Active Topics

 


Reply
Thread Tools
Posts: 60 | Thanked: 22 times | Joined on Feb 2008
#1
check out this http://www.appleinsider.com/articles..._platform.html

Intel is coming with a line of new super small super efficient CPU's for mobile computing applications. The smallest one runs at .8Ghz and consumes 0.7W. Their fastest consumes what TI's cpu in n800 does (2.5W roughly) but runs at 1.9Ghz some 5 times faster. I know clock rates are relative, software efficiency mean as much if not more, but the question remains why is Nokia is still chugging along on 3 year old technology from that technology laggard TI?
 
Bundyo's Avatar
Posts: 4,708 | Thanked: 4,649 times | Joined on Oct 2007 @ Bulgaria
#2
You may do a search before posting.
 
Johnx's Avatar
Posts: 643 | Thanked: 628 times | Joined on Mar 2007 @ Seattle (or thereabouts)
#3
Uhm-, so you're comparing the power draw of a cpu alone to the power draw of a system-on-chip. The Intel Atom still needs a northbridge/southbridge to even be mostly as functional as an omap. And I'll bet that even the *cheapest* Intel Atom alone costs more than Nokia pays for an OMAP2420. BTW, be sure to do some research and get some benchmarks before you *assume* that the Atom will even be as fast as your average ARM processor, clock-for-clock. Also, I'd be interested to see where you got the power consumption figures for an omap2420.

-John
 

The Following User Says Thank You to Johnx For This Useful Post:
Posts: 5,335 | Thanked: 8,187 times | Joined on Mar 2007 @ Pennsylvania, USA
#4
Originally Posted by Johnx View Post
so you're comparing the power draw of a cpu alone to the power draw of a system-on-chip.
This post from yesterday provides a good explanation of this.
 

The Following 2 Users Say Thank You to sjgadsby For This Useful Post:
Benson's Avatar
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#5
Originally Posted by Johnx View Post
Uhm-, so you're comparing the power draw of a cpu alone to the power draw of a system-on-chip. The Intel Atom still needs a northbridge/southbridge to even be mostly as functional as an omap. And I'll bet that even the *cheapest* Intel Atom alone costs more than Nokia pays for an OMAP2420. BTW, be sure to do some research and get some benchmarks before you *assume* that the Atom will even be as fast as your average ARM processor, clock-for-clock. Also, I'd be interested to see where you got the power consumption figures for an omap2420.

-John

Hmmm. 1500 mAH * 3.7 V / 2.5W = 2.22 hours
SOC indeed; I think that's close to the power consumption of the whole tablet, since it's hard to get less than 2 hours battery life.

Last edited by Benson; 2008-04-03 at 16:00.
 
Johnx's Avatar
Posts: 643 | Thanked: 628 times | Joined on Mar 2007 @ Seattle (or thereabouts)
#6
Heh. I like the idea of maybe doing a big comparison review between an N800/N810 and a bare, unsocketed Intel Atom chip. We can finally have a conclusive answer to the question "Which is the best piece of tech to draw 2.5W?" Maybe we can throw in a low-power night-light as kind of a baseline.

-John

PS: Whenever someone starts making silly comparisons, I'm always reminded of one of my favorite penny arcade strips:
http://www.penny-arcade.com/comic/2000/04/21/
 

The Following User Says Thank You to Johnx For This Useful Post:
Posts: 60 | Thanked: 22 times | Joined on Feb 2008
#7
Google for power consumption. Yes people who guessed that I was quoting full power consumption for n800 (reported in some IEEE symposium where they tried to measure effects of some esoteric cache settings on power consumption) vs. chip only for Atom as reported in media releases. The test results for n800 were remarkable in that the consumption was more or less constant 2.4W+/-0.2 regardless what the device was doing (wireless networks off). So my comparison was a little biased against omap2420. But it shouldn't be that much off, there is very little silicon on n800 outside the 2420, even speakers are driven directly by 2420 as far as I can tell. I believe that will also be the case for Atoms (system-on-chip). 2420 has also full cellular phone circuitry build into it but that must be turned permanently off on n800. The only major power drain on n800 outside the 2420 is the screen and it's LEDs, but those should be around 100mW at most I believe. If I got something wrong somebody will correct me I trust.

Parenthetically a good indicator of your CPU oomph is its ability to decode video especially h.264 (dedicated silicon has very little advantage here), and here n800 disappoints compared to iPod video not to mention Touch.

Last edited by directore; 2008-04-04 at 08:09.
 
Posts: 1,418 | Thanked: 1,541 times | Joined on Feb 2008
#8
Originally Posted by directore View Post
The test results for n800 were remarkable in that the consumption was more or less constant 2.4W+/-0.2 regardless what the device was doing (wireless networks off).
That is obviously not true. Otherwise, it would not last for several days when left alone while only lasting several hours when actively used.

But it shouldn't be that much off, there is very little silicon on n800 outside the 2420, even speakers are driven directly by 2420 as far as I can tell. I believe that will also be the case for Atoms.
Not true as well. The current Atom chips are not SoCs. They are standalone CPUs that require additional chipset to be attached. See here:

http://en.wikipedia.org/wiki/Silvert...m_architecture

You will also find from this article that an Atom CPU will consume no less than 650mW of power (this is the TDP for the least powerful Atom part), but may deep as low as 100mW when parts of the chip are hibernated.

2420 has also full cellular phone circuitry build into it but that must turned permanently off on n800.
And this is also not true. Please see the diagram of 2420 architecture here:

http://focus.ti.com/pdfs/wtbu/TI_omap2420.pdf

You will see that 2420 includes a DSP but no cellular-specific parts.

The only major power drain on n800 outside the 2420 is the screen and it's LEDs, but those should be around 100mW at most I believe.
Somehow, I tend to believe that this is also not going to be true
 

The Following 2 Users Say Thank You to fms For This Useful Post:
Posts: 2,152 | Thanked: 1,490 times | Joined on Jan 2006 @ Czech Republic
#9
Originally Posted by directore View Post
the consumption was more or less constant 2.4W+/-0.2 regardless what the device was doing (wireless networks off).
My N800 lasts days connected to wi-fi network when sitting on the shelf. And as Benson said the battery has 3.7*1.5=5.55 Watt/hours so with such power draw it would not last more than 2 hours.

I also left N810 turned on but completely unused (just tapped screen once approx. each week) and it lasted 31 days from full to flat battery.
Originally Posted by directore View Post
So my comparison was a little biased against omap2420. But it shouldn't be that much off.
No, your comparison was not little biased, you comparsion was completely wrong :-)

Originally Posted by directore View Post
2420 has also full cellular phone circuitry build into it but that must be turned permanently off on n800.
Yes, that proves you don't know what you are talking about :-) There is no full cellular phone circuitry, check OMAP 2420 site.
Originally Posted by directore View Post
The only major power drain on n800 outside the 2420 is the screen and it's LEDs, but those should be around 100mW at most I believe. If I got something wrong somebody will correct me I trust.
Not screen itself but backlight which is major drain in any similar mobile device. Also wi-fi radio is huge power drain. You also forgot bluetooth, SDRAM and flash memory and SD/MMC cards. As said above your comparsion is completely wrong for many reasons.
__________________
Newbies click here before posting. Thanks.

If you really need to PM me with troubleshooting question please consider posting it to the forum instead. It is OK to PM me a link to such post then. Thank you.

Last edited by fanoush; 2008-04-04 at 08:30.
 

The Following User Says Thank You to fanoush For This Useful Post:
Posts: 60 | Thanked: 22 times | Joined on Feb 2008
#10
That is obviously not true. Otherwise, it would not last for several days when left alone while only lasting several hours when actively used.

I meant variation in active states, not hibernation - go find that IEEE symposium to see their data. in any case I'm outta here.


--

edit: look the technology usually marches forward not backward, Atoms will be 45nm, 2420 is 65nm I believe, also atoms use hafnium based high-K dielectrics for much lower leakage and faster switching speed, atoms gotta be an improvement over 2420 a technology at least 3 year old from a company that never was at the front end of semiconductor technology.

Now I'm outta here for good.

Last edited by directore; 2008-04-04 at 08:41.
 

The Following 2 Users Say Thank You to directore For This Useful Post:
Reply


 
Forum Jump


All times are GMT. The time now is 05:57.