128mb Of Ram...

[Adventus]

It says here that the OMAP 3530 will only be available in the 2nd half of 2008:

http://www.iii.co.uk/investment/detail/?di...&id=3928029

Should we push back our expected release date?

Edit: Or have i just read something completely irrelevant...

 

[ldesnogu]

TRM not available. It should be SPRUF98

 

[MWeston]

It says here that the OMAP 3530 will only be available in the 2nd half of 2008

Yes, for general catalog ordering, this is true but we are a special case and one of TI's pet projects so we get access to things before others.

 

[atomicthumbs]

one of TI's pet projects

woot

 

[Exophase]

Is the 430mhz for the graphics chip?

No, it's for the DSP, which is a specialized second processor. Fans of Blackfin might go for this because it looks similar, if not more powerful per clock.

 

[icurafu]

EDIT: Exophase's answer above.

Is the 430mhz for the graphics chip?

http://focus.ti.com/lit/ds/symlink/omap3530.pdf

430MHz is the frequency of the DSP core. What gives us 720P. (audio video encoding and decoding) I guess we can use the for games, so the CPU does not have to handle audio and it seems to have a reasonably phat pipe to the DMA.

In terms of graphics performance it is confirmed in the doco above that:

10 Million polygons per second.

Says OpenGL ES 2.0 (Not OpenGL 2.0)

All the stuff we knew before like tile rendering and unified shading model.

ARM is big endian, DSP is little endian.

Memory has 2GB address space.

Memory is LPDDR (low power DDR)

In terms of SGX... It's hard to tell. There are a couple of references to it, but I'd just be guessing.

EDIT2: I could guess from reading it that it is... 110MHz at the max bus speed of 166MHz. At this point the CortexA8 is clock at 664MHz.

 

[ldesnogu]

No, it's for the DSP, which is a specialized second processor. Fans of Blackfin might go for this because it looks similar, if not more powerful per clock.

The C64x is a very interesting beast. It can run up to 8 instructions per cycle (no floating-point).
However it requires very powerful tools (optimizing assembler) which I am not sure are available unless you are ready to buy CodeComposer studio

 

[dockthepod]

Holy crap, this thing has a fast ARM, a dedicated GPU, AND a DSP coprocessor? It's going to be FUN to program for

 

[MWeston]

TRM not available. It should be SPRUF98

The TRM is broken up into many files under the User Guide section.

 

[ldesnogu]

Holy crap, this thing has a fast ARM, a dedicated GPU, AND a DSP coprocessor? It's going to be FUN to program for

I just hope you are not expecting talking directly to the GPU or the DSP, or you will be disappointed.
I am almost 100% sure the GPU will only be available by using OpenGL ES.
For the DSP, it's not obvious but my experience on other embedded platforms based on TI SoCs makes me fear we won't have direct access to the C64x

TRM not available. It should be SPRUF98

The TRM is broken up into many files under the User Guide section.

Thanks for pointing it! I had used the link found in one of the other TI documents, silly me

 

[goobers]

we are a special case and one of TI's pet projects so we get access to things before others.

Does this mean free marketing for Pandora essentially?

 

[flatmush]

QUOTE
I just hope you are not expecting talking directly to the GPU or the DSP, or you will be disappointed.
I am almost 100% sure the GPU will only be available by using OpenGL ES.
For the DSP, it's not obvious but my experience on other embedded platforms based on TI SoCs makes me fear we won't have direct access to the C64x sad.gif

This is a hobbyist platform, even if they don't give us the docs to mess around low-level then I'm sure someone will look into it. I know that I'm very interested in programming the GPU, by the looks of it opengl won't be able to access its full feature-set and it looks to be very good at multi-processing.

 

[Exophase]

TI has been announcing open source tools for developing for their DSPs, I'm sure that this as well as the low quantity availability of the OMAP35xx CPUs and their open declaration that they want to work with the open source community are all indications that the DSP will be available to us.

It'd be fun to program for it, maybe, just look out for the errata, it has some nasty bugs right now.

 

[ldesnogu]

TI has been announcing open source tools for developing for their DSPs, I'm sure that this as well as the low quantity availability of the OMAP35xx CPUs and their open declaration that they want to work with the open source community are all indications that the DSP will be available to us.

All I remember and could find was that TI granted Neuros the right to distribute c54x tools (ref). Nothing similar for c6x has been done, or at least I was unable to find any reference

QUOTE
It'd be fun to program for it, maybe, just look out for the errata, it has some nasty bugs right now.

You should have seen the kind of bugs the early c67 had. It could not load double float from memory :blink:

 

[hlidegp2x]

TI has been announcing open source tools for developing for their DSPs, I'm sure that this as well as the low quantity availability of the OMAP35xx CPUs and their open declaration that they want to work with the open source community are all indications that the DSP will be available to us.

All I remember and could find was that TI granted Neuros the right to distribute c54x tools (ref). Nothing similar for c6x has been done, or at least I was unable to find any reference


should this one be the right DSP found in OMAP3530 ?
http://www.google.fr/search?q=SPRU189&...lient=firefox-a

 

[ldesnogu]

should this one be the right DSP found in OMAP3530 ?
http://www.google.fr/search?q=SPRU189&...lient=firefox-a


Rather this: http://focus.ti.com/dsp/docs/dspsupporttec...ctName=spru732g

 

[Firefox]

ARM is big endian, DSP is little endian.

Yes, it does say that doesn't it?

Strange, I've never seen a ARM that wasn't running in little-endian mode...

 

[ldesnogu]

Strange, I've never seen a ARM that wasn't running in little-endian mode...

Most Xscale's (all?) are BE

 

[Exophase]

should this one be the right DSP found in OMAP3530 ?
http://www.google.fr/search?q=SPRU189&...lient=firefox-a


Rather this: http://focus.ti.com/dsp/docs/dspsupporttec...ctName=spru732g


Great, the instruction set/architecture is totally documented, so I don't see what will stop compilers or at least assemblers from being made, if they aren't already available.

If performance of GCC on Blackfin is any indication I'm thinking we might want to write ASM for this one.

 

[Firefox]

Most Xscale's (all?) are BE

Ah, never knew that...