Modularity and memory

[Grench]

IMHO - Get the currently planned edition of the Pyra out and sold and successful. Then consider selling a 'maxed out' version of the then existing SoC board with 8GB of RAM and a 256GB (or whatever is the highest capacity that will it then) eMMC as an upgrade board for $250+.

 

[FBnil]

Then consider s

I would wait until TexasInstruments comes with the successor of the current chip... the one that is twice as fast, and uses half the electricity... (Cortex-A57) THEN create a new daughterboard.


https://versus.com/en/arm-cortex-a15-vs-arm-cortex-a53

http://www.anandtech.com/show/7995/arm-shares-updated-cortex-a53a57-performance-expectations

disclaimer: it is actually 1.5 times as fast, but 2 sounds better

 

[Grench]

I would wait until TexasInstruments comes with the successor of the current chip... the one that is twice as fast, and uses half the electricity... (Cortex-A57) THEN create a new daughterboard.


https://versus.com/en/arm-cortex-a15-vs-arm-cortex-a53

http://www.anandtech.com/show/7995/arm-shares-updated-cortex-a53a57-performance-expectations

disclaimer: it is actually 1.5 times as fast, but 2 sounds better

Yes - assuming that the product sales support it, some day create a new daughterboard. This concept would not stand in for that eventuality. That will require a whole new multi-layer design, testing, iteration process, etc. Even starting this work is probably 2-3 years away.

I'm speaking to creating a maxed out daughterboard based on the then current designs that could be sold at a much higher price - after sales of the current version have established the product. I.e. minimal to no additional development work needed utilizing the exact same board production and population steps as the then current iteration of the production daughter board. This could happen 1-2 years after the current version gets released.

As far as current/upcoming production is concerned, 4GB of RAM and 32GB of eMMC is, frankly, plenty. I'm anxiously awaiting what they're putting the final touches on now. That product should keep us all quite busy for several years.

 

[jontheramer2]

I was just asking the question, but yes, the current version will keep me happy for years. I am not complaining about that and I ordered and will order a spare as well and I am recommending to my friends. Just saying, it could be less niche (aka more sales) with a bit higher specs as 'more normal' people expect that now (instead of in 3-5 years). But for us it will be fine, I am not too worried about that, it's just more future proof, for myself included with 8gb + 256 emmc on board.

 

[Letalis Sonus]

I would wait until TexasInstruments comes with the successor of the current chip...

Good joke. You do realize that the OMAP5 already is a canceled platform? The only reason why we didn't even get low power RAM is because the OMAP5 variant that supports it was never released, we're literally using the only OMAP5 chip that ever entered mass production. TI pretty much declared their ARM SoCs a dead end, if they do ever release an OMAP6 it will be an industrial embedded SoC that might as well be totally unfit as a replacement, TI does not develop for mobile scenarios anymore.

 

[Klumpen]

Good joke. You do realize that the OMAP5 already is a canceled platform? The only reason why we didn't even get low power RAM is because the OMAP5 variant that supports it was never released, we're literally using the only OMAP5 chip that ever entered mass production. TI pretty much declared their ARM SoCs a dead end, if they do ever release an OMAP6 it will be an industrial embedded SoC that might as well be totally unfit as a replacement, TI does not develop for mobile scenarios anymore.

Yep and we should be happy to get scraps since that's all we will ever get in such quantities and without a huge brand.

 

[FBnil]

You do realize that the OMAP5 already is a canceled platform?

I thought TI would not produce them themselves.... but still make a buck licensing it out to third parties like samsung and snapdragon? I then misread the post months ago... most tragic indeed.

 

[Pocak]

As memory is getting cheaper fast

What gave you that impression? Memory prices have been slowly rising for the past year.

I would pay substantially for that option.

Indeed. Napkin math says it'd have to be priced more than €100 higher than the 4GB version.

 

[EvilDragon]

TI's ARM department is anything but dead.
They have plenty of other SoCs for automotive and there are rumors they want to resurrect the DaVinci branch (which would be more consumer oriented SoCs, like the DM3730 used in the 1GHz Pandora (that's a DaVinci).

I guess with the internet of things-stuff becoming a major thing now, TI also wants to offer SoCs here.

Yes, it's far, far away until they could release a new low-power SoC suited also for mobile use, but it's not totally impossible.

 

[Letalis Sonus]

They have plenty of other SoCs for automotive

Mostly small scale microcontrollers and old stuff that lives off of the long availability guarantee that industrial and automotive applications require. Aside from a few mostly GPU-less OMAP4/5 knock-offs there hasn't been any new application class SoC since the OMAP5, their newest Cortex-A architecture is still the A15 and nothing newer has been announced so far.

I guess with the internet of things-stuff becoming a major thing now, TI also wants to offer SoCs here.

IoT devices are commonly implemented using microcontrollers, i.e. Cortex-M class chips that can't even run the Linux kernel due to their lack of an MMU.

 

[scft]

IMHO - Get the currently planned edition of the Pyra out and sold and successful. Then consider selling a 'maxed out' version of the then existing SoC board with 8GB of RAM and a 256GB (or whatever is the highest capacity that will it then) eMMC as an upgrade board for $250+.

Better not stick to the eMMC. There is UFS coming. More speed and total replaceability.
en . wikipedia . org/wiki/Universal_Flash_Storage

 

[slaeshjag]

OMAP5 doesn't support UFS anyway, so I don't see how it would be relevant to a "maxed out" OMAP5 board? (at least, I'm assuming that's what was hinted at.)

 

[scft]

OMAP5 doesn't support UFS anyway, so I don't see how it would be relevant to a "maxed out" OMAP5 board? (at least, I'm assuming that's what was hinted at.)

OMAP5 doesn't, but it is implied that there will be some kind of upgrade for the CPU.

 

[Grench]

OMAP5 doesn't, but it is implied that there will be some kind of upgrade for the CPU.

Could be if enough Pyra's sell to justify developing one. Could be, not will. It is not guaranteed, only possible - which is way more likely than upgrading nearly anything else.