Sign in with
Sign up | Sign in
Your question
Closed

Atom Benchmarked: 4W Of Performance

Tags:
  • Performance
  • Desktops
Last response: in Reviews comments
Share
Anonymous
July 29, 2008 8:30:03 AM

Originally developed for use in nettops, the Atom is now available for desktops. But how economical is the little platform, and is it powerful enough to handle daily work requirements?

Atom Benchmarked: 4W Of Performance : Read more

More about : atom benchmarked performance

Anonymous
July 29, 2008 9:37:22 AM

Where's Linux tests?.
Score
2
July 29, 2008 10:20:37 AM

This saved me time researching for an assignment :) 
Score
-2
Related resources
July 29, 2008 10:26:56 AM

Shouldn't it be on 9 - cooling and temperatures,be idle and load,instead of idle and idle? >.
Score
1
Anonymous
July 29, 2008 10:56:52 AM

When you look at the power consumption on load and compare it to the slowness of the chip while performing on load, it becomes clear why it only uses 4 watts more...
Score
0
Anonymous
July 29, 2008 11:08:03 AM

A CPU without a platform is useless.
Analyzing the Atom platform quickly from the power/performance perspective.
CPU name / idle W / load W/ Lame (seconds) / total Lame W used
Atom 230 / 40.5 W / 44.2 W / 773 s / 9.49 W
Celeron 220 / 44.9 W / 55.4 W / 375 s / 5.77 W
E2140 / 58.5 W / 69.5 W / 271 s / 5.23 W

Clearly the Atom platform is the most inefficient power/performance wise.
At idle you might win some W, but as soon as you try to do something you spend more power and waste more time.
There are other things you should consider, the frustration of having to wait for things that now we are used to do near instant and the inability to play HD video or use any significant graphics.

The only thing positive for atom is it's price. It's cheap. And maybe with a new chipset it might even be power efficient. But for now it's just cheap.

Score
3
July 29, 2008 11:48:23 AM

I wonder if it'd be able to play games if one were to equip it with a pci based 2400pro & 4gb of memory?
Also, I wouldn't be surprised if someone invents a voltmod for those boards, so they can increase voltage for cpu, mch & ich enabling 2ghz+ speeds
Score
-3
July 29, 2008 12:22:20 PM

so in other words atom is pretty much a failure unless it's pumped into a tablet or umpc? and even then apparently isn't cost effective or readily available, nice...sounds like a great product launch.
Score
2
Anonymous
July 29, 2008 12:32:35 PM

I ran a google search for ECS 945GCT-D mainboard, and could not find it.
I don't get it, why do you publish tests with products that don't even exist on the vendor web site?

Thank you, anyway
Score
-1
July 29, 2008 1:06:46 PM

Raiden, while your numbers are true, I think a fairer comparison would be to use 773s for all processors (meaning the rest go to idle alot quicker). I suppose you can power your computer down afterwards or start up a web browser etc.

Atom 230 (773s load / 0s idle) : 9.5W
Celeron 220 (375s load / 398s idle): 10.7W
E2140 (271s load / 502s idle): 13.4W
Sempron LE-1100 (43.9W idle, 70.4W load, 301s load / 472s idle): 11.6W

Sure this is biased against the Atom (not going idle at all) but with 4W delta between load and idle, I am too lazy to change the numbers already used.

I find it comparing the Atom to a Sempron LE-1100 more and Celeron 220 interesting:

"A Celeron at 1.20 GHz is 35% faster than an Atom at 1.60 GHz, but the Atom only consumes a fraction of the energy used by the Celeron. The AMD Sempron system, which uses almost the same energy in idle mode as the Atom system, is 43% faster."
Score
-1
July 29, 2008 1:16:08 PM

in anyway, despite of the low power and low performance, i still think it'll also be good as simple file server or home server besides as umpc. for experts, it'll be enough for some robotics and control application. just like the one used for Aiko in http://www.projectaiko.com
Score
0
July 29, 2008 1:24:46 PM

I've seen several articles on the Atom, but I have yet to see Tom's actually do a product comparison with a true Mini-ITX competitor, like Via. Is this because Via won't provide the hardware?

Score
2
July 29, 2008 1:53:32 PM

It would be interesting to see how to reduce power use by the 945GC when in a "headless" system e.g. firewall. I'd love to see how well this board performs with Endian Firewall or PFsense. Especially with all the Unified Threat Management (UTM) features running.
Score
0
July 29, 2008 1:56:45 PM

What about VIA's EPIA series? They have been around for ages and utilise less power than this, plus they have DVD and HD acceleration onboard. They are frequently run in cars and off batteries so they must have pretty low power requirements. They even released a dual-core one a while ago, surely this would kick atom's butt? (the DP-310 which has been out for years now).
Why not compare this like-for-like, surely you are aware of the VIA platforms and they are widely available with speeds up to 2GHz now. What's more VIA have announced that they will be providing boards with PCI-E x16 for proper graphics cards, giving them the edge over Atom which has been crippled to stop it affecting sales of Intel's more powerful and expensive products.

All this coverage of Atom would be far more balanced journalism if you compared it to a contender in it's own arena rather than more fully-featured desktop boards intended for a different market. I'd suggest you look at some of the Jetway boards for instance.

A long-time regular reader of Tomshardware
Score
4
July 29, 2008 2:42:35 PM

Please elaborate on this:
Quote:
Working with a screen resolution of 1280x1024 is possible, but compared to a traditional graphics card, it is a little blurred. At 1920x1200, the screen is washed-out and it is no longer practical to try to use on a daily basis.
I use a VGA connection to a 1920x1200 LCD panel all day long (through a KVM switch, no-less). It doesn't look washed out and it is completely usable on a daily basis. And how would an LCD panel look "blurred"? The pixels don't move, don't shift, and don't require focus? Are you using a CRT, and that is exhibiting timing issues with the VGA output that are not visible with a higher-quality VGA output driver chip? I just can't make sense out of your statements about video output.
Score
0
July 29, 2008 2:52:28 PM

Why isn't it compared to the mini-ITX offerings from VIA. Comparing it to full powered desktops is like comparing a bicycle to a motorcycle.
Score
2
Anonymous
July 29, 2008 3:12:54 PM

Seems like it's performance is comparable to an Athlon XP 2400 or so. I don't see how that's not acceptable for office use. Our whole site had PIIIs running XP up until 2005.
Score
-1
Anonymous
July 29, 2008 3:21:55 PM

I don't understand why Intel chose to have an elongated rectangular shaped die. Common sense dictates that the die be as square shaped as possible to minimize the surface area to pack more dies per wafer.
Score
0
Anonymous
July 29, 2008 5:30:15 PM

I know that these Taiwanese manufacturers have enthusiastically popped the Atom into these boards - but pitting a CPU designed to go after Transmeta, ARM, AMD Geode and VIA C7 type processors was aptly put as "pitting a bicycle against a motorcycle." This processor simply wasn't designed to compete in the desktop segment.
Score
1
July 29, 2008 5:32:10 PM

tennisballg: "Seems like it's performance is comparable to an Athlon XP 2400 or so. I don't see how that's not acceptable for office use. Our whole site had PIIIs running XP up until 2005."

Surely not Athlon XP 2400, but something twice slower, like Pentium 4 1,4GHz, or an Athlon 1 GHz. And same as Celeron M at 900Mhz which is used in EEE PC subnotebooks.
Score
0
July 29, 2008 5:44:12 PM

Typo on the "Core Temperatures Intel Atom 230" graph. I think the bottom bar is supposed to say "Load without fan" It says "Idle" right now.
Score
1
Anonymous
July 29, 2008 5:48:30 PM

Comparing the Atom processor to the latest dual core and more processors is just simply useless.

Much more I had preferred to see a comparison of the Atom processor to P4&P3, 800Mhz to 1,6Ghz from both Intel and AMD.
Due to the faster memory the Atom would probably turn out to be a winner there!
The Atom is perfect for running under Win XP and Linux, and like mentioned for chat/small game or web page servers, download platform, or to do the basic tasks of skype, internet, and editing html and text documents.

It'd also be nice to have a futuremark end-result, seeing that the atom only goes together with the Intel G950 (or was it 945?).

Further, remarks about the Sempron LE-1100 is basically useless, yes only on idle the difference is little, but noone would prefer the sempron above the Atom (power wise speaking). The extra cost of an atom processor over the LE-1000 is recovered in a matter of running the pc for less then a month.

the article mainly had the desktop section in mind when reviewing the Atom, and forgot that for a notebook this processor is very near to the perfect processor.
I want to wait until Intel has the guts to manufacture the G950 and Atom, and N/S-Bridge all in one less then 20WTDP package. That' give notebook manufacturers the option of manufacturing a laptop with more then 10 hours of battery.
Score
0
July 29, 2008 5:50:55 PM

Intel obviously needs a dual-core Atom version, tentatively named the Molecule :) 
Score
4
Anonymous
July 29, 2008 5:58:20 PM

edit on prev post
raiden_A CPU without a platform is useless.Analyzing the Atom platform quickly from the power/performance perspective.CPU name / idle W / load W/ Lame (seconds) / total Lame W usedAtom 230 / 40.5 W / 44.2 W / 773 s / 9.49 WCeleron 220 / 44.9 W / 55.4 W / 375 s / 5.77 WE2140 / 58.5 W / 69.5 W / 271 s / 5.23 WClearly the Atom platform is the most inefficient power/performance wise.At idle you might win some W, but as soon as you try to do something you spend more power and waste more time. There are other things you should consider, the frustration of having to wait for things that now we are used to do near instant and the inability to play HD video or use any significant graphics. The only thing positive for atom is it's price. It's cheap. And maybe with a new chipset it might even be power efficient. But for now it's just cheap.


Not true,
The Atom platform under load uses less energy then the others on idle.
clearly you're not going to use Atom based platforms as a workhorse. It's never intended for that.
If it takes you 15 minutes to check your mail, and read some web with the atom @ 40W, or it takes you 12minutes to do the same at a pc running at 50W idle, which one is the winner power-wise?
Score
0
Anonymous
July 29, 2008 6:04:18 PM

tennisballgSeems like it's performance is comparable to an Athlon XP 2400 or so. I don't see how that's not acceptable for office use. Our whole site had PIIIs running XP up until 2005.

It's performance is comparable to a 900-1,2 Ghz Celeron
Score
0
Anonymous
July 29, 2008 6:11:07 PM

ParkerochikI don't understand why Intel chose to have an elongated rectangular shaped die. Common sense dictates that the die be as square shaped as possible to minimize the surface area to pack more dies per wafer.

this is due to the fact that the little cache memory the Atom has, takes up nearly 1/3th of the processor's size.
In a situation with increased transistors a more squared version would be more likely.
Score
0
Anonymous
July 29, 2008 7:02:19 PM

The Atom should be compared to the Via Eden Line of processors.
Score
0
July 29, 2008 7:20:49 PM

Why does everyone assume all HD video is h.264? The 945 chipset DOES have mpeg2 acceleration but nobody tries that beyond simple DVD. I use the HDHR to record all my HD video OTA in 1080i or 720p HD mpeg2. It can be fairly taxing on single core systems but this processor with the 945 just might be able to handle it. Why is nobody who benchmarks this processor trying HD mpeg2? Am I going to have to buy one to find out if I can build a slim client for OTA HD?
Score
0
July 29, 2008 7:43:28 PM

TeraMediaPlease elaborate on this:I use a VGA connection to a 1920x1200 LCD panel all day long (through a KVM switch, no-less). It doesn't look washed out and it is completely usable on a daily basis. And how would an LCD panel look "blurred"? The pixels don't move, don't shift, and don't require focus? Are you using a CRT, and that is exhibiting timing issues with the VGA output that are not visible with a higher-quality VGA output driver chip? I just can't make sense out of your statements about video output.


LCD's can be very blurry when say you have a 640x480 source and are trying to display it on a 1650x1050 native LCD display the output will look very "blurry" since the image will be stetched.
Score
0
July 29, 2008 8:15:22 PM

I'm not impressed, actually. I guess the efficiency is impressive, but a 1.6 Atom can't even touch a 1.2 Ghz Celeron? Come on.

Now, I know these aren't gonna be used for anything hardcore (gaming, 3D, video conversions, etc.), and it's just a mini-system, but still, that's not very good imo.

I'm sticking with VIA and ARM for the mini stuff.
Score
0
July 29, 2008 8:37:45 PM

omicron_15, fair enough re: blurring caused by scaling. But the author makes a comparison between the IGP and "traditional graphics cards", claiming that the IGP is more blurry when doing such scaling (though with many displays, you can tell it not to scale and avoid such blurriness). That would implicate the display's scaling engine, not the GFX chip.

And I still am curious to know what the author means about the image looking washed out at 19x12? I ask, because I am curious if the VGA output is not functioning properly. Obviously a screen shot wouldn't help, but a photo of the screen compared to the same screen and image from a different computer would be useful to back up such a statement. I wouldn't want to purchase a MB w/ a faulty VGA output, especially for use as a simple DVR or whatnot. Autoboy above is contemplating exactly this, but he might reconsider if the VGA output does not function properly.
Score
0
July 29, 2008 9:46:44 PM

Thats the funny thing. Intel wanted this for a UMPC chip and I think people puleld too much hype out of it. For UMPCs it will probably be a great chip and the HT will help boost performance while using less power.

Too bad. Maybe when Intel gets a better chipset for it (based on the 45nm process and a better IGP) maybe it would be viable for the desktop but as it stands Atom is a UMPC chip at heart.
Score
0
July 29, 2008 11:00:34 PM

fazers_on_stunIntel obviously needs a dual-core Atom version, tentatively named the Molecule

Agreed :D 
Score
0
July 30, 2008 2:49:23 AM

Celeron is the slowest processor before, now we welcome ATOM. I think Sempron is much better.
Score
0
July 30, 2008 5:24:05 AM

Ok, my question is, how did ASUS managed to achieve 20W under full load with their EEE Box? I mean, all these custom builds with ITX boards with atom consume a minimum of 40W, where's the catch?
Score
0
July 30, 2008 6:19:43 AM

Any chance of adding to this comparison another designed-to-be-low-power CPU?
The Core 2 Duo U7600 (as in the Sony Vaio TZ series, Toshiba Portege R series, Fujitsu Lifebook T series and likely some ultraportables I've forgotten) with its 1.2GHz and around 12W power draw would make and interesting contender for the Atom's power consumption/performance crown.
Most of the above laptops use the mobile 945 chipset too, which I know doesn't entirely level the playing field on power consumption but should at least provide a fair performance comparison.
Score
0
July 30, 2008 8:09:58 AM

good tests, linux testing (for example on most popular Ubuntu distro) would be great ;)  I wouldn't buy it to run windows on it. pleeeease ;) 
Score
0
Anonymous
July 30, 2008 9:58:43 AM

TeraMedia> There is little difference when you use D-SUB and not DVI on a LCD panel.
Score
0
July 30, 2008 10:13:38 AM

Quote:
Intel obviously needs a dual-core Atom version, tentatively named the Molecule


Better yet, name it something along the lines of valance electron or chained pairs or something like that...

Sub-atomic particles? Anyone? No. Okay then.
Score
0
Anonymous
July 30, 2008 11:08:55 AM

ProDigit80 - 15m*40W=12m*50W so they are equal so no advantage power-wise for the atom with your example. It takes longer to do the same thing with the atom and the power savings just doesn't justify it.
You are not going to use the atom as a workhorse true, but the fact is that the atom is slow and most of the time it will be used at high load even doing simple things and as I already demonstrated, the atom (platform) is inefficient power/performance.
The problem is not with the atom, the CPU has excellent power/performance ratio, but with the chipset witch uses lots of power. The rest of the CPUs tested have worse power/performance ratio but are performing better and the chipset power use is about the same. The power used by the chipset dilutes the power/performance of the atom processor and that is a shame.
I know some of you will likely contradict me on the power saving issue (estimating power usage on different scenarios) , but really it doesn't matter if I'm right or wrong, it's a desktop and it's only a few W, it won't change the global warming.

I consider the atom CPU a great product, but not for desktop use.
Other (bad) things to consider about the platform:
The CPU is week. Users expect the same level of performance from a ITX as a midrange computer, after all it is a desktop. Already there are ITX boards that use regular desktop CPUs and there are no thermal issues with a small from factor. People buying SFF are interested in design, performance, features not necessary power savings, and as already demonstrated, the atom desktop platform isn't delivering those power savings.
The price needs to be low. Lower then the current ITX platforms since the atom doesn't perform as well. To keep the prices down the platform will use the cheapest chipset/chips available. That's why we have a chipset that uses so much power. It doesn't have a modified mobile chipset or any power efficient chips. All of those cost extra money and will make the atom platform more expensive than the current ITX lineup.
In the end I don't think the users will go for that. Atom for desktop needs to be more powerful and already Intel is working on a dual core version and probably will increase it's speed also.
If Intel pushes atom to a level of performance that the users will be comfortable with, the platform can change the market the same way ultra portables did. Intel might even design a cheap power efficient chipset so the whole platform to be power/performance and price/performance efficient. Imagine a 200$ sexy SFF ... might convince many to abandon the uglier, noisier cousin.
Intel is going in the right direction, but it has a long road ahead and the competition isn't staying idle.
AMD has great chipsets and VIA has power efficient CPUs.
Sorry for the long post
Score
0
Anonymous
August 1, 2008 12:05:56 AM

I don't found the complete test bed used. I'm looking especialy for the power suply, that should be responsible for more than half of the power consumption in those tests, leaving the results and conclusions partialy meaningless.

Simply going from an (gross oversized)550W to an (very oversized) 220W PSU shaves 20W from the total system power consumption on this atom plataform. More here: http://www.xtremesystems.org/forums/showpost.php?p=3176...
Score
0
August 4, 2008 4:40:16 PM

Given these products are targeted at basic workloads, I would have expected a little more testing on the web surfing side. The web page loading is good, but it really depends on what kind of page and the content. With most sites and advertising based on Flash 9, my personal experiences on these new CPUs is that they come to a real crawl (70-100%) CPU utilization when surfing even kid's sites like "disney.com" and "nickjr.com". Any chance you will be going deeper on web sites and show how these platforms stack up?
Score
0
Anonymous
August 7, 2008 2:46:39 PM

i have intel945GCLF motherboard that also the same with this ecs system. and i'm very happy with it. i've install many thing and perform very smooth with all yada yada enabled. ubuntu 8 super fast. i also have supercomputer and this this tiny thing just slip in my 5.25 bay drive. now i have 2 x86 inside one cpu case. so while i'm just browsing the web or downloading stuff. i even write this on post in it , i switch to the atom gear... that;s what i called speedstep. and one more thing. i bought it only $69. i think it because the ecs way. try another mobo with atom please..
Score
0
Anonymous
October 11, 2008 3:09:57 AM

"I don't understand why Intel chose to have an elongated rectangular shaped die. Common sense dictates that the die be as square shaped as possible to minimize the surface area to pack more dies per wafer."

2x8 arrangement of rectangles is the same as a 4x4 arrangement of squares. And a rectangle has more perimeter, so more sideways conduction of heat. And a hexagonal array would pack even more dies per wafer than squares.
Score
0
Anonymous
November 13, 2008 5:49:42 PM

i just purchased an atom 230 with 2 gig 160gb sata2 on a gigabyte gc230d. it is replacing a q6600+4850 system for day to day office duties. both running xp. and you know what? as long as you don't ask it to encode video or play proper games it wurks fine. think single core athlon 64 with 1 gig ram speeds. it burns dvds, plays itunes and surfs with mozilla all at the same time. the only downside is the load times of apps. horses for courses. as they say in the uk.
Score
0
Anonymous
December 11, 2008 3:11:38 PM

my question is, , qhere is the nano from via ?
I want to compare before buy. but I cant find it anywhere, on stores to know the price and think ...

Score
0
Anonymous
February 2, 2011 11:37:35 PM

I have an Inspirion Mini 10 and it can't play games for shit thanks to this supposed "fix all". FUCK INTEL!
Score
0
Anonymous
February 2, 2011 11:38:28 PM

I have an Inspirion Mini 10 and it can't play games for shit thanks to this supposed "fix all". FUCK INTEL!
Score
0
!