SoC Poll - What's most important for you?

[TrashyMG]

It's interesting enough to see (has it to be proven ?) the non-linearity of the CPU Hz/V .

Whenever you want. I just did it on the OP:

1GHz@1GHz

./configure

time make

real 5m15

user 4m45

sys 0m15

I'd be curious to see your make -j1 and make -j4

Of course, i know it's a non-benchmark (not the same compiler, libs), but it gives an idea.

This wasn't compiled with any optimizations(neon) and such


cpufreq-set -f 500Mhz


real    4m29.961s


user    3m38.516s


sys     0m38.453s


Ran make clean


cpufreq-set -f 1000Mhz


real    2m22.887s


user    1m49.961s


sys     0m16.797s


Ran make clean


cpufreq-set -f 1500Mhz


real    1m32.000s


user    1m14.117s


sys     0m11.391s


Edit:


make -j1 @1500Mhz


real    1m43.865s


user    1m14.352s


sys     0m11.141s


make -j4 @1500Mhz


real    0m54.564s


user    1m20.766s


sys     0m12.781s


Edit2:

make -j6 @1500Mhz


real    0m51.775s


user    1m23.156s


sys     0m12.141s

 

[Linux-SWAT]

And it's the big cold wave... No need heat sink.

 

[WizardStan]

The OMAP5 running at 500Mhz compiled a little faster than the 1Ghz at 1Ghz (and 1/3 time at 1.5Ghz)... That certainly adds up with all the other numbers. The A80 is a beast that makes me think it's time to replace my ageing BeagleBoard, but for the Pyra I seriously think I'd be more than happy with an OMAP5 under the hood.

 

[Linux-SWAT]

It's interesting enough to see (has it to be proven ?) the non-linearity of the CPU Hz/V .

Whenever you want. I just did it on the OP:

1GHz@1GHz

./configure

time make

real 5m15

user 4m45

sys 0m15

I'd be curious to see your make -j1 and make -j4

Of course, i know it's a non-benchmark (not the same compiler, libs), but it gives an idea.

This wasn't compiled with any optimizations(neon) and such


cpufreq-set -f 500Mhz


real    4m29.961s


user    3m38.516s


sys     0m38.453s


Ran make clean


cpufreq-set -f 1000Mhz


real    2m22.887s


user    1m49.961s


sys     0m16.797s


Ran make clean


cpufreq-set -f 1500Mhz


real    1m32.000s


user    1m14.117s


sys     0m11.391s


Edit:


make -j1 @1500Mhz


real    1m43.865s


user    1m14.352s


sys     0m11.141s


make -j4 @1500Mhz


real    0m54.564s


user    1m20.766s


sys     0m12.781s


Edit2:

make -j6 @1500Mhz


real    0m51.775s


user    1m23.156s


sys     0m12.141s

Thanks !

On my x86 AMD Phenom 9500 Quad-Core Processor 2.2GHz

make -j1

real    1m3.819s


user    0m55.340s


sys     0m8.274s

 

[WizardStan]

On my i5-3570K at 3.4Ghz

time make -j1

real 0m20.650s

user 0m17.417s

sys 0m1.450s

But that's hardly unexpected.

For slight more "Apples to apples" comparison, my CC Pandora running at 500Mhz

...

"/usr/bin/sed: can't read =/usr/lib/libXext.la: No such file or directory"

Well that's not good.

 

[ShleeDragon]

We need to stop chasing the SOC dragon.. we will never catch up

 

[Natsu]

We need to stop bowing down to the weakest of the chips here and actually see if the other ones are

possible.

 

[klapse]

But snapdragon won't give a huge leap in performance over OMAP5. Ed and the team have OMAP devboards available NOW and could get down to working on the OS and designing the hardware.

 

[comradekingu]

Remember when we used to have only the difference in speed that we now think is important to pile on top available to us. Now there are people on lawns, and i dont know what.

Strongest chip to me is the most linux support and usability. To me pure power is like having range on a cannon, if you cant have accuracy point the argument on itself.

 

[Natsu]

But snapdragon won't give a huge leap in performance over OMAP5.

Except the fact that it does.

 

[comradekingu]

For what use-case? Secondary, with what optimizations?

Im sure the rifle is powerful, but if you end up shooting your foot because its heavy you have lost your aim. Show me how snapdragon doesnt have huge problems with proprietary dependencies, and linux as opposed to android support.

Edit: if you guys are running on pre 3.11 linux kernels without the fixes-omap5 commit and the board in question really is that uEVM one from the bugfix then its potentially damaging and all curves invalid.

 

[God Ginrai]

But snapdragon won't give a huge leap in performance over OMAP5.

What are you basing this on? For the record, I'm still supporting the A80 over the Snapdragon. However, that doesn't mean the Snapdragon doesn't deserve consideration.

-God Ginrai

 

[God Ginrai]

I think based on the reply I got from Allwinner, it would throw us at least 6 months back, which is way too much.

EvilDragon, I have to say that I'm not really sure how the A80 could set you back 6 months.

I just found this video from CES of an interview with the AllWinner spokesperson that was at CES:

https://www.youtube.com/embed/kuCeKjZ7-8Q?feature=oembed
He talks a little bit about the A80 and the Optimus at the beginning, but the bulk of the information about them starts ~4:27.

Not only does he confirm right near the end (~6:35 is where it starts) that they are shipping the OptimusBoard in Q1 2014, but he also claims that they expect to have customers shipping devices with the A80 in it within the first half of this year.

At the end, when asked about if anyone could buy the OptimusBoard or only special customers, he said "We have quantities allotted for early customers, but if anyone is interested in these we're more than happy to start talking to people immediately."

He even comments ~6:30 that this will be a "dual architecture for android and other OSes". That sounds to me like Linux support could be very likely.

Also, for those wondering, he confirms that the A80 is using the PowerVR 6 series with OpenGL ES 3.0 at ~4:49.

With these specs, and this time table, I honestly believe that we couldn't have a better option for the SoC.

-God Ginrai

 

[ShleeDragon]

The A80 is the perfect SOC on paper.. Quad core A15 and Quad core A7! big.LITTLE! low power! PowerVR6! GLES3! linux drivers! cheap! OOOOOOH GOD I WANT IT NOW!

but we don't want bleeding edge. Let alone bleeding edge with a 6 month wait time and limited support from day 1...... this is risky. beautiful. attractive risk.

1) What's the Failure Rate and Confidence of this product?

2) Does the allwinner have good linux support on current SOCs? I mean good support with mainline kernels, not "hacked togethered custom kernel" in github mostly done by Android kang mod developers.
Right now.. if we wanted to use the A20... is that supported and useable?

 

[God Ginrai]

The A80 is the perfect SOC on paper.. Quad core A15 and Quad core A7! big.LITTLE! low power! PowerVR6! GLES3! linux drivers! cheap! OOOOOOH GOD I WANT IT NOW!

but we don't want bleeding edge. Let alone bleeding edge with a 6 month wait time and limited support from day 1...... this is risky. beautiful. attractive risk.

Like I said, I don't believe that it is a 6 month wait time. Considering they are shipping the OptimusBoard Q1 2014, (read: any time now) then it wouldn't have any longer of a wait time than the Snapdragon would.

Also, who said we don't want bleeding edge? The OMAP 3430 was bleeding edge when it was picked for the Pandora. Also, the original reason EvilDragon settled with the OMAP5 was because Samsung wouldn't give him their bleeding edge SoC.

-God Ginrai

 

[ShleeDragon]

The A80 is the perfect SOC on paper.. Quad core A15 and Quad core A7! big.LITTLE! low power! PowerVR6! GLES3! linux drivers! cheap! OOOOOOH GOD I WANT IT NOW!


but we don't want bleeding edge. Let alone bleeding edge with a 6 month wait time and limited support from day 1...... this is risky. beautiful. attractive risk.

Like I said, I don't believe that it is a 6 month wait time. Considering they are shipping the OptimusBoard Q1 2014, (read: any time now) then it wouldn't have any longer of a wait time than the Snapdragon would.


Also, who said we don't want bleeding edge? The OMAP 3430 was bleeding edge when it was picked for the Pandora. Also, the original reason EvilDragon settled with the OMAP5 was because Samsung wouldn't give him their bleeding edge SoC.

-God Ginrai

ooooooooh  Did the advertising and public relations guy in the suit tell you they were shipping something soon? I'd never expect to hear bullshit from somebody selling something... until it's for sale in our hands. It's vapour!

also, the OMAP5 is an excellent SOC. It's pretty damn good..... and that aguement makes no sense - I doubt Samsung would have given Craig the SOC when the Pandora 1 was being designed!

 

[comradekingu]

http://www.allwinnertech.com/en/product/A31.html  this lists android and windows.

http://www.allwinnertech.com/en/devkit/ProA31Kit.html supposedly has software:android sdk  and it links to http://www.wits-tech.com/  im finding nothing there.

Edit:  Doc:  Full Doc     > < Wheeeeeeeere?

Its a powerful toy running toy operating systems, prove me wrong.

Edit2: Support on http://linux-sunxi.org which belongs to arm-netbook community...

mv /usr/lib/arm-linux-gnueabihf/mesa-egl/ /usr/lib/arm-linux-gnueabihf/.mesa-egl/Awkward, but at least gets you something workable.

 This is an actual quote from http://linux-sunxi.org/Binary_drivers   I dont even...

The sunxi kernel mainlining effort is a massive task, and using upstream code today is very much a developer only undertaking. The upstream code does not support NAND, MMC or SATA, you can only boot off a USB harddisk. Similarly, it lacks sound, display, 3D, 2D and media support. If you simply want a more-or-less usable device now, then use our own 3.4 branch.

The Mainline Linux 3.14 kernel has SMP support for Allwinner Sunxi platforms. I have much respect for the allwinner community guys, but i would be suprised if allwinner became a serious company with chips on offer 6 months from now.

Imo freescale i.MX 6 or marvell armada 1500 make _much_ more sense.

 

[ShleeDragon]

^^^^^^^ sounds like a rabbit hole of KANG'ED source to me.

Supports Microsoft Windows 8: Why can't I hold all of these WATS!

 

[God Ginrai]

ooooooooh  Did the advertising and public relations guy in the suit tell you they were shipping something soon? I'd never expect to hear bullshit from somebody selling something... until it's for sale in our hands. It's vapour!

 We've been looking into the Snapdragon not even knowing yet if Qualcomm will give us one.

also, the OMAP5 is an excellent SOC. It's pretty damn good..... and that aguement makes no sense - I doubt Samsung would have given Craig the SOC when the Pandora 1 was being designed!

My argument makes sense, you just didn't bother to read it. I said that the reason EvilDragon settled with the OMAP5 was because Samsung would not give their SoC to him. That was obviously a comment about the Pyra, and had absolutely nothing to do with Craig.

-God Ginrai

 

[ShleeDragon]

You're also not a fan of facts. The OMAP5 was the first SOC that ED was able to source (and he sourced it quickly after discussions with TI)..... Once the OMAP5 was secured, he was able to greenlight the project as a whole...

There is nothing wrong with shopping around for something better but he didn't choose it because he got rejected from another source. That's crazy talk.

The OMAP5 isn't unattractive sister you only took to the prom because you got rejected by your crush.