possible soc updates.


monstercameron

Well-Known Member
Joined
Oct 19, 2010
Messages
1,001
Website
www.thinkteletronics.com
so, time has passed and we all have better knowledge of the possible tech that could power p2.


Tegra 3


Freescale i.mx6 quad


Exynos 4412


Allwinner a20


Qualcom snapdragon 400-600


They all have opensource drivers, are all powerful and have been producec for many months so supply should be there.


Right now, I have changed my position on tegra and would like to see this chip used in p2 even though the snapdragon 600 is as powerful as my beloved amd z-60's bigger 18 watt brother, the e-350, in both cpu and gpu bound benchmarks.
 
Last edited by a moderator:
Don't know for the other chips but Tegra 3 is feaseble if you would release a P2 now...i think that when P2 will be released (who know when..)

maybe Tegra4(just annunced) or 5 will be good to have.
 
Personally i would wait with the Pandora 2 until Arm8 chips are out, so that the improvements are significant enough to justify the high price of the Pandora.

I also think it would be good to have native support for Gles, OpenGl and DirectX, so an Arm8 Tegra might be nice.
 
Last edited by a moderator:
There will never be native support for Desktop DirectX or OpenGL on embedded arm chips I think.
 
There will never be native support for Desktop DirectX or OpenGL on embedded arm chips I think.
Tegra DirectX and OpenGL are different from desktop versions?

I didnt know that.

But at least a layer could be written then, that would add some level of support.
 
The gpu might feature all the stuff needed for directx or opengl, but the mobile driver most likely won't.


Edit: A layer for opengl is already in the works by lunixbochs.


There are some other implementations already but far from complete.


The powervr chip used in the pandora is also available on an atom board, where it has directx and opengl drivers, but only for x86.
 
Last edited by a moderator:
There is always the next best thing.. Let's keep chasing that dragon

Lots of the new SOCs are still using the last gen GPUs - that don't include opengl ES3.
 
Last edited by a moderator:
Id say the next Pandora would be at least 2 years out. So at least a Tegra 4 possibly the next gen after that.
 
64-bit is great and all but we don't even know when the first SoCs will be available. It might not be until 2015. So far the only one who has even confirmed a roadmap for a ARMv8 SoC release is nVidia with Tegra 6, and it's pretty much certain that won't be out before 2015.
 
Last edited by a moderator:
So why not produce a Myriad/P2 in 2016?
And sell Pandora 1 for another 3 years? Good luck with that.. the thing really needs a lot of things redesigned and it's getting harder to sell it today.. I'd sooner ask, why is is 64-bit so critical? I have 8GB of RAM on my desktop and laptop but I multitask way more than would be practical on Pandora-sized device. The thought of keeping hundreds of browser tabs and dozens of editor windows, terminals, etc all open on something with so much less usable screen real estate makes me cringe..
 
How about waiting for user opinion on the iCP2's controls, and for a suitable screen to be available?
 
How about waiting for user opinion on the iCP2's controls, and for a suitable screen to be available?
Is that a response to what I said or is it a general comment on this topic?
If you're responding to me, then why do you think Pandora 2 needs 3 years to get user opinion on iCP2 controls (assuming iCP2's designers will even be working on the Pandora 2 in question) and why do you think a suitable screen isn't available today?

If you want user opinion then the user base has already spoken via countless threads and polls over the last several months. The userbase, or at least the users here on this forum, want a device that's about the same size as Pandora 1, same basic controls but better shoulder buttons, at least one but preferably two full size SD slots, full USB, similar audio quality, HDMI or DisplayPort, ~1280x720+ screen with better response time, etc etc.

There could be remaining details to be discussed about keyboard layout (has that even been revealed for iCP2?) but it's not really that critical for most people and it's not something that would hold up development substantially.
 
Last edited by a moderator:
I was thinking that if waiting for ARM v8 to turn up would take too long (a reasonable viewpoint), then is there anything that would be worth waiting for?

Last time I asked ED, the copyright on the designs belongs to the hardware dev(s), so It would be feasible to use the iCP2 front panel if its any good.

I say that needs at least 6 months times 1000 users to make a good decision

Which screen(s) do you think are currently suitable?
 
I don't see all waiting as equal here. Something could be worth waiting 1 year for (2014), but that doesn't mean it's worth waiting 3 years.

DaveC owns the Pandora case design and AFAIK the iCP2 one, but if ED is doing a new Pandora I wouldn't automatically assume he wants to use either as a base. And let's say iCP2 based is on the table, what user input is really needed at this point? What has to be verified by extensive testing? The important parts of the controls, the d-pad and buttons and all, are already pretty much vetted.

There are lots of 1280x720-ish or even 1920x1080 screens around 5" out there, some AMOLED ones with fast response times as well.. I don't see any big new display breakthrough around the corner.
 
.. and since I didn't really comment on the opening thread, let me say this: going with anything made on TSMC 40nm or a similar process is outrageous. That's going to be two nodes behind by the end of 2014. Nodes transitions don't come around that often these days (pushing 2.5-3 years), but they mean way too much for perf/W to fall behind on. So that'd be the number one thing I'd qualify as worth waiting for, but more importantly than that, don't use one that's already behind.

So that rules out Tegra 3, i.MX6 and A20. I can't wrap my head around why you'd pick any of those over Exynos 4412 given similar availability - even if you saved $5-10 (and I doubt you would, look at how cheap HardKernel's 4412 boards are), there's no way it's worth it. Good job to nVidia somehow convincing people Tegra 3 is worth more than it is, no idea how they pulled it off, but it shouldn't be on anyone's list for a Pandora 2.

Of course ideally you'd do much better than Exynos 4412. My top pick right now is Exynos 5410, which is just now showing up in some Galaxy S4s. If that can work for, let's say a mid 2014 release, and it doesn't have some serious flaw or drive up costs tremendously then I say do it. Next choice is a Qualcomm Snapdragon S800 or S600. Both of these solutions offer a good amount of peak processing power plus some form of asynchronous multiprocessing and a good amount of GPU power. For Pandora 2 you want something that lets you run a big heavy thread and one or two weaker threads that can soak up background OS stuff, light emulation tasks like audio processing or screen filters, 3D API overhead, things like that. And you want cores that have good perf/W while under a low but steady load like you'll get running an NES emulator.
 
Last edited by a moderator:
No, I didn't forget the keyboard. I just don't think there needs to be this big back and forth on how it needs to be made. There wasn't for Pandora, there wasn't for iCP2, and I don't see a huge uproar over botched key technology. I trust whoever is making this to be able to make a reasonable decision based on some degree of limited testing. It doesn't need exposure to 1000 people.

I've seen a few displays on panelook that meet the specifications I gave, I'm not going to enumerate them. Maybe you think 1920x1080 is the requirement now, I don't, don't think there's a big community push on this, and don't think ED is latching on to it. So definitely not a driver, although yes, there are 1920x1080 displays available. I'm almost certain you'd get more community support going with a fast response 720p display than a slow response 1080p display. That doesn't mean you can't have both, but I've seen a lot of LCD datasheets and it's quite rare to find one with better response time than Pandora's - which is not good. For that reason AMOLED seems like a safer bet to me, but I can't tell how your particular choice stacks up because panelook makes you pay to look at datasheets.

Do you have to keep calling it Myriad? Until you get some confirmation that someone with interest and capability of actually making it wants to call it that it's kind of obnoxious..
 
Back
Top