the open pandora 2 must use an amd apu!

[_wb_]

A Pandora can use up to 3.7W if you use everything at 100%: display backlight, wifi, cpu, everything. I estimate that about 1W comes from the cpu and gpu, about 1.2W from the display, maybe 1W from wifi and the rest from all the rest. When idle it uses about 0.1W, I have no idea how that breaks down to cpu and the rest.


So no, 3.3W for the cpu alone is not comparable to the Pandora. It's a number in the right ballpark, but still about 3-4 times as much.

 

[Binky]

3.3W for the cpu alone

That what I missed.


3.3W for the whole device, fine.


3.3W for the processor alone; not good.

 

[monstercameron]

Slightly refined, more things to see what is happening...

https://www.youtube.com/embed/rw2cxlbPOq8?feature=oembed

As for the other comments...why are exophases wasted on me?


_wb_ the amd apu would compare better to newer multicore socs like ti omap5 or exynos 5 in terms of power consumption...Just to point out but the a5x\a6 uses alot of power, so that and the scrren forced apple to double the battery in the ipad 3

 

[Binky]

Tablets are a different kettle of fish - more room for battery, more screen to backlight, other hardware the same as smartphones


Also, apple stuff is ludicrously thin already, so a double battery matters less than it would in a pandora.


I guess that a smaller proportion of the thickness of a tablet is a battery compared to a pandora.


Who says the battery doubling was entirely down to the processor, anyway?

 

[Exophase]

monstercameron, the things I said about power consumption reported by Joulemeter for GPU usage in games applies just as much to how much power the video playback engine uses. I wish I realized this and said so sooner before asking you to run more tests. Suffice it to say I am 100% confident that Joulemeter isn't telling you how much power the video decode engine is using. It's only telling you how much power is being used by the CPU, which is not a lot because it's not doing much work. It showing the same power consumption for every type of video you played should be a dead giveaway.


There's also a reason why the "base" part is always the same - it's because the power model doesn't correlate it with anything.


If you want a measurement of power consumption the best thing I can think of is either read the current draw off of the battery using a multimeter or run it from full load to nothing. This can only measure full system load, so I guess if you can do it with the screen 100% OFF you should try that.\


Your videos are really long so if you want to point me to something specific I'm missing could you please give me a time to go to?

_wb_ the amd apu would compare better to newer multicore socs like ti omap5 or exynos 5 in terms of power consumption...Just to point out but the a5x\a6 uses alot of power, so that and the scrren forced apple to double the battery in the ipad 3

The only time A5X uses more power than A5 is when you stress the GPU heavily, because it's so wide. But it probably still consumes significantly less than C-50's does at full load. It's obvious that the screen made most of the difference in iPhone 3 moving to the huge battery. On the other hand, for iPhone 5 Apple only increased the battery capacity by about 20%: they could do this because the screen is larger and they gain some by using a slightly more efficient (3.8V vs 3.7V) battery chemistry. But the SoC has a 2x faster CPU and GPU and more screen pixels to boot, yet I don't see people talking about iPhone 5's battery life being the < 2 hours it'd be while stressing a Hondo in a similar configuration.


Fact is, with stuff like Hondo AMD is at the disadvantage not just because they're using a CPU uarch that's intrinsically not that competitive in perf/W but because they're falling a processor node behind. Hondo is still on TSMC 40nm while OMAP5 will be TSMC 28nm and Exynos and A6 use Samsung's 32nm process.

 

[monstercameron]

Who says the battery doubling was entirely down to the processor, anyway?

the display aswell

Who says the battery doubling was entirely down to the processor, anyway?

how would i do this via battery drain? Run sper pi on it from full charge?

 

[Exophase]

A Pandora can use up to 3.7W if you use everything at 100%: display backlight, wifi, cpu, everything. I estimate that about 1W comes from the cpu and gpu, about 1.2W from the display, maybe 1W from wifi and the rest from all the rest. When idle it uses about 0.1W, I have no idea how that breaks down to cpu and the rest.


So no, 3.3W for the cpu alone is not comparable to the Pandora. It's a number in the right ballpark, but still about 3-4 times as much.

Both people who quoted that number said they had something on USB. Calling Pandora's LCD at full brightness "overkill" doesn't really begin to cover it. Suffice it to say that most of that power is dissipated behind the screen so it doesn't become a cooling headache to support like the stuff on the PCB.


If I wanted to give a very conservative number for power consumption on Pandora on "real" heavy use and not pathological conditions that no one's really interested in, I'd say about 8 hours of total run time. That puts it a little under 2W. I don't have a great way to estimate how that's split up but a pretty random stab would be something like 1W for the SoC, 0.5W for the display, and the rest for everything else.

 

[monstercameron]

there is also undervolting and underclocking which should reduce power consumption.


Alot of comparison between this chip and ~4yr single core, old omap3 in the p1...how does it stack up to newer socs? tegra3,omap5,exynos5,i.mx6 etc.


Also havent battery densities or price/density(makes sense?) gone up enough to compensate for higher power use over those from p1?

 

[Binky]

there is also undervolting and underclocking which should reduce power consumption.

So when underclocked (don't these things automatically underclock when under light loads?), does it make better use of each joule drained from the battery than an equivalent ARM SOC?

 

[monstercameron]

it underclocks but only to about 800Mhz -in windows it depends on the power mode- and I don't think is is effectively better than the comparative arm soc when it is also underclocked but sane undervolting should reduce power though...but if one is simply listening to music, browsing the web or watching a movie I feel it could hit ~4-6 hours on the pandoras current battery

 

[Exophase]

there is also undervolting and underclocking which should reduce power consumption.

You can reduce the clock speed and voltage of every mobile SoC around. Most of them will do it at least as well as AMD's. What difference does this make?


Unless you're saying OPT can hard-cap the clock speed and voltage below its rated value to try to trim down the TDP. It's not like you have this huge abundance of performance you can shave off here, 1GHz dual core Bobcat is already weaker than other options.

Alot of comparison between this chip and ~4yr single core, old omap3 in the p1...how does it stack up to newer socs? tegra3,omap5,exynos5,i.mx6 etc.

Well you tell me - which one can you find in phones and which one can't you? Now what's the max CPU performance for running in phones? If the phone's enclosure can handle the TDP that gives you a hard limit on how many watts it's producing.


And if Hondo could be put in a phone then at the very least AMD would be marketing it this way, even if there were no takers. Intel has been marketing Atom chips for phone usage for years now.

Also havent battery densities or price/density(makes sense?) gone up enough to compensate for higher power use over those from p1?

No. Battery capacity density hasn't gone up a lot, it barely goes up at all. Pandora's base is already pretty large so if there's density savings they'd probably be applied to making the unit thinner and lighter.


The maximum power consumption will probably go up with Pandora 2, and the battery life will probably go down - that's unavoidable. But the important thing is that it gets at least the same battery life while performing the same tasks. Would your Hondo solution? I can confidently say "no" when I see that running PS1 and N64 emulators use a good half the CPU time on both cores. There's no way that plus GPU is using < 1W.

it underclocks but only to about 800Mhz -in windows it depends on the power mode- and I don't think is is effectively better than the comparative arm soc when it is also underclocked but sane undervolting should reduce power though...but if one is simply listening to music, browsing the web or watching a movie I feel it could hit ~4-6 hours on the pandoras current battery

Oh, so apparently AMD is actually offering much worse DVFS (dynamic voltage and frequency scaling) than every SoC competitor I'm aware of. That's reassuring.


4-6 hours for light tasks on Pandora's battery sucks. No one wants that.

 

[klapse]

Mr Cameroon


Batteries are not improving significantly (in the purchasable batteries of past 4 years at least) -- so any increase in TDP needs to be matched by a physically larger battery to keep us at 10 hours usage.


We are also not going to run Windows as our OS, so the thread, while entertaining, is not very productive with regards to the Pandora2.

 

[monstercameron]

We are also not going to run Windows as our OS

didnt say it would be a windows only affair...this is a cut down c-50 hardly any differences...there is always the opensource drivers or the propietary drivers(fiddling maybe involved)

I get it that you guys love your arm processors, this is just for fun...its a spam forum afterall so play along with it pass on your critiques and opinions but stop bashing the idea of it...besides after seeing all those games I demoed how can you guys say that this has less performance than current gen arm parts?


another video on how powerful the gfx perf is...this is warface running on the cryengine 3...on a tablet!


http://www.youtube.com/watch?v=qP91dWe-c90

 

[Exophase]

Dude, I know you think posting a video of running some game is an airtight argument against EVERYTHING EVERYONE SAYS but it's not. We get the idea. No one needs further convincing on that one solitary point.


I'm really tired of your allegations that we're just biased (well okay, some of the people here are, but not the ones you're talking to right now). If there were an x86 SoC that could offer competitive perf, perf/W, features, and price I'd say go for it. There isn't, and as far as choices go I'd pick a Medfield or Clovertrail well before I picked anything AMD has announced. I can say Hondo has less CPU performance than Cortex-A15 SoCs because I know a lot about how both work and you don't, you just spam game videos and think that proves anything you want. Yes, the GPU is quite good. GPU isn't everything. You asked me how I thought the GPU would compare with OMAP5 for instance and I gave you a detailed explanation as to why I thought they'd be about equal. But you just ignored it like you ignore most of what I tell you.


I know this might seem like a contradiction but just because this is called the spam forum doesn't mean it's a license to spam. Maybe they should just rename the forum since some people aren't getting this.

 

[mcobit]

Well, those videos you are posting show stuff, that could easily be running on modern arm socs with the same performance.


It should only be a question of if the studios will port their engines to arm.


This of course is my opinion

 

[monstercameron]

dude I am not ignoring what you said its just that I am not such a good debater and the only move I know is to move with the punches, I am sure if I had a more technical understanding of this topic I could argue the perceived benefits to sacrificing the battery life to use this chip.

 

[monstercameron]

well their is unreal 3, it doesn't really look as good on mobile as on desktop -or tablet in this case- infinity blade looks good but devil may cry 4 looks better -that game runs 20-30 fps at 720 on my notebook.

 

[bismuthdrummer]

I do see merit in your arguments for x86, but I think Exophase summed up the community's fear about this:

4-6 hours for light tasks on Pandora's battery sucks. No one wants that.

The Pandora userbase at large holds preference to a long battery life, read the reviews on these forums. I think maybe one or two have suggested the battery life is too long or could be shorter... everyone else praises and adores the lasting charge they get.


Versatility (full USB, dual SDHC, Bluetooth, full desktop environment, touchscreen, keyboard & controls, et cetera) and portability remain the two hallmarks of this device. Not power or desktop compatibility. If we went x86 it would simply not be the Pandora 2 because of reduced battery life and well, incompatibility with Pandora 1.

 

[_wb_]

Just did some power consumption tests for the c-50 and the idle is~.03W, the web browsing is ~1.5W, hd video is (via uvd) .5W and demanding 3d game is ~3.3W

The Pandora 1 uses about 1.5W for the whole system, not just the cpu when you're playing a demanding game (something that keeps both cpu and gpu quite busy) at normal backlight settings (not max brightness).


So that's about as much as what your proposal uses just for the cpu when web browsing! (note that "web browsing" is a very ill defined thing but if it includes just viewing pages, not cpu hogging applets/flash, then it allows the cpu to go idle quite a lot).

 

[ghostpatch]

And more seriously do you know what the crazy Germans here put on popcorn? Sugar! Sugar I tell you. The mind reels.

One thing the Germans did right is making their popcon sweet! salty popcon baaahh!