess of their support and Allwinner is the only decent option around, although there are multiple examples around for successful designs that use Freescale and TI chips on their modules/boards.
Maybe it's time for a reality check?
Please say which are the successful ones.
John
http://www.eetimes.com/document.asp?doc_id=1323198
Quote from: JohnS on July 26, 2014, 01:44:00 PM
Please say which are the successful ones.
John
Phytec, Variscite, Emtrion all have AM335x based modules and there are others, they're quite easily found when using google for example.
Quote from: olimex on July 26, 2014, 05:40:38 PM
http://www.eetimes.com/document.asp?doc_id=1323198
Ehhm yes. I am sure you have some sort of point but it eludes me. Could you explain what you are trying to say?
Quote from: farlane on July 27, 2014, 10:06:41 AM
Quote from: olimex on July 26, 2014, 05:40:38 PM
http://www.eetimes.com/document.asp?doc_id=1323198
Ehhm yes. I am sure you have some sort of point but it eludes me. Could you explain what you are trying to say?
That was just a reply to the discussion of TI and Freescale exiting the mobile CPU market now adding broadcom as well.
Quote from: jlucius on July 27, 2014, 10:29:49 AM
That was just a reply to the discussion of TI and Freescale exiting the mobile CPU market now adding broadcom as well.
But that is completely irrelevant to the topic, the Sitara isn't even a 'monile CPU' but mostly positioned as an industrial controller.
IMHO Olimex is trying to make excuses for not properly engineering this module. My opinion is that Olimex has to do her work properly or stop development of this module altogether and be honest to the customers ( and herself )
Quote from: farlane on July 27, 2014, 12:27:08 PM
Quote from: jlucius on July 27, 2014, 10:29:49 AM
That was just a reply to the discussion of TI and Freescale exiting the mobile CPU market now adding broadcom as well.
But that is completely irrelevant to the topic, the Sitara isn't even a 'monile CPU' but mostly positioned as an industrial controller.
IMHO Olimex is trying to make excuses for not properly engineering this module. My opinion is that Olimex has to do her work properly or stop development of this module altogether and be honest to the customers ( and herself )
+1 freescale and ti and atmel are much more industrial and used in the industry! Automotive field for instance.
Quote from: farlane on July 27, 2014, 12:27:08 PM
IMHO Olimex is trying to make excuses for not properly engineering this module. My opinion is that Olimex has to do her work properly or stop development of this module altogether and be honest to the customers ( and herself )
Sorry I think it was me dragging this thread to an off-topic discussion, but it was very interesting.
For an answer see post #2 here from olimex where they say that there are problems and they don´t get proper support.
I am also still interest in this board ...
Quote from: jlucius on July 27, 2014, 03:53:17 PM
For an answer see post #2 here from olimex where they say that there are problems and they don´t get proper support.
I'm not buying that, why would TI not support a customer which potentially could sell a lot of their CPUs? I think Olimex just isn't motivated enough to make this thing work. And that is their prerogative but just stop making a fool of yourself and your customers and abandon the damm thing.
Seriously, more than 1.5 years for designing a module (which is still not working aparently) is just ridiculous and you can't simply blame that on TI.
farlane, indeed we work on this modues with low priority for the reasons above, did you ever got Linux related support from TI? who is your contact there?
Quote from: olimex on July 27, 2014, 05:27:16 PM
farlane, indeed we work on this modues with low priority for the reasons above, did you ever got Linux related support from TI? who is your contact there?
You work on this module with low priority because 'more interesting things' come around as you have told me back in June 2013. Don't blame your unwillingness to take this project seriously on any other than yourself because that would be nonsense.
I have no contact for Linux related support @TI because we don't use TI but maybe you could have a look at http://processors.wiki.ti.com/index.php/Linux_Consultants_and_Commercial_Linux_Providers and check if any of those can help out? You could in any case bring your hardware designs online and see if any of your users could help out with that or has any tips to improve that side of things. OSHW design and all ...
Quote from: farlane on July 27, 2014, 05:58:05 PM
Quote from: olimex on July 27, 2014, 05:27:16 PM
farlane, indeed we work on this modues with low priority for the reasons above, did you ever got Linux related support from TI? who is your contact there?
You work on this module with low priority because 'more interesting things' come around as you have told me back in June 2013. Don't blame your unwillingness to take this project seriously on any other than yourself because that would be nonsense.
I have no contact for Linux related support @TI because we don't use TI but maybe you could have a look at http://processors.wiki.ti.com/index.php/Linux_Consultants_and_Commercial_Linux_Providers and check if any of those can help out? You could in any case bring your hardware designs online and see if any of your users could help out with that or has any tips to improve that side of things. OSHW design and all ...
Well these are commercial support engineers, but it also links to Sitara Forum (http://e2e.ti.com/support/arm/sitara_arm/default.aspx)
Ever tried asking there? Looks very active.
Quote from: farlane on July 27, 2014, 05:58:05 PM
You work on this module with low priority because 'more interesting things' come around as you have told me back in June 2013. Don't blame your unwillingness to take this project seriously on any other than yourself because that would be nonsense.
aah this explains your attitude above :)
sorry to have dissapointed you but this is something we do for fun, so as you wrote above if you want to work now with AM335x modules there are plenty of offerings most of them with commercial support.
I can't tell you when this prject will be finished it may be next month or next year, everything depend on if we can get help for the Linux support, so far we have got none although we searched mostly from the community around BB as some of these people we know and see what they did
Quote from: jlucius on July 27, 2014, 06:26:45 PM
Well these are commercial support engineers, but it also links to Sitara Forum (http://e2e.ti.com/support/arm/sitara_arm/default.aspx)
Ever tried asking there? Looks very active.
I have read few comments about TI official SDK being used obsolete kernel and lack of support, compared to BB community Linux distribution, this is why we looked at BB forums first
Quote from: farlane on July 27, 2014, 10:06:41 AM
Phytec, Variscite, Emtrion all have AM335x based modules and there are others, they're quite easily found when using google for example.
It's completely news to me that these are "successful". Really???
Frankly they don't seem to me to be so, but YMMV.
John
Quote from: olimex on July 27, 2014, 07:07:34 PM
aah this explains your attitude above :)
sorry to have dissapointed you but this is something we do for fun, so as you wrote above if you want to work now with AM335x modules there are plenty of offerings most of them with commercial support.
I can't tell you when this prject will be finished it may be next month or next year, everything depend on if we can get help for the Linux support, so far we have got none although we searched mostly from the community around BB as some of these people we know and see what they did
If my 'attitude' is that i am criticizing the fact that Olimex does not seem to take it's projects seriously then yes.
Quote
I can't tell you when this prject will be finished it may be next month or next year
Maybe you should inform your users a bit better that you have this 'attitude' about projects you announce and avoid this weird situation altogether. Keep it fun for your customers too so to say.
Quote from: JohnS on July 27, 2014, 07:34:11 PM
It's completely news to me that these are "successful". Really???
Frankly they don't seem to me to be so, but YMMV.
Successful in the sense that they are actual working modules. If they're commercially succesful i would not know.
Quote from: farlane on July 27, 2014, 08:18:48 PM
Successful in the sense that they are actual working modules.
How do you know these are working modules?Do you use these? What Linux Kernel they run? What features are supported by the kernel and drvers?
Quote from: olimex on July 27, 2014, 09:54:38 PM
How do you know these are working modules?Do you use these? What Linux Kernel they run? What features are supported by the kernel and drivers?
I'm assuming they are working because they are actually sold. Maybe you should ask one of these companies for a bit of help?
Could you also react to the other points? I find it rather typical that you pick out a single remark and start nitpicking over that instead of just openly discussing all the concerns i'm raising. (it's not a fight you know, there is no winning or losing here)
As i said, it is your prerogative to approach such a project the 'we might get it done, or we might not' just be clear about it.
Quote from: olimex on July 27, 2014, 07:10:19 PM
Quote from: jlucius on July 27, 2014, 06:26:45 PM
Well these are commercial support engineers, but it also links to Sitara Forum (http://e2e.ti.com/support/arm/sitara_arm/default.aspx)
Ever tried asking there? Looks very active.
I have read few comments about TI official SDK being used obsolete kernel and lack of support, compared to BB community Linux distribution, this is why we looked at BB forums first
The latest SDK is from March 2014 and is based on Linux 3.12. See
http://www.ti.com/tool/linuxezsdk-sitara
I would stop trying to get help through the Beagleboard and get directly to TI. Any larger production company should be able to get help via a TI field engineer.
I just got a Beaglebone Black and will try a few things with that, but it would really be nice having the Olimex board.
But you also must understand that people will try do do business with your (olimex) stuff and in that case people have to rely on you to deliver stuff at a schedule and with continuous quality. And the SOMs are most likly used in business products. I think that´s what makes farlane angry.
Hello,
I'm studying Electrical Engineering and currently I'm writing my Master Thesis. I planned some Time ago to use your AM335-SOM and designed a own Baseboard using your Baseboard Sources. Unfortunatly you didn't sold me some of your Boards:P ...even after I explained my Problem and that I would buy them without Software Support (I called you in Bulgarie)... So I have to change to the SOM from Phytec. They are working correctly. Phytec is using PTXDIST to build the Kernel. Their free downloadable BSP is using Linux 3.2. I managed it to change everything to my own Baseboard. Only my Ethernet Port isn't working currently but I'm using a TLK110 Chip and until now I didn't found Information about integrating this Chip in Linux correctly (it is only for Future use in my Projekt , so it isn't a must have for me). Phytec is using a SMCS Chip and it works. So it could not be a Hard thing to adopt their BSP to your Board even I managed it within 1 Month and I never build a Kernel before...
So I think you should Download PTXDIST and the BSP from Phytec and build a Linux. Then I'm still interested in your Boards... (I would buy some also with no Software Support!!!!!)
Try this Link:
http://www.phytec.de/de/support/faq/faq-phycore-AM335x.html and http://phytec.com/wiki/images/c/c4/L-775e_3.pdf
If you want, I can send you my Boardfile, my Schematics and Layout from my Baseboard.
Hi Noy
Thanks for the tip and I really sorry for your problems around your thesis
We also have a bit of progress with AM3352-SOM meanwhile (it become the longest developed board in our company).
After several e-mail exchange with TI support we figured out that TI SDK is quite deep binded with the TI PMU chip and the Kernel have quite a number of calls to it and expect the PMU to reply. We do not use PMU and there boot hangs.
Dimitar Gamishev made dummmy I2C device on the same address which always answer OK instead the PMU and now the Kernel from TI SDK boots and we have everything working exept the second LAN.
I hope we will be able to run the second LAN in the next days and finally move AM3352-SOM forward.
Of course the current solution is ugly, but at least prove our hardware works so far.
Our Kernel patching knowledge do not allow to make elegant solution for the moment, but I hope once we release these there will be good people from the community which will help :)
Tsvetan
It is an "ugly" Solution, but even a Solution. So is it possible to buy some Boards in the near Future?
Hint:
Phytec is also providing "Dummy-Regulators" in their Board-File. I think it is caused to some similar Problems, which you have.
Quote from: Noy on September 09, 2014, 06:44:31 PM
Hello,
Only my Ethernet Port isn't working currently but I'm using a TLK110 Chip and until now I didn't found Information about integrating this Chip in Linux correctly (it is only for Future use in my Projekt , so it isn't a must have for me).
I would advice against using a TLK110 PHY, we had some problems with it in combination with some 1Gb switches when the device is unpowered. TI herself indicates that there is a recommended replacement part ( http://www.ti.com/product/dp83848i/description&lpos=See_Also_Container&lid=Alternative_Devices )
This device also is a bit simpler from a software point of view with respect to link detection.
Just to let you know, YMMV.
Thank you for this Information. Do you know if the TQFP Chips are Pin-Compatible? Cause my PCBs are already produced for the TLK110.
>> now the Kernel from TI SDK boots and we have everything working exept the second LAN.
WOOHOO GOOD news....ETA on rolling out the first batch? Most people probably don't need the second LAN.
Would be interesting to run a poll on what about the am3352 has people so interested and if an All Winner SOM the same size as the am3352 would attract the same attention.... ie A10-SOM?
Quote from: bruceuk on September 16, 2014, 07:18:52 PM
>> now the Kernel from TI SDK boots and we have everything working exept the second LAN.
WOOHOO GOOD news....ETA on rolling out the first batch? Most people probably don't need the second LAN.
Would be interesting to run a poll on what about the am3352 has people so interested and if an All Winner SOM the same size as the am3352 would attract the same attention.... ie A10-SOM?
But there are already two Allwinner SOMs, the A13 and A20. For me it´s the TI mainline linux support and the two LAN interfaces I am interested in.
Quote from: Noy on September 15, 2014, 09:32:32 AM
Thank you for this Information. Do you know if the TQFP Chips are Pin-Compatible? Cause my PCBs are already produced for the TLK110.
Yes they are pin compatible :) For us it just required to do adjust the link detection code a bit.
Quote from: olimex on September 10, 2014, 04:04:41 PM
Dimitar Gamishev made dummmy I2C device on the same address which always answer OK instead the PMU and now the Kernel from TI SDK boots and we have everything working exept the second LAN.
I hope we will be able to run the second LAN in the next days and finally move AM3352-SOM forward.
Any news?
Any chance we could get some form of ETA on these boards? Maybe a smaller quantity run at a higher cost?
Scratch that.....my A13-SOM arrived, don't think I'll need the A3352-SOM anymore ...this little A13-SOM is VERY VERY cool.
Although not sure about the processor package, think BGA would be more robust. The pitch on the pins is ridiculous :)
What is status on the AM3352 SOM?
I'm developing a system utilizing AM335x PRU for wide bus high-speed data transfers, at the same time I need HDMI which makes the BeagleBone pretty useless: PRU1 pins is taken by HDMI/LCD, and BBB uses PRU0 pins for HDMI ctrl SPI1, which I hope to relocate to free PRU0. I can develop without screen on BBB but final product need HDMI. Hence my search for a more flexible SOM like yours.
edit:typos
Quote from: kjetilh on January 07, 2015, 02:12:00 PM
What is status on the AM3352 SOM?
I'm developing a system utilizing AM335x PRU for wide bus high-speed data transfers,
The AM3352 (nor the AM3354) does not contain a PRU, sorry.
Quote from: jmyreen on January 07, 2015, 08:38:37 PM
The AM3352 (nor the AM3354) does not contain a PRU, sorry.
Really?
In that case, TI does a less than average job of clarifying. I skimmed the DS just now without really getting this, and visiting the AM3352 product page (http://www.ti.com/product/am3352) it clearly states 2x PRUs under device features.
EDIT: I see now the PRUs is not a specified feature in the parametric product selector table, nevertheless IMO this should be made a lot clearer.
EDIT2: thanks a lot!!!