Omap Dev Kit Playing A Video

[JakeK]

Thanks Chip and Prophet for the explanation. I am not really in tune with all of this stuff.

On the topic of video, will the media player be more open in the way of what formats and codecs it can play? This probably is a question that would be best answered further down the track but if it can it would be great. Because as I said in another thread, if the media player is anything like the one on the F200 I would hurl the Pandora into a wall.

(Sorry if I went off topic, I just want to find this out)

 

[Squidge]

We can't even be sure it's the same SoC. It could very well be a 3430.

It's a 3530, we just had to call it a 3430 previously for NDA reasons.

Squidge : what is your devkit ? the OMAP 3430 with Virtual Platform ?

This is the dev kit:
http://www.linuxdevices.com/news/NS5754452968.html

No virtual platform, all hardware.

You upload code to the dev kit, watch it crash, fix code, re-upload and keep going round that circle until you have something that works

On the topic of video, will the media player be more open in the way of what formats and codecs it can play?

Well, the current media player is mplayer, so yes, it's fully GPL, and so fully open source. However, DSP code for hardware acceleration of videos may be closed-source (as it will be licensed IP).

 

[ldesnogu]

We can't even be sure it's the same SoC. It could very well be a 3430.

It's a 3530, we just had to call it a 3430 previously for NDA reasons.

Squidge : what is your devkit ? the OMAP 3430 with Virtual Platform ?

This is the dev kit:
http://www.linuxdevices.com/news/NS5754452968.html

No virtual platform, all hardware.
So TI is providing some 3530 dev kits? I thought they only had 3503 chips (ref: http://focus.ti.com/docs/toolsw/folders/pr...dxevm3503.html).
Basically a 3503 is a 3530 with no 3d HW and no c64x+.

QUOTE
You upload code to the dev kit, watch it crash, fix code, re-upload and keep going round that circle until you have something that works

This surely can get frustrating quickly

 

[Squidge]

So TI is providing some 3530 dev kits? I thought they only had 3503 chips (ref:

I don't think the 3530 chips are in production yet, which would explain that.

 

[ldesnogu]

I don't think the 3530 chips are in production yet, which would explain that.

Heh, nice one, you avoided my implicit question
No problem, I can understand.

 

[Svartalf]

It's a 3530, we just had to call it a 3430 previously for NDA reasons.

And the only reason you're able to call it what it is right now and the info's coming out is that TI finally announced the beastie. Typical.

QUOTE

This is the dev kit:
http://www.linuxdevices.com/news/NS5754452968.html

No virtual platform, all hardware.



Heh... I'm hoping that the first round for devs (if I've got a slot) for the Pandora's NOT quite as pricey as that EVM. $1500 is seriously pricey. I can do it pretty effortlessly, really, but that price point put it down the list a bit as I've got several other embedded projects that need that same level of capitalization and I don't want to overspend myself.

QUOTE

You upload code to the dev kit, watch it crash, fix code, re-upload and keep going round that circle until you have something that works



Yep. That's how the first bring-up pass usually works with embedded hardware.

Still, it's cool to see that someone's got the EVM for the SOC in hand and working through issues. Knowing how much everyone's spent up to this point means people ARE serious about this whole thing. You don't make multilayer boards unless you mean it. You don't plunk down ~1.5k on an EVM (Nor, do they just let people HAVE an unannounced unit that easily unless TI is taking you serious as well...).

Cool stuff.

QUOTE

Well, the current media player is mplayer, so yes, it's fully GPL, and so fully open source. However, DSP code for hardware acceleration of videos may be closed-source (as it will be licensed IP).



I can live with the DSP code being proprietary (I won't like it, but I'll live with it...) as long as everyone honors the rights grants by the respective licenses. Which it looks like this is what's happening.

So TI is providing some 3530 dev kits? I thought they only had 3503 chips (ref: http://focus.ti.com/docs/toolsw/folders/pr...dxevm3503.html).
Basically a 3503 is a 3530 with no 3d HW and no c64x+.



Heh... They've got to do bring-up themselves; when they announce the thing, it's been in the labs and in select people's hands for at least a couple of months. Been there, done that. The odds are good that the unit's the real thing.

QUOTE

This surely can get frustrating quickly



You've little idea just how fast it can. >

I've done this with other ARM systems doing work for Project 25 Trunking systems (P25, for those that don't know, is the interoperability spec for the non-spread-spectrum digital police/fire/emt radios in current use. Trunking allows you to increase the number of "channels" by allocating "talk groups" and assigning a channel out of a pool if there's one available to the push-to-talk that a user makes...)- it's a serious pain in the backside if it doesn't come together fairly quickly.

 

[jadon]

Guys, don't go too nuts on this, but the board has an OMAP3530, not OMAP3503. You can only get SUPPORT for the OMAP3503 features right now. I believe that is the answer you get if you call and ask, but we don't want folks buying the board expecting full support to be disappointed.

 

[Svartalf]

Guys, don't go too nuts on this, but the board has an OMAP3530, not OMAP3503. You can only get SUPPORT for the OMAP3503 features right now. I believe that is the answer you get if you call and ask, but we don't want folks buying the board expecting full support to be disappointed.

When would the full support be arriving? I'd think they'd already have the bulk of the 3D rattled out by now or they'd not be offering EVM's with that SOC unless you could develop for it in the first place. I know I'd be VERY peeved if I couldn't get 3D driver support for the eval board for a chip that it's main purpose is providing both the DSP AND the GPU.

 

[jadon]

When would the full support be arriving? I'd think they'd already have the bulk of the 3D rattled out by now or they'd not be offering EVM's with that SOC unless you could develop for it in the first place. I know I'd be VERY peeved if I couldn't get 3D driver support for the eval board for a chip that it's main purpose is providing both the DSP AND the GPU.

And there you go. :rolleyes:

The current eval board is for the OMAP3503 (non GPU/DSP) evaluation. It would just suck to have to send out new chips to everyone once the full support comes on-line. Craig & co. have seen the 3D and DSP working, so there's no chance of not getting the support. It'll be (early, most likely) 2nd half of this year.

 

[DJWillis]

The current eval board is for the OMAP3503 (non GPU/DSP) evaluation. It would just suck to have to send out new chips to everyone once the full support comes on-line. Craig & co. have seen the 3D and DSP working, so there's no chance of not getting the support. It'll be (early, most likely) 2nd half of this year.

Not to mention there is no shortage of things to do with the current EVM . There is plenty of hardware that is not tied up in the PowerVR/DSP arena that can be fun to mess about with.

Note to self: I will get the scalar to play nicely.

 

[JakeK]

Curse you people and your mumbo jumbo! XD

 

[cappuchok]

So, just to get everything straight, is this summary accurate (I guess only jadon and Craig & co. can answer that one)?

* Currently generally available evaluation boards are using 3503 chips as opposed to 3530.
* Craig & co. has an actual rare 3530 evaluation board (not generally available).
* The 3530 evaluation kit currently lacks drivers for GPU/DSP functions, so development for the 3530 at this moment is limited to the 3503 feature set.
* Driver support for the full 3530 feature set will not be available in time for the general Pandora release (may/june) so it will take a few firmware updates before GPU/DSP can be used, making it likely that early ports of programs that would otherwise be better off using GPU/DSP, will have to be completely non-hardware-accellerated (using only raw CPU power) to begin with.

Does that sum up the situation? Or am I missing something?

 

[SSK_Ceazer]

At least that shows that Squidge solved his screen problems

Is that TI OMAP3430 dev kit?

It's TI OMAP3530 dev kit

 

[Squidge]

So, just to get everything straight, is this summary accurate (I guess only jadon and Craig & co. can answer that one)?

* Currently generally available evaluation boards are using 3503 chips as opposed to 3530.
* Craig & co. has an actual rare 3530 evaluation board (not generally available).
* The 3530 evaluation kit currently lacks drivers for GPU/DSP functions, so development for the 3530 at this moment is limited to the 3503 feature set.
* Driver support for the full 3530 feature set will not be available in time for the general Pandora release (may/june) so it will take a few firmware updates before GPU/DSP can be used, making it likely that early ports of programs that would otherwise be better off using GPU/DSP, will have to be completely non-hardware-accellerated (using only raw CPU power) to begin with.

Does that sum up the situation? Or am I missing something?

Pretty much, although there might be some driver support for the GPU at Pandora launch (but probably not final code) and DSP interface code is already there. Just no actual DSP code is running.

 

[ldesnogu]

Pretty much, although there might be some driver support for the GPU at Pandora launch (but probably not final code) and DSP interface code is already there. Just no actual DSP code is running.

I hope it's only the kernel driver glue that is missing for the GPU. My guess is that the OpenGL ES 2.0 high-level part is done by imgtec, and I hope that this is ready.

 

[Arialia]

Pretty much, although there might be some driver support for the GPU at Pandora launch (but probably not final code) and DSP interface code is already there. Just no actual DSP code is running.

I hope it's only the kernel driver glue that is missing for the GPU. My guess is that the OpenGL ES 2.0 high-level part is done by imgtec, and I hope that this is ready.

You can already download the SDK of imgtec for OpenGL ES 2.0 looks great with tutorial

grrr my laptop have problems with Opengl :angry:

 

[ldesnogu]

You can already download the SDK of imgtec for OpenGL ES 2.0 looks great with tutorial

That's "only" an OpenGL ES 2.0 to OpenGL 2.0 translator.

 

[Svartalf]

That's "only" an OpenGL ES 2.0 to OpenGL 2.0 translator.

Heh... It's enough to start migrating code over. Since SDL's presumed to be there as well as ES, one can do 2D testing against whatever ARM unit (2x or N8x0...), and start the dogsbody work of things for ES with the translation layer driver, making most of the needed changes including any fixed point work that you care to do to speed things up. I'd rather I had the nice toys in hand and a more gelled firmware set, but I'm seeing enough to start taking this serious enough to start bludgeoning code into place as time permits.

 

[dizang]

It's playinig on Dev kit not Pandora board. It doesn't prove any significance.

I will beleive it when it's running on the Pandora's board.

Saying in an engineering student sense.

 

[DJWillis]

It's playinig on Dev kit not Pandora board. It doesn't prove any significance.
I will beleive it when it's running on the Pandora's board.
Saying in an engineering student sense.

Valid (if very cynical) point, but in the same vein I think the intention was to show the project is very serious and not vapour. Not to show any 'final' demos as such. Shots of stuff running on Pandora PCB's will be along in good time I am sure.