Tegra K1


ED have one confirmed. He just hasn't told us about it yet. Although I think it's pretty safe to say by now that it's an OMAP5 that we will see in Pyra.
 
You are aware that there's there's not really a current-gen SoC which has an OpenSource Driver available?


And that TI offers both Linux and Android drivers for the OMAP whereas most other current SoC vendors only offer Android drivers?


Is there any other current-gen SoC where you have full Linux support available?


Links?
The great defender of the OMAP 5.

:p
 
no, the logic is solid, I'm not arguing that it's probably the best choice given the options available as far as what can be obtained. I'm just disappointed at the only option(s) available are lackluster. I thought we were in 2014, not 2012 is all.

If the options were balls to the wall performance and shitty support and awesome support and lackluster performance I'd choose lackluster performance. But right now it's shitty support and lackluster performance. See what I'm saying?

rockchip rk32xx series is supposed to have linux and android support and mali gpu which is friendlier for open source development

tegra 4 is also has active linux support (I've booted linux on my shield)
 
Last edited by a moderator:
2012? The OMAP5 is a CortexA15, which is the same as all other current SoCs... Big.little would have been neat, but that's more about power saving, not horsepower.

Rockchip is horrible when it comes to design guides, and the Rk32xx is a CortexA12 which is slower than an A15... So you'd rather have something slower...?
 
2012? The OMAP5 is a CortexA15, which is the same as all other current SoCs... Big.little would have been neat, but that's more about power saving, not horsepower.

Rockchip is horrible when it comes to design guides, and the Rk32xx is a CortexA12 which is slower than an A15... So you'd rather have something slower...?
OMAP5 then? So much for my pipe dream of a Z3770. Oh well. Are you ready to sell me a Pyra yet?

How about now?

Now?

Can I pre-order it yet?

How about now?

Now?

How about in 2 Months? :)
 
Like I said in my middle paragraph, better support and lackluster performance to me is preferred to awesome performance and shitty support.

Finding something that is as open and supported as possible is best. Either officially or community. Getting an obscure chipset only found in cars and dev boards isn't going to be enough I feel.

Minor differences in specs aren't going to matter a couple years down the road, the support will though.

I was just poking fun, but I do remember hearing about omap 5 almost 2 years ago, even if not available till recently.
 
But support from TI is better than from rockchip, mediatek, Samsung or nvidia (unless you use an nvidia-certified design company).

TI has most of the documentation public and offers both Linux and Android sample OS incl. GIT for kernel and drivers publically.

Where is nvidias official kernel source code for Linux on the Tegra4?
 
So the only 'possible' chipset for Pyra is OMAP5 ? :unsure:

The Pandora chipset (OMAP3) when presented was much more competitive (i think) even 2 years later when the product became effectively available.

Now most of the products have quad core CPUs and better GPUs than the OMAP5 and a product that want to be appetible to tech entusiast/geeks (like us) and hit the shelves on end 2014/ begin 2015 need to offer a better deal IMHO.

Anyway i think i will buy the same the Pyra with OMAP5 (or even with a 6502 or z80), only because i don't wanna miss such a great community and you ED :)
 
Last edited by a moderator:
Now most of the products have quad core CPUs and better GPUs than the OMAP5
Like ED said, if you have leads on something better then please suggest it. If OMAP5 is the best that can be bought then there's no point in complaining.
 
Like ED said, if you have leads on something better then please suggest it. If OMAP5 is the best that can be bought then there's no point in complaining.
The best one that I know for sure can realistically be bought in small quantities is Z3770. That's obviously off the table, as we've known for some time. That said, OMAP5 does have advantages. It's the best solution in terms of backwards compatibility, easily. There's a realistic shot of PND software using the video scaler the same way they use it on Pandora, without having to do a lot of complex emulation in the device driver.
 
Like ED said, if you have leads on something better then please suggest it. If OMAP5 is the best that can be bought then there's no point in complaining.
The best one that I know for sure can realistically be bought in small quantities is Z3770. That's obviously off the table, as we've known for some time.
Yes - it would have been so cool. Sadly yes, things do seem to have gone another route.
 
Where is nvidias official kernel source code for Linux on the Tegra4?
I believe it has been merged into mainline

edit:

https://github.com/torvalds/linux/tree/master/arch/arm/mach-tegra

tegra 4 = tegra114
That is good and all, but only if they would sell their SoCs to people like ED.
Like I said before, I'm not questioning the logic behind the decision on the SoC that was chosen. I completely 100% understand the thought process and why it's basically the only option. I'm just disappointed in what the options are what they are. I was assuming the cards would have been dealt a little bit better given the time since the first design was chosen.

-linux support

-availability in quantity

-reputable manufacturer

-performance

-open source as possible

it was probably in that order too. I just frown on powervr as I've heard way too many people complain about how it's so hard to reverse engineer it for open source purposes. Tegra and mali have active community driven open source gpu driver development, nvidia is even helping out where they feel they can.

http://www.phoronix.com/scan.php?page=article&item=nvidia_tegra_3d

But he asked a specific question, I gave the answer to that question. Now if that answer changes anything... I don't know, but it's what he asked for.
 
Last edited by a moderator:
Well the K1 with dual core 64bit armv8 will be out in products around mid this year, making the OMAP5 at least 1 gen old. By the time the Pyra launches - hazards a guess - next year, it could be 2 gens old.   But if that's the only chip ED can get, well it's disappointing as SNESFAN said, but what can can you do - it is what it is. 

I'm just really concerned about OMAP5 from a different perspective - longevity and support. I know they have moved away from mobile SOC's and will focus on vehicle system SOCs , but that space is also becoming very competitive .  I don't have a good feeling about their mid term prospects.

ED - You may as well officially say the OMAP5 is what you have at the moment.  I think the cat is well and truely out of the bag now.
 
Last edited by a moderator:
ED - You may as well officially say the OMAP5 is what you have at the moment.  I think the cat is well and truely out of the bag now.
Unless that's all part of the master plan to mislead us until he's contractually able to announce that he's sourced the Intel Z3770.

Seriously, there could be contractual reasons to not 'verify' things like that until certain dates or milestones are met.
 
Well the K1 with dual core 64bit armv8 will be out in products around mid this year, making the OMAP5 at least 1 gen old. By the time the Pyra launches - hazards a guess - next year, it could be 2 gens old.   But if that's the only chip ED can get, well it's disappointing as SNESFAN said, but what can can you do - it is what it is.
Umm, generations in smartphone business tend to change gen each 8 months these days...

So would you like to spend 600 EUR for a new Pandora-successor each few months?

BTW: Does anyone say that the VITA is already 2gens behind, as it uses an old CortexA9 SoC?

Nope, no one does, because no one compares the VITA to a smartphone. Why the heck does ANYONE compare the Pandora / Pyra to smartphones?

That's just plain stupid.

The Pandora is now 6 years old technology, and the last year has been amazing with software releases.

It runs a lot of stuff that has been optimized way better than my Cortex A9 tablet, even though the Cortex A9 is one generation further.

What do you EXPECT of a Pandora successor?

The latest available technology that will be way better than that of huge companies that produce their own SoCs?

Sorry, that was never the point of it, and whoever thinks something like that should probably leave right away, as that will never ever ever happen.

The Pyra should have a lifetime of at least 5 - 6 years, so whatever we choose and release will be last-gen half a year later anyways.

I'm just really concerned about OMAP5 from a different perspective - longevity and support. I know they have moved away from mobile SOC's and will focus on vehicle system SOCs , but that space is also becoming very competitive .  I don't have a good feeling about their mid term prospects.
Automotive/Industrial has ALWAYS been a WAY bigger part of TI than the mobile part.

The OMAP3 has a 10-year-availability rate, TI ensures you can buy it until at least 2018.

The OMAP5 should be similar, because that's the most important thing in the industrial business.

Mobile stuff can be EOL anytime, which is death for such small companies as we are. I can't simply stockpile a few tens of thousands of SoCs in case they would be EOL suddenly...
 
Back
Top