will there be a new pandora with a quad core?


cdoublejj

Still Fresh
Joined
Jan 8, 2014
Messages
12
will there be a new pandora with a quad core? or is pandora on the decline with so many android devices out now? if there was one with a tegra 4 or quadcore snap dragon or something pandora would be o nthe leading edge again due to it's battery life and controlls.
 
will there be a new pandora with a quad core? or is pandora on the decline with so many android devices out now? if there was one with a tegra 4 or quadcore snap dragon or something pandora would be o nthe leading edge again due to it's battery life and controlls.
My guess is that it will not be a quadcore in the Pyra, but a dual core OMAP5 from TI. Nvidia, Qualcomm and Samsung are not interested in selling their SoCs to small-scale projects like ours. OMAP5 would be just as cutting edge though, it has a Cortex-A15 just like the Tegra 4 and Exynos 5. The difference between 2 cores and 4 cores is not as big as it seems. Sure, 4 cores can get hot and drain your battery quicker, but you can only really get more performance out of it if you do lots of multitasking on it, or if you can run software on it that isn't bottlenecked by a single thread or by the GPU.

Pandora is not on the decline at all -- it can run Android too, but the fact that most people in the Pandora community prefer to use the full GNU/Linux is telling.

Pandora is still cutting edge because it still is the only of its kind: there simply aren't any other handhelds with a full keyboard and real gaming controls.
 
Will there be any ARM8 cortex-A53 soc's out aytime soon?

Maybe it would be an idea to make a concept for an arm 8 pyra? since we dont know yet when the pyra is due anyway, it probably wouldnt hurt to research the possibility of arm 8 in the pyra .

And no, the pandora doesnt decline, its actually sold out.

However its becoming increasingly harder to find parts to create new pandora,s.
 
Last edited by a moderator:
 ^^ from my perception pyra is already as much advanced as the pandora was when the preorder started... So 2 years from now :D (but I doubt ED will be open to large change on the project)
 
https://en.wikipedia.org/wiki/OMAP#OMAP_5

hhhmmmmm, well it is a dual core, and it's arm v7 and it's 1.7 ghz. would this compete with ayn current phones or tablets? the gpu looks likea mild at best stepup being a PowerVR SGX544MP2 @ 532 mhz vs PowerVR SGX530 @ 200 MHz

https://en.wikipedia.org/wiki/Pandora_%28console%29#Pandora_1_GHz

then again it could be mroe than i know about. what i'm interested in is gfx capabilities like this.

http://www.qualcomm.com/media/blog/2013/01/11/inside-snapdragon-800-series-processors-new-adreno-330-gpu

https://www.youtube.com/embed/us1tc-JKCIM?feature=oembed

it would also be ncie if it was inthe 300 USD range isntead of 500 USD, that or it would need some serious CPU/GPU power for 500 USD.
 
Last edited by a moderator:
First of all this device isn't trying to compete with current devices. If the OMAP5 is chosen it still would be a significant boost vs the current hardware.
 

it would also be ncie if it was inthe 300 USD range isntead of 500 USD, that or it would need some serious CPU/GPU power for 500 USD.
I don't think that is possible as this is a small niche developed and built by a small group of people, vs a big corporation that are selling in the millions..  Not to mention I can argue that a smart phone isn't cheaper than the Pandora especially in the US.. I know the "Data" plan alone out of my phone bill could buy a Pandora a year on top of paying $200 dollars upfront for it.
 
Last edited by a moderator:
Sure, 4 cores can get hot and drain your battery quicker, but you can only really get more performance out of it if you do lots of multitasking on it, or if you can run software on it that isn't bottlenecked by a single thread or by the GPU.
Outside of a very small margin cores that aren't used don't use power.
 
Sure, 4 cores can get hot and drain your battery quicker, but you can only really get more performance out of it if you do lots of multitasking on it, or if you can run software on it that isn't bottlenecked by a single thread or by the GPU.
Outside of a very small margin cores that aren't used don't use power.
Sure, but what's the point of having extra cores if you don't use them?

My point is not that you cannot keep 4 cores somewhat busy. My point is that you can rarely get a linear speedup from parallelization, so 4 cores will in general not at all be 4 times as fast as 1 core or twice as fast as 2 cores.
 
Sure, but what's the point of having extra cores if you don't use them?

My point is not that you cannot keep 4 cores somewhat busy. My point is that you can rarely get a linear speedup from parallelization, so 4 cores will in general not at all be 4 times as fast as 1 core or twice as fast as 2 cores.
Your original statement was still misleading. You said it'll use more power but you only get more performance in some cases. Strongly implying that you always use more power. You didn't say anything about the ratio of increased performance to increased power consumption.

You rarely get a linear speedup from parallelization, but that doesn't mean that the sub-ideal scaling is all lost in redundant work. If a core is waiting for another core to finish something it can be clock gated or possibly even power gated. In some systems its voltage can be dropped independently. The point is, it's more complex than saying distributing a task to N cores will use N times the power no matter what the performance scaling is.

On the flip side, distributing a task to more cores can use less power if the clock speed is lower. If you cut the clock speed in half you will generally use a lot less than half the power, at least if you were anywhere near the higher end of the clock range. This can work asynchronously to an extent. If you have one task that needs 1GHz and another that needs 0.7GHz it can be better to execute them on two cores at 1GHz, clock gating one of them some of the time, than one at 1.7GHz. This is especially true if the SoC needs especially heroic voltage scaling to hit their higher clock levels, as many SoCs these days have (Cortex-A15 was never really intended to go past 1.5GHz in phones)
 
I'd love to just see at least 4gb of flash so we can have a far more robust distro and utilities installed by default... we all like to dream right? :)
 
Sure, but what's the point of having extra cores if you don't use them?


My point is not that you cannot keep 4 cores somewhat busy. My point is that you can rarely get a linear speedup from parallelization, so 4 cores will in general not at all be 4 times as fast as 1 core or twice as fast as 2 cores.
Your original statement was still misleading. You said it'll use more power but you only get more performance in some cases. Strongly implying that you always use more power. You didn't say anything about the ratio of increased performance to increased power consumption.


You rarely get a linear speedup from parallelization, but that doesn't mean that the sub-ideal scaling is all lost in redundant work. If a core is waiting for another core to finish something it can be clock gated or possibly even power gated. In some systems its voltage can be dropped independently. The point is, it's more complex than saying distributing a task to N cores will use N times the power no matter what the performance scaling is.


On the flip side, distributing a task to more cores can use less power if the clock speed is lower. If you cut the clock speed in half you will generally use a lot less than half the power, at least if you were anywhere near the higher end of the clock range. This can work asynchronously to an extent. If you have one task that needs 1GHz and another that needs 0.7GHz it can be better to execute them on two cores at 1GHz, clock gating one of them some of the time, than one at 1.7GHz. This is especially true if the SoC needs especially heroic voltage scaling to hit their higher clock levels, as many SoCs these days have (Cortex-A15 was never really intended to go past 1.5GHz in phones)
You're right, as usual :)

However, what happens in typical idle situations? E.g. if you put a load of, say, 2% on all four cores clocked at 1GHz? Would that use the same, more, or less power than having a load of 8% on one core?

I would assume that in such a use case, more cores would still consume more power because of overheads/latencies of powering on and off. Maybe I'm wrong, or maybe the overhead is negligible.
 
What good is another bloody poll going to do?   We get whatever SOC ED can get,  and it seems he already has one locked in...

Do you really think a poll on how many cores the new SOC should have would be in any way useful? 

It makes sense to make a poll when there is something to be decided, or to gather some kind of useful information,  but I've never understood the why people insist on making a poll for the sake of making a poll...

- Neelix
 
*shudders*  clearly you are wrong,  as everybody includes me, and I most assuredly do not like useless polls. :p

- Neelix
 
please, no offtopic about polls in this absolutely essential P2 spam thread...

now getting back on subject: can we have a poll on the number of cores wanted? I'd want pi core myself...
 
Last edited by a moderator:
Back
Top