Overview of SoC options


_wb_

Microbe
Staff member
Joined
Apr 5, 2012
Messages
5,390
Age
42
Location
Brussels, Belgium
I know there are already a lot of threads about this, but I want to try to get a more or less accurate list of the actual options, and their main advantages and disadvantages. Feel free to add information or to correct this information. I'll try to keep the list up to date.

Intel Bay Trail-T (22nm)

  • CPU: 1.46 GHz (Turbo: 2.39 GHz) quad-core Atom Z3770 (Silvermont)
  • GPU: Intel HD Graphics (4 EU) @ 311 MHz (Turbo: 667 MHz)
  • RAM: LPDDR3 1067 (17.1 GB/sec)
  • Advantages: cutting edge SoC, probably very good performance and perf/W, compatibility with x86, good FOSS support, probably available
  • Disadvantages: not backwards compatible, notaz and some others are not interested in x86
Texas Instruments OMAP 5 (28nm)

  • CPU: 1.7 GHz dual-core ARM Cortex-A15
  • GPU: PowerVR SGX544MP2 @ 532 MHz
  • RAM: LPDDR2 532 MHz (8.5 GB/sec)
  • Advantages: backwards compatible, available, relatively good documentation and support
  • Disadvantages: no FOSS driver for GPU, it's a bit behind the cutting edge

Samsung Exynos 5420 (28nm)

  • CPU: big.LITTLE 1.9 GHz quad-core ARM Cortex-A15 + 1.3 GHz quad-core ARM Cortex-A7
  • GPU: ARM Mali-T628 MP6 @ 480/533 - 600 MHz
  • RAM: LPDDR3e 933 MHz(14.9 GB/sec)
  • Advantages: probably the best ARM SoCs at the moment (together with Snapdragon 800); GPU might have a FOSS driver at some point (lima)
  • Disadvantages: not available to us; 8 cores is probably overkill
Nvidia Tegra 4 (28nm)

  • CPU: 1.9 GHz quad-core ARM Cortex-A15
  • GPU: 672 MHz NVIDIA
  • RAM: LPDDR3 up to 933 MHz (14.9 GB/sec)
  • Advantages: should be mostly backwards compatible, state-of-the-art
  • Disadvantages: poor FOSS support, probably not available to us
HiSilicon K3V3 (28nm)

  • CPU: big.LITTLE 1.8 GHz dual-core ARM Cortex-A15 + 1.2 GHz dual-core ARM Cortex-A7
  • GPU: ARM Mali-T658
  • RAM: ?
  • Advantages: see Exynos (it seems to be rather similar), might be available to us
  • Disadvantages: poor documentation/support, probably not a very reliable company?
Rockchip RK3188 (28nm)

  • CPU: 1.8 GHz quad-core ARM Cortex-A9
  • GPU: Mali-400 MP (Quad core, 533 MHz)
  • RAM: LPDDR2, DDR3, DDR3L
  • Advantages: used in several devices already, GPU has a FOSS driver (Lima), probably available to us
  • Disadvantages: previous-generation CPU and GPU, poor documentation/support, probably not a very reliable company?
Qualcomm Snapdragon 800 (28nm)

  • CPU: Up to 2.3 GHz quad-core Krait 400
  • GPU: Adreno 330 @ 550 MHz
  • RAM: LPDDR3 933 MHz (14.9 GB/s)
  • Integrated Wifi/Bluetooth/GPS
  • Advantages: probably one of the best ARM SoCs at the moment (together with Exynos 5420); GPU might have a FOSS driver at some point (freedreno)
  • Disadvantages: doesn't seem to be available to us (please prove me wrong here!), probably not as low-power as big.LITTLE


For the purpose of comparison, here is what the current Pandora uses:

Texas Instruments DM3730 (45nm)

  • CPU: 1.0Ghz single-core ARM Cortex-A8
  • GPU: PowerVR SGX530 @ 200 MHz
  • RAM: LPDDR 200 MHz
and this is what the CC / ReBirth models use:

Texas Instruments OMAP3530 (65nm)

  • CPU: 600 MHz single-core ARM Cortex-A8
  • GPU: PowerVR SGX530 @ 110 MHz
  • RAM: LPDDR 167 MHz
 
good work thanks for this :)

So basically, 2 from unreliables companies, 3 not sourcables it left 2 :

- TI OMAP5

- Intel Bay trail (I bet this wont be sourcable either)

In summary, either ED manage to get some qualcomm stuff or we are stuck with TI ? Am I reading this right ? (still better than nothing at all ;P)
 
amlogic m802 (28nm)

  • CPU: Up to 2 GHz quad-core A9
  • GPU:Mali-450 @unknown MHz
  • RAM: LPDDR3? 64bit
  • Advantages: possible to obtain in relatively small quantities; GPU might have a FOSS driver at some point (lima)
  • Disadvantages: probably not as low-power as big.LITTLE, gpu no gles 3 support
Freescale i.mx6 (40nm)

  • CPU: Up to 1.2 GHz quad-core A9
  • GPU: vivante gc1000
  • RAM: 2x32bit ddr3
  • Advantages: can be bought in small quantities, sata available on some skus, gpu possible FOSS driver
  • Disadvantages: built on older 40nm process, low cpu clock, low gpu performance, relatively higher power draw
 
In before Monstercameron mentions AMD..
amlogic m802 (28nm)

  • CPU: Up to 2 GHz quad-core A9
  • GPU:Mali-450 @unknown MHz
  • RAM: LPDDR3? 64bit
  • Advantages: possible to obtain in relatively small quantities; GPU might have a FOSS driver at some point (lima)
  • Disadvantages: probably not as low-power as big.LITTLE, gpu no gles 3 support
Freescale i.mx6 (40nm)

  • CPU: Up to 1.2 GHz quad-core A9
  • GPU: vivante gc1000
  • RAM: 2x32bit ddr3
  • Advantages: can be bought in small quantities, sata available on some skus, gpu possible FOSS driver
  • Disadvantages: built on older 40nm process, low cpu clock, low gpu performance, relatively higher power draw
LOLOLOLOL

couldn't help but laugh

Ok, my take on all this, it would be cool if we made a P2 with one of the ARM ones to start with. Then, later, if people are interested, make one with a Bay Trail. The ARM would be the priority, but I will still like to later see an x86 one, just maybe not right off the bat.
 
_wb_, thank you for putting together the list.

good work thanks for this :)

So basically, 2 from unreliables companies, 3 not sourcables it left 2 :

- TI OMAP5

- Intel Bay trail (I bet this wont be sourcable either)

In summary, either ED manage to get some qualcomm stuff or we are stuck with TI ? Am I reading this right ? (still better than nothing at all ;P)
Buried in one of the other SoC threads, ED made a post that he had a relationship with one of Intel's listed Bay Trail suppliers.  I don't think we should discount that as a possibility until we know for sure.

Since Intel traditionally sells it's CPUs and SoCs through supply vendors rather than direct, it could actually be easier to get the Z3770 in the quantities needed than many of the alternatives.  That is something ED would need to check into if he's interested in the Bay Trail SoC.

Of the options listed, I think the Bay Trail is a great possibility.  It's as if Intel looked at the Pandora and built the perfect chip for a successor, just to see if we noticed.  They gave us everything we ever wanted in an SoC (bleeding edge, solid graphics, FOSS, 64bit, low power consumption, high performance per watt, cheap to buy @$37ea) except that it is X86.  Could they move us off of ARM?

No, I don't think that Intel cared a bit about the Pandora or it's community.  But wow - they pretty much nailed the need.
 
Last edited by a moderator:
I just say, use whatever is available to us, has a decent price, good support and allows 10hr+ gaming (you need to add power consumption imho as a criteria).

Oh, and whatever makes our devs happy. Maybe some chipsets are more "easy" to use than others?
 
I just say, use whatever is available to us, has a decent price, good support and allows 10hr+ gaming (you need to add power consumption imho as a criteria).


Oh, and whatever makes our devs happy. Maybe some chipsets are more "easy" to use than others?
I domt thinknwe really know the power consumption yet
 
good work thanks for this :)

So basically, 2 from unreliables companies, 3 not sourcables it left 2 :

- TI OMAP5

- Intel Bay trail (I bet this wont be sourcable either)

In summary, either ED manage to get some qualcomm stuff or we are stuck with TI ? Am I reading this right ? (still better than nothing at all ;P)
Altough I'm a little "underwhelmed" with the Ti option, I guess you can remove Intel Bay trail from the list :mellow:
 
good work thanks for this :)

So basically, 2 from unreliables companies, 3 not sourcables it left 2 :

- TI OMAP5

- Intel Bay trail (I bet this wont be sourcable either)

In summary, either ED manage to get some qualcomm stuff or we are stuck with TI ? Am I reading this right ? (still better than nothing at all ;P)
Altough I'm a little "underwhelmed" with the Ti option, I guess you can remove Intel Bay trail from the list :mellow:
Remove Bay Trail from the list based on what?
 
good work thanks for this :)

So basically, 2 from unreliables companies, 3 not sourcables it left 2 :

- TI OMAP5

- Intel Bay trail (I bet this wont be sourcable either)

In summary, either ED manage to get some qualcomm stuff or we are stuck with TI ? Am I reading this right ? (still better than nothing at all ;P)
Altough I'm a little "underwhelmed" with the Ti option, I guess you can remove Intel Bay trail from the list :mellow:
Remove Bay Trail from the list based on what?
No hard facts, just looking at the lack of interest Ed has shown in the x86 alternative => Ed is quite conservative if it comes to choosing new things (at least when its pandora related), so the probability is high that he won't deviate much from the current design.
 
I know there are already a lot of threads about this, but I want to try to get a more or less accurate list of the actual options

...
IMHO the first thing to be considered when looking at actual options is whether a particular SOC could be reliably sourced by a company like OP GmbH for the projected lifetime of the product.

Sad to say that it looks like you can cross Exynos & Tegra off the list if the viability of the supply chain is to be considered.
 
I just say, use whatever is available to us, has a decent price, good support and allows 10hr+ gaming (you need to add power consumption imho as a criteria).


Oh, and whatever makes our devs happy. Maybe some chipsets are more "easy" to use than others?
I domt thinknwe really know the power consumption yet
No, not from the future Handheld but maybe from all of these chipsets. There should be data sheets available that tell more about the power consumption of each chipset I guess. ;)
 
good work thanks for this :)

So basically, 2 from unreliables companies, 3 not sourcables it left 2 :

- TI OMAP5

- Intel Bay trail (I bet this wont be sourcable either)

In summary, either ED manage to get some qualcomm stuff or we are stuck with TI ? Am I reading this right ? (still better than nothing at all ;P)
Altough I'm a little "underwhelmed" with the Ti option, I guess you can remove Intel Bay trail from the list :mellow:
Remove Bay Trail from the list based on what?
No hard facts, just looking at the lack of interest Ed has shown in the x86 alternative => Ed is quite conservative if it comes to choosing new things (at least when its pandora related), so the probability is high that he won't deviate much from the current design.
I wish notaz will learn adjust to love x86 so Bay Trail could be used  :)   I'm not taking any sides (ARM vs x86), I'm just going with what's more power, availability, and support  :)
 
Last edited by a moderator:
Of the options listed, I think the Bay Trail is a great possibility.  It's as if Intel looked at the Pandora and built the perfect chip for a successor, just to see if we noticed.  They gave us everything we ever wanted in an SoC (bleeding edge, solid graphics, FOSS, 64bit, low power consumption, high performance per watt, cheap to buy @$37ea) except that it is X86.  Could they move us off of ARM?

No, I don't think that Intel cared a bit about the Pandora or it's community.  But wow - they pretty much nailed the need.
...  Perfect huh?  Let's analyze this claim.

The density of Aluminum is 2.7 grams per cubic centimeter.  A single cubic center worth of Aluminum would actually represent a rather significant heat spreader for a compact & passive system like we're talking about.

The specific heat of aluminum is 0.91 J/g*degrees Celsius.  If we assume an initial temperature to be a comfortable 24 degrees Celsius and a final temperature a uniform blister 70 degrees Celsius, then that requires 113 Joules per cubic centimeter of Aluminum.  Ignoring cooling options it would take 57 seconds at Intel's 2W SDP to heat the heat spreader to this temperature.

So let's account for cooling making some generous assumptions.  Let's assume that this cubic centimeter of Aluminum is setup in a thin 5cm flat square, with only the outward surface radiating heat in a meaningful way, and it is able to radiate heat perfectly as a black body.  Let's also ignore efficiency loss due to re-radiation from the environment for the moment.  Under these unrealistic idealized conditions the heat spreader would stabilize at around 72 degrees Celsius with 2 Watts.

In reality it would stabilize much higher due to the local environment becoming a heat trap resulting in massive efficiency loss to reradiation, nevermind the heat sink isn't a idealized blackbody and the area is on the excessively improbable side as a heat spreader, rather then an actively convection cooled finned heat sink ala the Shield.  Naturally an enclosed and effectively sealed box doesn't lend itself to convection, and case conduction isn't something you want to be betting on in the constraints of the normal thermal envelope, rather then its extremes as SDP is not the TDP.

intel-sdp-slide.jpg


In short the Z3770 should cook itself in around a minute in a Pandora form factor, and is almost but not quite perfectly unsuited.  Bay Trail-T is designed for tablets, not handhelds.  If you want to advocate Intel the Merrifield platform is what you should be talking about, which is set to arrive in 2014 and thus Intel's not currently in a position to really offer to anyone.
 
Last edited by a moderator:
Jebe there isn't an SoC on this list that won't consume > 2W under heavy load, and the most power hungry ones there are all being put in phones. There are known cases of games draining phones with 8+Wh batteries in well under 2 hours. Just because Intel doesn't want to make quad core SoCs for phones doesn't mean that they're very common amongst other SoCs, and as far as cores go Intel has the most power efficient one. A Pandora is better equipped to handle heat dissipation than a phone because its display isn't sandwiched against the rest of the thing.

Z3770 has a base speed of 1.46GHz. Passively cooled tablets can run all four cores at the max turbo of 2.4GHz for fairly lengthy periods of time. But clearly the reason why the base speed is so much lower than the max turbo is to give a broad dynamic range so it can handle lower power caps. The cores at 1.46GHz are going to use far less power than the ones at 2.4GHz. Using a dual core SoC doesn't help your perf/W at all, in fact it makes it worse under well threaded situations. It only uses more power overall if you allow it to, and this is most likely configurable (worse case scenario, it'll thermally throttle if it gets too hot, something a lot of phones already do)
 
Last edited by a moderator:
@ Jebe : You certainly seem to have a good grasp of thermal dynamics, if you could comment on the practicality of using no cooling solution for a current gen SOC in this thread it would be much appreciated.

Could you explain a little more as to why in your opinion a passively cooled SOC would work fine in a tablet / phone but not in a handheld mini computer?
 
Last edited by a moderator:
I wish notaz will learn adjust to love x86 so Bay Trail could be used  :)
What makes you think I don't know x86?
It's not that you don't know it, it's that you've stated previously that you wouldn't want to work your 'magic' in terms of firmware / OS / emulator development with an x86 based SOC.

As you don't seem to have much of an ego you might not realise how much your work is appreciated / viewed by the Pandora community. Without your input P1 would be nowhere near as awesome as it is today. This of course goes to all major contributors, but anyone who has followed the Pandora project will know that you have put in hundreds (possibly thousands?) of hours of hard work backed up with genuine expertise to help make the Pandora what it is today. The fact that this work has been unpaid and for a device with such a small user base makes it all the more awesome IMHO.

Your free time and what you do with it is of course 100% your own concern, I for one would have no problem whatsoever if you decided you didn't want to work on an ARM based P2. Obviously I'd really hope that would not be the case but I think anyone taking it for granted you would work as an unpaid volunteer on an ARM based P2 like you have on P1 would be 'bang out of order'.

From what I can see your lack of desire to work as an unpaid volunteer on an x86 based SOC for P2 is one of the main reasons why people would want to stick with an ARM SOC even if Bay Trail proves to be the 'best' choice, in terms of availability & sales potential, for a SOC for P2.

Please don't see this as a request that you reconsider your dislike of working with x86 as stated above, how you choose to spend your free time is nobody's business but your own. :)
 
Back
Top