no, this is not the 1st of april ... x86 beats ARM at power efficency in Tablets

[Movieman]

Performance was not a issue to crack, but now that x86 with Atom is more power efficient than ARM, it is good enough that I come out of my hole to share

Ok, fair enough but wrong type..Now about these new IB xeons....

[Tony]

Anand often jumps in with graphs like this and overlooks other variables. The batteries would need to be standardised for the tests, same screens, etc etc then decide which is most efficient.

Its nothing to do with ARM CPU's being less efficient, its all to do with the Samsung platform they tested being very efficient overall.

Apples to apples people, anything else is just wasted editorial and confuses things...(when comparing CPU's etc )

[highoctane]

If nothing else it shows x86 can be power competitive with ARM in certain scenarios, who would have made that kind of argument before.

[PedantOne]

heh, Atom, effective? - No way! Still in order slow c*ap, sorry. Like Tony said, we need apples to apples comparison.

[kl0012]

heh, Atom, effective? - No way! Still in order slow c*ap, sorry. Like Tony said, we need apples to apples comparison.

Your post is funny. Why do you need "apples to apples comparison" if you have already decided that Atom is c*ap? And what if "apples to apples comparison" will not support your idea? It seems someone already entered a "denial mode".

[Manicdan]

theres many ways to do a fair comparison as long as things are mentioned and attempted to be compared in the exact why specified.

for example:
"best total battery life for similar priced and weight" would indicate that you can have 2 completely different platforms with way different specs, but whats important is price and weight. that would be apples to apples in my opinion.

another could be to compare with similar screen size and resolution while watching video. one might perform much better, but then also cost much more. such details would need to be pointed out.


its just way to easy for people to read a single chart and think they know the whole story, thats why marketing is 90% charts.

[Lanek]

Maybe it is better to wait the full review is online...

But looking this tab is running a full win8 version, sport a 11.6" 1368x768 screen ( PLS if im right ), 2gb of DDR + 64gb of storage.. seeing the battery life is looking really good with the new clover trail is an excellent surprise.

[kuroikenshi]

theres many ways to do a fair comparison as long as things are mentioned and attempted to be compared in the exact why specified.

for example:
"best total battery life for similar priced and weight" would indicate that you can have 2 completely different platforms with way different specs, but whats important is price and weight. that would be apples to apples in my opinion.

another could be to compare with similar screen size and resolution while watching video. one might perform much better, but then also cost much more. such details would need to be pointed out.


its just way to easy for people to read a single chart and think they know the whole story, thats why marketing is 90% charts.

I would agree that price is importanr, tbh I never cared for how much a device weights,

[Manicdan]

I would agree that price is importanr, tbh I never cared for how much a device weights,

my brother as an old asus netbook, its a bit thicker than these tablet/keyboard combos and a bit heavier. but for 2 year old tech and only 400$ that does things nearly identically as these with similar battery life, its kinda funny how much ascetics sell for.

[DeadlyFire]

So your here..Ok, there is a charge for doing the PR and that charge is some news on next gen Intel chips..

^ what this guy said

[zalbard]

A more interesting number would be performance per watt.

Its nothing to do with ARM CPU's being less efficient, its all to do with the Samsung platform they tested being very efficient overall.

CPU is a part of platform. So it has a lot to do with the CPU.
You won't see these CPUs sitting on the same motherboard, it is simply not possible. Best platform offer vs best platform offer comparison is perfectly valid.
As already mentioned, the devices have similar batteries, and x86 tablet has a larger display, so it's actually at disadvantage.
If you're trying to say that ARM platforms are inefficient, it's not like consumers can somehow swap things around to take advantage of a "seemingly more efficient" CPU.

[SKYMTL]

Anand often jumps in with graphs like this and overlooks other variables. The batteries would need to be standardised for the tests, same screens, etc etc then decide which is most efficient.

Its nothing to do with ARM CPU's being less efficient, its all to do with the Samsung platform they tested being very efficient overall.

Apples to apples people, anything else is just wasted editorial and confuses things...(when comparing CPU's etc )

Totally agree on these points. I think there is a whole lot of jumping to conclusions in this thread.

[haylui]

PR thread for Intel in the name of Anand??

[Spectrobozo]

if this numbers can be trusted http://www.anandtech.com/show/6340/i...dows-8-tablets
it looks quite competitive, and it have a huge advantage when it comes to software compatibility....
now what about the price of this SoC compared to the ARM competition?

[kl0012]

Some early owner review:

http://www.umpcportal.com/2012/11/re...art-pc-500t/1/

[BoredByLife]

Nice propoganda but what about the performance? Who cares if your battery last 100 hours but if you have to wait 10 hours before a page is loaded? Or things go shocky or slow?
Performance per watt is what defines tables imo. Not just Energy Efficiency

Also these remarks:
- different process node (probably intel 22nm compared to 28 or 32nm ARM cores which are based on generic proces nodes)
- if above, who says it is architecture related? It probably has nothing to do with x86... but all to do with a very good Intel manafuctaring process... to which arm has no access to.

[CrimInalA]

How can you compare these different processors/devices when they all have a different battery capacity ..

[onewingedangel]

Nice propoganda but what about the performance? Who cares if your battery last 100 hours but if you have to wait 10 hours before a page is loaded? Or things go shocky or slow?
Performance per watt is what defines tables imo. Not just Energy Efficiency

Also these remarks:
- different process node (probably intel 22nm compared to 28 or 32nm ARM cores which are based on generic proces nodes)
- if above, who says it is architecture related? It probably has nothing to do with x86... but all to do with a very good Intel manafuctaring process... to which arm has no access to.

Clovertrail is 32nm - the power saving is likely coming mostly from the extra sleep states and other power optimisations that have been developed to cover the whole product range, not just Atom. Clovertrail is just the first product to ship with some of these features.

Silvermont will probably be another big leap forward in performance and efficiency with a new Atom core design on a power optimised 22nm process - which I suspect will match/beat ARM in most if not all performance metrics.

[kl0012]

Nice propoganda but what about the performance? Who cares if your battery last 100 hours but if you have to wait 10 hours before a page is loaded? Or things go shocky or slow?
Performance per watt is what defines tables imo. Not just Energy Efficiency

Also these remarks:
- different process node (probably intel 22nm compared to 28 or 32nm ARM cores which are based on generic proces nodes)
- if above, who says it is architecture related? It probably has nothing to do with x86... but all to do with a very good Intel manafuctaring process... to which arm has no access to.

Here are some performance numbers (vs. Tegra 3):

http://www.anandtech.com/show/6385/m...face-review/10

[BoredByLife]

Here are some performance numbers (vs. Tegra 3):

http://www.anandtech.com/show/6385/m...face-review/10

Well, thats pretty bad is it... And + the huge size of the bloated win8 OS for tablets...
I'd prefer ARM for the moment. Especially since Tegra3 isnt the best chip...

[kl0012]

Well, thats pretty bad is it... And + the huge size of the bloated win8 OS for tablets...

Hm... How is that bad? 30-50% faster then Tegra3 and runs fully functional Windows8 (not pathetic Windows RT) while comparable in battery life and you call it bad? Then what's good?

I'd prefer ARM for the moment. Especially since Tegra3 isnt the best chip...

Tegra3 is one of the best. Quad core exynos is better, but as far as I know there are no WinRT devices using this chip. Another option is dual-core Krait which is slower then Tegra3 in multithreaded workloads.

[Hans de Vries]

Hm... How is that bad? 30-50% faster then Tegra3 and runs fully functional Windows8 (not pathetic Windows RT) while comparable in battery life and you call it bad? Then what's good?

Tegra3 is one of the best. Quad core exynos is better, but as far as I know there are no WinRT devices using this chip. Another option is dual-core Krait which is slower then Tegra3 in multithreaded workloads.

These benchmarks tell you exactly nothing at all about ARM performance versus Atom performance.
They may tell you that the Chrome browser is much faster as Windows 8 plus IE 10 in javascript
except for an intriguing SunSpider result,

KRAKEN (lower is better) single threaded javascript (jit compiler), 

 9733    Chrome     1.70GHz Cortex A15
14229    Chrome     1.66GHz Atom (N570)
33855    MS IE10    1.80GHz Atom (Z2760)
49595    MS IE10    1.30GHz Cortex A9

http://www.anandtech.com/show/6385/m...face-review/10
http://www.anandtech.com/show/6422/s...s-cortex-a15/6

The new Intel multimedia benchmark used by AnandTech: TouchXPRT 2013 is another brainchild
of dr Who's boss Shervin Kheradpir, General Manager of Intel's Performance Benchmarking and Analysis
Group and founding President of Bapco (via HDXPRT/Principled Technologies)

http://www.hdxprt.com/blog/2012/10/2...the-fast-lane/ AnandTech was the first to use the test
http://intel-mydreampc-1829796403.us...Whitepaper.pdf

The 1.3GHz Quad core A9 Tegra 3 comes out worse as the 1.8GHz Dual core Atom. Well for bandwidth limited
multimedia benchmarks that's not that hard to do considering the 32bit bus on the Tegra 3 versus the 64bit
bus on the Z2760, Apple uses 128 bit buses and Samsung's Exynos 5450 will have a 128 bit bus as well I guess.
Most of the other new SOC's use 64 bit buses.

Hans

[kl0012]

These benchmarks tell you exactly nothing at all about ARM performance versus Atom performance.
They may tell you that the Chrome browser is much faster as Windows 8 plus IE 10 in javascript
except for an intriguing SunSpider result,

KRAKEN (lower is better) single threaded javascript (jit compiler), 

 9733    Chrome     1.70GHz Cortex A15
14229    Chrome     1.66GHz Atom (N570)
33855    MS IE10    1.80GHz Atom (Z2760)
49595    MS IE10    1.30GHz Cortex A9

http://www.anandtech.com/show/6385/m...face-review/10
http://www.anandtech.com/show/6422/s...s-cortex-a15/6

It's really doesn't matter in this case since WinRT is mostly limited to IE (MS didn't open Win API to third party, so no Google V8 javascript engine is expected in the near future). Someone was worried about how fast Atom can load web pages in Win8 - so it's pretty much fast (faster than tegra3 in IE and a lot faster when using desktop chrome)

The new Intel multimedia benchmark used by AnandTech: TouchXPRT 2013 is another brainchild
of dr Who's boss Shervin Kheradpir, General Manager of Intel's Performance Benchmarking and Analysis
Group and founding President of Bapco (via HDXPRT/Principled Technologies)

http://www.hdxprt.com/blog/2012/10/2...the-fast-lane/ AnandTech was the first to use the test
http://intel-mydreampc-1829796403.us...Whitepaper.pdf

You mean the benchmark is crippled?

The 1.3GHz Quad core A9 Tegra 3 comes out worse as the 1.8GHz Dual core Atom. Well for bandwidth limited
multimedia benchmarks that's not that hard to do considering the 32bit bus on the Tegra 3 versus the 64bit
bus on the Z2760, Apple uses 128 bit buses and Samsung's Exynos 5450 will have a 128 bit bus as well I guess.
Most of the other new SOC's use 64 bit buses.

Hans

Doesn't matter either. The only cpus currently supported by WinRT are tegra3, krait and something old from TI.

[Drwho?]

as Usual, Hans forget to say that the benchmark is open source, and he failed to find anything wrong with it ... (This is call FUD technic)

And then, well, Hans, please disclose your interest into this please? like, who is employing you etc ...



...

At least, I am open on my agenda, and don't hide who is my employer.

[Hans de Vries]

Someone was worried about how fast Atom can load web pages in Win8 - so it's pretty much fast (faster than tegra3 in IE and a lot faster when using desktop chrome)

The ARM Chromebook is twice as fast in Browsermark as this Clovertrail based system.

217,031 Chrome, dual core 1.7GHz A15 Exynos
101,644 MS IE10, dual core 1,8GHz Atom, Clovertrail

http://www.anandtech.com/show/6422/s...s-cortex-a15/6
http://hothardware.com/Reviews/Intel...review/?page=3

They simply don't intent to compete against the $250 Chromebook considering
the $800 which is asked here for a (20% slower) version using a 1.5GHz Clovertrail.
Or a similar Clovertrail system from Dell for $829...

So from <$300 Atom based netbooks we are now going to $800 Atom based systems
We'll see how this all ends...

Hans