The Intel Atom Thread

[Sweepr]

Zenphone 2 is very competitively priced, nice job ASUS + Intel. The base model (with Moorefield) costs ~$190, for that price you're getting 64-bit Lollipop out of the box and better CPU & graphics performance than Snapdragon 801. The top dog with 1080p screen and 4GB RAM costs ~$285.

Asus releases the Zenfone 2, slaps sub-$300 price tag on the 4 GB RAM model

ZE551ML (5.5″, 1080p display, 4GB RAM, 32GB) – 8990 NT$ (~US $285)
ZE551ML (5.5″, 1080p display, 2GB RAM, 32GB) – 6990 NT$ (~US $221)
ZE550ML (5.5″, 720p display, 2GB RAM, 16GB) – 5990 NT$ (~US $190)
ZE550CL (5″, 720p display, 2GB RAM, 16GB) – 4990 NT$ (~US $158)

www.whatsontech.com/2015/03/09/asus...-starting-from-4990-nt-dollars-us-158-dollars

 

[Sweepr]

AnandTech: Dell Venue 8 7000 Review

Whether a tablet's performance is acceptable or not will often be based largely on how much the tablet costs. Class leading performance can't really be expected from inexpensive tablets, and for expensive ones having class leading performance is a must. At $399, the Venue 8 is priced at the same point as the Nexus 9, and it trades blows with it across our different tests. With generational improvements to their processors, it's not hard to imagine Intel becoming a major performance leader in the mobile space. Performance isn't the only metric considered when a manufacturer is deciding on the processor for their device, but I wouldn't be surprised if we end up seeing more design wins for Intel in the future.

Going over the Venue 8's strengths beyond its design, we find that it does quite well in our CPU tests. The Intel Atom Z3580 manages to hold its own against the competition's high end ARM SoCs, and at this point the list of applications that don't run on Intel devices consists of only a tiny handful of Android NDK apps. Its speed also doesn't come at the cost of battery life, with the Venue 8 trading blows with the competition in our battery tests.

www.anandtech.com/show/9034/the-dell-venue-8-7000-review

Important reminder, Silvermont x86 cores still pack a punch among modern mobile CPUs. Snapdragon 805 was outclassed in all CPU tests and Moorefield even managed to beat Tegra K1 Denver in some tests (WebXPRT and Basemark OS II 2.0 Web). There's more to benchmarks than Geekbench.

This would have been a great mobile SoC for phones/tablets by mid-2014. Better CPU performance than S805, slightly faster GPU than S801 and 64-bit support before everyone else. Heck, same can be said about Cherry Trail-T. Intel needs to execute faster in mobile.

 

[witeken]

Too bad the amount of Moorefield devices compared to Snapdragon 800 series is negligible.

 

[Nothingness]

Important reminder, Silvermont x86 cores still pack a punch among modern mobile CPUs. Snapdragon 805 was outclassed in all CPU tests and Moorefield even managed to beat Tegra K1 Denver in some tests (WebXPRT and Basemark OS II 2.0 Web). There's more to benchmarks than Geekbench.

Yes, there's more than Geekbench, but WebXPRT is made by a company sponsored by Intel, PrincipledTechnologies. They even make "reports" commissionned by Intel. I can't trust them.

 

[Arachnotronic]

Yes, there's more than Geekbench, but WebXPRT is made by a company sponsored by Intel, PrincipledTechnologies. They even make "reports" commissionned by Intel. I can't trust them.

Someone could probably make some serious cash by writing a competitor to Geekbench 3...

 

[Nothingness]

Someone could probably make some serious cash by writing a competitor to Geekbench 3...

Perhaps. Not an easy task though if you want to do it properly (I know too much about benchmarks to even dare trying :biggrin.

I find Geekbench very useful and accurate (cf an old post about correlation with SPEC scaling) when properly used. But one should not use it as a single score (due to the infamous crypto bias) or without looking at other benchmarks.

BTW we should see the new SPEC CPU this or next year. It will likely have the same issues as all benchmarks for which source is available (compiler overtuning, e.g. icc), but will surely be interesting.

 

[Arachnotronic]

Perhaps. Not an easy task though if you want to do it properly (I know too much about benchmarks to even dare trying :biggrin.

I find Geekbench very useful and accurate (cf an old post about correlation with SPEC scaling) when properly used. But one should not use it as a single score (due to the infamous crypto bias) or without looking at other benchmarks.

BTW we should see the new SPEC CPU this or next year. It will likely have the same issues as all benchmarks for which source is available (compiler overtuning, e.g. icc), but will surely be interesting.

IMO, the best way to test performance is to use a wide array of real world applications. On PCs, it's pretty easy to do that, but on mobile devices it's much harder, which is why there is so much emphasis on these synthetic benchmarks.

 

[Nothingness]

IMO, the best way to test performance is to use a wide array of real world applications. On PCs, it's pretty easy to do that, but on mobile devices it's much harder, which is why there is so much emphasis on these synthetic benchmarks.

I agree that app testing is the best thing to do.

The problem with using real world applications is that many of them have been tuned for Intel chips for instance, probably making AMD CPU look worse than they are.

On mobile chips, you might have some ARM bias (though in my experience the contrary happens as seasoned devs are used to tune for x86 and don't know how to properly use gcc flags for ARM).

Pick the most used app on mobile device: the browser. You could not compare CPU's just using a browser for obvious reasons (not the same, version, OS specifics, etc.).

Difficult task

 

[teejee]

IMO, the best way to test performance is to use a wide array of real world applications. On PCs, it's pretty easy to do that, but on mobile devices it's much harder, which is why there is so much emphasis on these synthetic benchmarks.

I don't agree in general. If you test exactly the applications you use yourselves with similar workload then it is useful for you, but not for anyone else.
The problem with your approach is that you don't know how each applikation stresses the CPU, so the mix of applications/workload is most likely unbalanced.
There are literally billions of "real world usages".

 

[Arachnotronic]

I don't agree in general. If you test exactly the applications you use yourselves with similar workload then it is useful for you, but not for anyone else.
The problem with your approach is that you don't know how each applikation stresses the CPU, so the mix of applications/workload is most likely unbalanced.
There are literally billions of "real world usages".

Look at how the most-used apps within a given category perform.

 

[Sweepr]

Doesn't get any cheaper than this. To think that only 2 years ago $200 Windows tablets were a distant dream (just look at the first pages of this thread).

Meet a $48 Windows tablet

Want a Windows tablet but don’t want to spend a lot of money? In the United States your cheapest option is probably the Winbook TW700 which is available from Amazon or Micro Center for $70.

In China there’s an even cheaper option.The Ployer MOMO7W is a Windows tablet that sells for just 299 Yuan. That’s about $48 US.

...The Ployer MOMO7W features a 7 inch, 1024 x 600 pixel display, 1GB of RAM. 16GB of storage, and Windows 8.1 software. It also comes with a 1-year subscription to Microsoft Office 365. It has an Intel Atom Bay Trail processor, WiFi, and a microSD card slot.

While the screen resolution is disappointingly low, the tablet does have an HDMI port — which means you could plug it into an external display and use the tablet to stream videos to a TV.

http://liliputing.com/2015/03/meet-a-48-windows-tablet.html
https://mikecanex.wordpress.com/2015/03/12/the-us48-windows-tablet

Might not be the best tablet all around but you're getting quite a lot of performance and features for that price, 1/10th of an iPad Air 2 (base model, WiFi only).

 

[frozentundra123456]

I have the TW700, and while it is decent for the price, I really wish it had more ram. Even just idling on the desktop it uses about 50%.

 

[krumme]

Doesn't get any cheaper than this. To think that only 2 years ago $200 Windows tablets were a distant dream (just look at the first pages of this thread).

Meet a $48 Windows tablet

http://liliputing.com/2015/03/meet-a-48-windows-tablet.html
https://mikecanex.wordpress.com/2015/03/12/the-us48-windows-tablet

Might not be the best tablet all around but you're getting quite a lot of performance and features for that price, 1/10th of an iPad Air 2 (base model, WiFi only).

Cheap. Except for Intel and ms shareholders that is.
Shows how consumers value it. Next step pay user to use the products

 

[Idontcare]

Cheap. Except for Intel and ms shareholders that is.
Shows how consumers value it. Next step pay user to use the products

Are they are the AMDs of the mobile world?

 

[AtenRa]

1: Windows are free for sub 9" tablets
2: Intel BayTrail-T is no more expensive than ARM APUs due to Contra-Revenue
3: There are lots of ARM 7" Tablets at sub $50

1+2+3 = Intel Windows 7" Tablet at sub $50

 

[kpkp]

Benchmarks of the 4GB version of the Zenfone 2.

 

[Idontcare]

1: Windows are free for sub 9" tablets
2: Intel BayTrail-T is no more expensive than ARM APUs due to Contra-Revenue
3: There are lots of ARM 7" Tablets at sub $50

1+2+3 = Intel Windows 7" Tablet at sub $50

What does Apple charge for iOS8.3 for my iphone and ipad?

What does Google charge for Android XYZ?

The fact that Microsoft's business is broken relative to the 21st century world of commerce (where you charge for app ecosystem access and not for OS access) is Microsoft's problem, not the consumer's and not Intel's.

Intel may be subsidizing BoM for hardware, but at least their business model is 21st century (sell hardware, get revenue), but Microsoft's model is very much unlike their competitors in all the wrong ways.

 

[krumme]

Are they are the AMDs of the mobile world?

Yeaa you can say so. Mubadala is dumping big desktops cpu on the market. But at least they dont wrap them in dollar bills. But its the same happening in minor scale.

One can understand the usual user avoid win on tablets because its useless for their need - apps whatever. But its crazy how much Intel must lower their prices - literally pay oem - to use their product. Beats me - it looks like an average product - why pay oem for a product that seems midrange? i dont get the bom argument. Looking at the last year 4b loss we are -way- beyond that. The oem must simply avoid Intel products - or is it x86? - like a plague for some what strategic reasons ?.

 

[III-V]

Are they are the AMDs of the mobile world?

Intel would have to go to extraordinary lengths to tarnish their brand to the level of AMD.

There is a big, yet subtle difference between selling your products for less than you'd like because you can't make them cheap enough (Intel's case), and selling your products for less because your competitor(s) are more desirable due to their reputation (AMD's case). Additionally, Intel just wants money -- unlike Apple, they're okay with using their elite reputation to peruse non-elite markets. I think Apple's stance on that may change, and they'll start catering to developing markets as Samsung and others have begun to challenge Apple's dominance, but that's an idea for another thread.

Even if AMD had a superior cost structure, superior performance -- the manufacturing capacity, execution (and whatever else I'm missing) -- to beat Intel by a significant margin, they still wouldn't be better than Intel. It's a household name. It's like Coca-Cola, or Apple. Consumers are hooked to the Intel heroin, and there's just no substitute.

Sure, AMD's financials would be better, but the world would still feel like something's missing. It's extremely difficult to change the market's perception of your business, unless you do something phenomenally amazing or phenomenally stupid. AMD's claim to fame was being the budget alternative to Intel -- they were allowed to clone Intel's chips and become a secondary supplier to IBM. They were, and always will be, a distant second.

________

I think that's a big part of the divide between members here. There's no doubt there's quite the dichotomy in thinking on this forum. Some people despise Intel, because they underhandedly gained their position. Others love them, because -- in spite of their transgressions -- they've given people incredible products, time and time again. Really, though, I think there's a third group -- those that can see that Intel's "incredible products" are truly incredible sometimes, but other times they're only incredible because Intel wants you to think they are, despite their competitors having something comparable or better.

On the AMD side of the fence, some love them, because they seem to break the corporate self-interest stigma, and do some really altruistic things. Others have gotten so tired of their inability to produce competitive products that they've written them off entirely. A small few can recognize where AMD has done great things, but recognize the odds have always been stacked against them, and they are now more than ever.

I feel like a lot more of us are in the open-minded category than others may be willing to believe, and although there are a small handful of people on both sides of the fence that are truly toxic. They ruin this forum for themselves and for everyone else.

 

[VirtualLarry]

I used to be a fan of AMD. They used to offer competitive technology at very competitive prices to Intel. (My AXP 1800+ and board, and RAM I think, only cost me $140 at the time, for a 1.4Ghz CPU. Intel's PIII CPUs alone would have cost me more for the same performance.)

But unfortunately, lately, they've seemingly fallen off of the wagon. Bulldozer really didn't help things, and they have had to spend the last five years digging themselves out of the hole that they put themselves in.

I thought Thuban was OK, probably the last truely decent CPU that AMD made.

I'm looking at mini/nano-PCs now, and AMD doesn't have anything of value to offer in that space, due to their lacking performance / watt, and performance in that form-factor. I struggled for a year or two with their C-70 APU in a nano-PC, and that was enough. Finally built myself an ITX rig, with an H61 and G1610, MUCH MUCH better, for not a huge amount more power-consumption. (Ok, so I lost my USB3.0 ports.)

 

[Arachnotronic]

Are they are the AMDs of the mobile world?

AMD at least generates a positive gross profit margin for the PC chips they sell.

Intel's mobile group is still GM negative.

 

[kimmel]

AMD at least generates a positive gross profit margin for the PC chips they sell.

Intel's mobile group is still GM negative.

Why don't you compare the PC chips sales to the PC chips sales between the companies? Oh wait, it doesn't fit the narrative.

 

[III-V]

AMD at least generates a positive gross profit margin for the PC chips they sell.

Intel's mobile group is still GM negative.

At least Intel has strong potential to change that. The delays are disappointing, but all signs point towards profitability, even if it takes longer than Intel or its shareholders would like. They're missing out on some serious profits, but the picture's pretty clear, IMO -- it's not a question of whether or not they'll get there, but when.

Even if things got delayed even further, Intel is still in a very, very good position to become profitable. They are making serious progress with expanding their product portfolio, and increasing the competitiveness of those products. Even if everything were to get pushed out another 6 months (e.g., Broxton in 2H 2016 or early 2017), Intel would still be in a very good place.

At least for me, since I haven't quite had the savings to invest in Intel yet, I have more time before Intel's profits pick up. For everyone that's already invested though, the rate of return is certainly disappointing. When you take money out of the picture, and look purely at the technological progress side of things, Intel's delays are really frustrating. At the end of the day however, we can still take peace in the fact that things will move forward.

For AMD, we have no idea... wish they'd be more clear with the direction they plan to take.

 

[frozentundra123456]

I am less optimistic than you on intel's prospects in mobile (phones). Dont get me wrong, i would love to see someone put ARM in its place. Problem is, intel was late to the party, and 14nm needed to be on time, and a game changing product. Unfortunately, it was neither. I can see a market for Intel in tablets, and of course laptops and convertibles, because there is a place for x86 in those markets. Phones is quite another story. Even myself, who really hates android in a tablet or larger device, am satisfied with it in a phone. I know intel can run android, but in a phone native ARM/ android just seems like the obvious choice. I once thought intel could leverage their process advantage to make a far superior performance per watt chip that everyone would want to use it. However, the 14nm delays and mediocre improvements have me seriously doubting this.

 

[witeken]

However, the 14nm delays and mediocre improvements have me seriously doubting this.

A mediocre improvement?

Don't forget that 22nm is already superior to any foundry's process node. There's no competition for Intel when it comes to transistor performance per watt, and it will only get worse/better.