I can almost, with 90% certainty saythat the p2 should use an amd apu!


I want all the wonderful people who have been working recklessly be known, and be respected.
Ok all you reckless souls... it's time to stand up and be counted!

... for so long you've been looked down upon for your casual disregard of consequences, your impulsive ill-concidered decisions...

Well, that's all about to change... It must be our perspectives that are flawed... for who are we to say that you aren't wonderfull and deserving of the utmost respect? 

Oh, and all you diligent and responsible workers...

GET BACK TO WORK!    ;) :lol:
 
As long as we get full drivers next time for the DSP & GFX I don't care what processor it uses.
 
As long as we get full drivers next time for the DSP & GFX I don't care what processor it uses.
The DSP is fully usable and doesn't need any drivers, but no one coded for it yet.


And there's no 3D GPU out there with full opensource drivers, except for an older Mali which is not being used anymore in new SoCs.


So for 3D it's not possible and for DSP not needed.
 
What...Exo, ED, & WizardStan don't qualify? o_O
I'll be the first to admit that I don't know what I'm talking about. :p I just did math from known numbers. Pandora's battery is 15Wh, a 7W CPU alone will suck it dry in (15/7 = 2) hours, even less if you factor in the wifi and LCD and the rest of the fiddly bits. The current 1Ghz Pandora with a full CPU load, full brightness, and wifi transferring maxes out at under 4W. That should be our target number: find a CPU, wifi chip, and screen which combined and running at approximately the same speed as the current Pandora (underclocked as needed) are around 4W.
 
And there's no 3D GPU out there with full opensource drivers, except for an older Mali which is not being used anymore in new SoCs.
AMD's GPUs (like the one in the SoC monstercameron so desperately wants us to use) have open source drivers available as an alternative. Intel's GPUs, which will be used in the BayTrail variants of the upcoming Silvermont SoCs, are fully open source.
 
I know the thread is almost dead  and I understand that the majority prefer arm for legacy reasons but...I must defends the apus honor!

http://www.amd.com/us/products/notebook/tablets/pages/tablets.aspx#4

Power projections based on calculations carried out by AMD Performance Labs measuring total system and individual component power at Windows Idle and under  various system loads while web browsing and/or viewing a 9:57 minute online video in h.264 format, viewed at 1080P setting at 100 nits.  The AMD “Larne” reference platform is projected to measure APU power at 1.2 W at idle, 1.40 W during web browsing, 2.35 W during video playback and .02 W during a system S3 “sleep” state.   Total system power for the reference platform is projected at 2.8 W at idle, 3.7 W during web browsing, 5.3 W during video playback and .07 W during a system S3 “sleep” state.   Battery life calculations were derived using a 35Whr battery pack at 98% utilization. The power projections are based on the “Larne” reference system with a configuration including the A4-1200 Dual  Core 1.0GHz APU, AMD Radeon™ HD 8180 series graphics, 2GB DDR3-1066 system memory and Microsoft Windows 8.
the apu during video playback is 5.3W, with max cpu & gpu utilization I doubt it would go over 8W...but look at that, the apu itself doesnt use that much power, so with flash memory, 2GB ddr3L 1066, a 5" display and a 15-20Whr battery, the battery life could go from 8hrs-2hrs under normal mixed usage, it might not be single core a8 levels of power usage but its not that far off besides how much power do you think exynos 5440, tegra 4, apple a6x or even a snapdragon s800 uses?
 
Last edited by a moderator:
I'm going to say it again and I'm going to say it as many times as it takes until you can give an actual answer: if Temash's power consumption is so low then why is this 11" laptop with 30Wh battery only lasting 3-4 hours in web browsing? Do you think Acer is using so much higher consuming parts for its laptop?

Even if we're taking AMD's marketing numbers seriously they're still not amazing. 1.2W at idle is unacceptably high and much worse than you'll get from competing SoCs. Web browsing numbers can be manufactured to represent pretty much any kind of load. Video playback numbers aren't that useful for estimating CPU power draw. Figures under known CPU loads are conspicuously absent. And 2x1GHz Jaguar is too weak all around.

The review you linked to before seems to think it's 3W for the CPU because it apparently uses 3W more under full load than idle (it's not especially clear since he only gave vague footnotes of a live demonstration that's not available). That's operating under the conclusion that the quad core uses almost nothing while idle. That's probably wrong.

How much power this needs at various CPU levels - near idle, very low loads, medium loads, high loads - is absolutely crucial. Exynos 5250 for instance uses a lot of power at 2x1.7GHz, more than I'm really comfortable with, but I'm willing to consider this SoC as approaching acceptable because it doesn't use very much power at around 2x800MHz. It's essential that Pandora 2 doesn't use much more power than Pandora 1 when performing the same kinds of tasks. The SoC needs to be able to deliver a medium load, mainly single threaded, at well under 1W, and it needs to be able to deliver idle at much lower, and the software needs to be able to manage this. It also needs to be able to offer much much higher single threaded performance while using a "reasonable" amount of power (< 2W).
 
Last edited by a moderator:
I'm going to say it again and I'm going to say it as many times as it takes until you can give an actual answer: if Temash's power consumption is so low then why is this 11" laptop with 30Wh battery only lasting 3-4 hours in web browsing? Do you think Acer is using so much higher consuming parts for its laptop?
Laptop screens need big backlights...
 
Laptop screens need big backlights...
You aren't following me. I'm comparing this laptop with this one. The screen size and battery capacity is identical.

BTW, DDR3L is still going to be worse for power than LPDDR2 or LPDDR3 that all of the mobile-oriented ARM SoCs support.
 
Laptop screens need big backlights...
BTW, DDR3L is still going to be worse for power than LPDDR2 or LPDDR3 that all of the mobile-oriented ARM SoCs support.
You mind to answer a question for me? I would take that look as a yes.

If CPU uses less power by underclock itself and bring down the voltage and current that's on it.

Can memory do the same thing when they are under very low load? Because, I think LPDDR 3 has a much lower working voltage than DDR3L.

If memory can't underclock itself like the CPU and GPU do, then X86 machines wiill always have much higher idle power consumption, right?

Also, How did you derived the single core soc medium and high load power consumption expectations? 1w on medium and 2 w on high seems a bit unreasonable to me.

Isn't this even lower than pandora's cortex A8?
 
exophase(gulp :unsure: ) I havent said that you or any one is wrong with their critisms...

here is an update http://www.umpcportal.com/2013/05/amd-temash-a6-1450-performance-acer-aspire-v5/

I'm going to say it again and I'm going to say it as many times as it takes until you can give an actual answer: if Temash's power consumption is so low then why is this 11" laptop with 30Wh battery only lasting 3-4 hours in web browsing? Do you think Acer is using so much higher consuming parts for its laptop?
the v5 was using an ssd, 4db ddr3 (not ddr3L/RS afaik) and 50% brightness on its ips screen, I dont know for sure what was amds draw or the rest of the system but 7-9W total system power browsing isnt too bad, although it is ~4W aboves amds larne test system with the dual core.

The review you linked to before seems to think it's 3W for the CPU because it apparently uses 3W more under full load than idle (it's not especially clear since he only gave vague footnotes of a live demonstration that's not available). That's operating under the conclusion that the quad core uses almost nothing while idle. That's probably wrong.
.I was watching the live stream, he wasnt using precision tools to measure power draw(windows tool perfmon) so I am also skeptical but he also uses perfmon on intel ultrabooks all the while so the numbers he gave must be in the ballpark.

How much power this needs at various CPU levels - near idle, very low loads, medium loads, high loads - is absolutely crucial. Exynos 5250 for instance uses a lot of power at 2x1.7GHz, more than I'm really comfortable with, but I'm willing to consider this SoC as approaching acceptable because it doesn't use very much power at around 2x800MHz. It's essential that Pandora 2 doesn't use much more power than Pandora 1 when performing the same kinds of tasks. The SoC needs to be able to deliver a medium load, mainly single threaded, at well under 1W, and it needs to be able to deliver idle at much lower, and the software needs to be able to manage this. It also needs to be able to offer much much higher single threaded performance while using a "reasonable" amount of power (< 2W).
ok, i have lost the perfromance and efficiency argument (objectively speaking) so now I will stress compatibility, foss, future proofing, support and legacy argument (which is too subjective to declare a winner, hence a war of attrition :rolleyes: ).

http://www.anandtech.com/show/6536/arm-vs-x86-the-real-showdown/7

also from anandtech a15 review

idle-a15-total.jpg


quad a9s tegra and snapdragon are ~3W idle, dual a15 is 3.5W -display may be to blame. amd larne -read somewhere it is a 13" display was idling at supposedly 2.8W and chippy, from his latest  post 8W idle -higher than the live stream

Test done on high-performance profile under battery power.

Test 2 – ALU

Standing usage immediately before test: 7.9W (7% utilization, ave 1Ghz clock in quiescent state)

Test max power drain (ALU Test): 12.3W (94% CPU load by Crystalmark Process)

Test power: 4.4W

Normalised: 790mw/Core/Ghz

Test 2 – FPU

Standing usage immediately before test: 8.1W (7% utilization, ave 1Ghz clock in quiescent state)

Test max power drain (FPU Test): 13.1W (94% CPU load by Crystalmark Process)

Test power: 5W

Normalised: 900mw/Core/Ghz
 
Last edited by a moderator:
You mind to answer a question for me? I would take that look as a yes.

If CPU uses less power by underclock itself and bring down the voltage and current that's on it.

Can memory do the same thing when they are under very low load? Because, I think LPDDR 3 has a much lower working voltage than DDR3L.

If memory can't underclock itself like the CPU and GPU do, then X86 machines wiill always have much higher idle power consumption, right?
LPDDR2 and LPDDR3 both use 1.2V vs DDR3L which is 1.35V. But they also have a bunch of other power saving technologies. I don't think that anyone is applying DVFS to memory.

Like usual it's not an x86 problem, Atom phone and tablet SoCs can currently use LPDDR2, but AMD's APUs can't.

Also, How did you derived the single core soc medium and high load power consumption expectations? 1w on medium and 2 w on high seems a bit unreasonable to me.

Isn't this even lower than pandora's cortex A8?
I'm talking about the CPU part of the SoC only. When you get 8 hours of battery life on Pandora that means that the entire system is using ~1.9W. Take away the display, sound chip/amp/speakers, PMIC (including efficiency loss), memory, display controller and various other peripherals on the SoC, and a few other components on the PCB that may not be negligible, and I think you'll find that the CPU itself uses well under 1W. Note that Cortex-A8 wasn't the most power efficient design to begin with, ARM's numbers for A9 suggest appreciably higher perf/W, and of course even DM3730 is 1-1.5 processing nodes behind what we can get now (OMAP3530 is another full node behind that).

To match Cortex-A8 @ 1GHz a single Cortex-A15 core @ around 500-700MHz is sufficient, and even Exynos 5250 can pull that off in under 500mW. So it can at least satisfy this.
 
the v5 was using an ssd, 4db ddr3 (not ddr3L/RS afaik) and 50% brightness on its ips screen, I dont know for sure what was amds draw or the rest of the system but 7-9W total system power browsing isnt too bad, although it is ~4W aboves amds larne test system with the dual core.
7-8W for web browsing, assuming similar actual loads, is bad vs the Exynos 5250 Chromebook which is something of an upper bound (power-wise) for the competing choices for Pandora. That's a few W higher. I can't find anywhere that says what kind of RAM is used in this Acer machine except that I have seen some claim that it's only DDR3-1066 (nothing about voltage). Chromebook is using DDR3L-1600 (disregard what Amazon says, it's wrong). I also found an unboxing that says the Acer machine is using an HDD, that can make an even bigger difference, but it shouldn't be that huge while running web browser tests because the storage should be mostly idle unless something really bad is happening. 4GB vs 2GB uses a little bit more power but nowhere close to 2x because you only actually access one stick at a time (unless they're on separate channels). I remember seeing reports from some papers that showed it was something like 20%. It'll also depend on if the memory is a single SODIMM, two SODIMMs, soldered on the board, or what. But none of this sounds like it should contribute to that much of a difference by itself.

quad a9s tegra and snapdragon are ~3W idle, dual a15 is 3.5W -display may be to blame. amd larne -read somewhere it is a 13" display was idling at supposedly 2.8W and chippy, from his latest  post 8W idle -higher than the live stream
Of course it's fair to expect a 10" display to use a few W. Also for Tegra 3 in Surface RT it apparently doesn't have good regulation of its GPU so it's always using too much when the display is on.

Standing usage immediately before test: 7.9W (7% utilization, ave 1Ghz clock in quiescent state)

Test max power drain (ALU Test): 12.3W (94% CPU load by Crystalmark Process)

Test power: 4.4W

Normalised: 790mw/Core/Ghz

Test 2 – FPU

Standing usage immediately before test: 8.1W (7% utilization, ave 1Ghz clock in quiescent state)

Test max power drain (FPU Test): 13.1W (94% CPU load by Crystalmark Process)

Test power: 5W

Normalised: 900mw/Core/Ghz
These numbers are all over the place. First it's 7-8W while web browsing, now it's 8W idle.. I'm having a hard time taking this seriously. I want real power measurements by reviewers who know what they're doing.

Anyway, this is following the still dubious practice of estimating CPU power consumption by taking peak minus idle, that might work well when your CPU uses several dozen W at full load and is known to have decent power savings at idle like an IB or Trinity. But I can't look at 8W idling for this platform and really believe the CPU is using almost. Besides that 7% average utilization isn't low enough for me to consider it really almost idle.

Check out this review of a different Acer Aspire v5 model (v5-171) with i5 and 37Wh battery, similar screen etc:

http://www.anandtech.com/show/6467/acer-aspire-v5171-notebook-review-the-death-and-rebirth-of-the-netbook/4

There are two interesting facets from this:

1) SSD + DDR3L improves power consumption by 13% at idle

2) This only uses about 6W at idle with the HDD + 6GB DDR3 (over two SODIMMs) configuration

Meaning that unless Acer is using much worse parts for this v5 (doubtful) either the idle power consumption is significantly worse or chippy's idle numbers are not really representative. Note that the E-350 chip gets by far the worst idle power draw here (8.5W); part of that is due to the larger 13.4" screen but maybe part of it is just worse idle performance for the CPU.
 
Last edited by a moderator:
yeah he was using an ssd during the live tests, he seems to be using the hdd in the current tests. the ram was reported by task manager as running at 800MHz -so 1600MHz effective-  and another peculiarity it seems that the apu might support ddr3L RS, which runs at 1.25v

http://www.amd.com/us/Documents/AMDGSeriesSOCProductBrief.pdf

FIRST GENERATION SOC DESIGN

>

Delivers up to 70% overall improvement over AMD G-Series APU

6

>

Integrates Controller Hub functional block as well as CPU+GPU+NB

>

28nm process technology, 24.5mm x 24.5mm BGA package

“JAGUAR” CPU CORE WITH PERFORMANCE INCREASES

>

Dual-core and quad-core, up to 2MB shared L2

>

113% CPU performance improvement over AMD G-Series APU

7

NEXT GENERATION GRAPHICS CORE WITH PERFORMANCE INCREASE OVER PREVIOUS GENERATIONS

>

20% compute performance improvement over AMD G-Series APU when running multiple industry-standard graphics-intensive benchmark

>

DirectX®11.1 graphics support

IMPROVED POWER SAVING FEATURES

>

Power gating added to Multimedia Engine, Display Controller & NB

>

DDR P-states for reduced power consumption

MEMORY SUPPORT: SINGLE-CHANNEL DDR3

> Up to DDR3-1600 – 1.35V and 1.25V voltage levels supported

> Up to 2 UDIMMs or 2 SO-DIMMs

> ECC support

INTEGRATED DISPLAY OUTPUTS

>

Supports two simultaneous displays

>

Supports 4-lane DisplayPort 1.2, DVI, HDMI™ 1.4a

>

Integrated VGA

>

Integrated eDP or 18bpp single channel LVDS

UPDATED I/O (FEATURES MAY BE SKU DEPENDENT)

>

Four x1 links of PCIe® Gen 2 for GPPs

>

One x4 link of PCIe Gen 2 for discrete GPU

>

8 USB 2.0 + 2 USB 3.0

>

2 SATA 2.x/3.x (up to 6Gb/s)

>

SD Card Reader v3.0 or SDIO controller
where did you find the power numbers for the sammy chromebook?
 
Last edited by a moderator:
Back
Top