Secrets of the Atom cracked, Wind PC revealed

Everyone wants a new generation ultra-portable pumping the Atom processor right? After all, it's Intel's latest and greatest Centrino proc. Not so fast... literally. Impress got their hands on an unofficially released MSI Wind PC (the mini-PC, not the low-cost laptop) and subjected the 1.33GHz Atom processor to a suite of standard tests. While the source is in machine-translated Japanese, that table comparing the Atom with Eee PC 900's Celeron and a typical ULV Core 2 Duo found in a wide range of full-sized, ultra-portable laptops speaks for itself. Keep in mind though that Intel's Atom is supposed to be less power hungry than its peers and CPU benchmarking is hardly the last word when it comes to determining real-world laptop performance -- read/writes to disk, memory, and other system nuances must also be considered. Nevertheless, based on this table and other mounting evidence, first and second generation Eee PC owners shouldn't feel any compulsion to rush out and upgrade to Atom on day 1.

[Via Technophone and Notebook Italia]

[Hellios @ May 21st 2008 5:58AM]

I'm not sure how the graphics score was different in the EEEs seeing they both use exactly the same chip (and excluding the screen, everything else as well)

[Albert @ May 21st 2008 6:12AM]

because the CPU is faster, therefore less latency in commands received. Anyway imagine comparing a Pentium 4 with a x1950 agp verus a Core 2 Quad 9300 with a x1950 pci-e. I'm sure you'll agree the 9300 will get better frame rates.

[Blackstar @ May 21st 2008 11:43AM]

I'm waiting for the next-gen "Blows PC".

[Akshat @ May 21st 2008 6:00AM]

This might be a dumb question, but is it possible that the CPU score was lower on the Wind due to Vista?

[gad get @ May 21st 2008 6:25AM]

No.

[gad get @ May 21st 2008 6:27AM]

Or possibly yes.

[Mile @ May 21st 2008 7:10AM]

Very well maybe.

[white_ultras @ May 21st 2008 8:12AM]

There should be speed/watt table, then ATOM would probably come on top,
ATOM is *much* slower then Celeron, but is also much less power hungry about ~2 Watts!

[MMaster23 @ May 21st 2008 8:37AM]

I don't think so because PCMark scores tend to be higher on Vista (look at other review, it really is :))

[SAS @ May 21st 2008 6:02AM]

I have a question under those system nuances does operating system affect the performance because the Wind is running vista compared to the eee pc's XP

[harveyhanson @ May 21st 2008 6:09AM]

Wont things like RAM and FSB speeds change these results though, I understand that CPU score shouldnt be affected by RAM, but it doesnt say what setup they are on

[jouh @ May 21st 2008 6:11AM]

Perhaps the different score on the Eee PC's is caused by the RAM. (More memory assigned to the graphics...)

[CosterMonger @ May 21st 2008 10:29AM]

First thing I notice :)
Why are the memory scores different?
why is the MSI Wind with 1GB of DDR2 @333Mhz better then then the ASUS EeePC 900 with 1 GB of DDR2 @400Mhz.

still need to see a multi-threaded application test for the Atom.

[Velorium @ May 21st 2008 12:32PM]

Ever heard of latency?

[CosterMonger @ May 22nd 2008 10:17AM]

FSB
wind 533Mhz
Eeepc 400Mhz

[Richard Lai @ May 24th 2008 12:43PM]

@CosterMonger: wouldn't the slower RAM cause a bottleneck though since it doesn't match the Wind PC's FSB?

[Jeremiah @ May 21st 2008 6:13AM]

As some others have stated, I'm curious if Vista could have held the MSI Wind's scores back a tad bit. I know from experience that PCMark scores can very by a couple hundred points between Vista and XP.

[pete @ May 21st 2008 6:15AM]

This is a totally skewed test, NONE of the 4 machines had the same OS or ram. We all know that vista is a ram hog and vista needs 2 GB ram to run averagely. The atom is tested with 1 GB of ram.

I'm not pro or against Atom, this test is worthless unless its in a controlled environment.

[Reikon @ May 21st 2008 3:39PM]

For everyone complaining about the unfair comparisons because of differences between OSes, this is a review of the whole system, not the processor.

[Darak @ May 21st 2008 6:16AM]

I could be mistaken, but I'm fairly sure that its been common knowledge that the Atom processor is slower than a Celeron 900 for quite awhile. I distinctly remember discussions about it on other forums.

[Mile @ May 21st 2008 7:11AM]

That was speculation though because neither Marty nor KingJoeDoom had gotten their hands on an Atom yet.

[iofthestorm @ May 21st 2008 11:27AM]

It's been common knowledge even here, people just like to ignore that fact or something.

[Oinquer @ May 21st 2008 6:19AM]

anyhow...the Atom on Wind seems to be a 1.33 and not a 1.6...and it seems other than consumptions its taking hell of a beat from 900 MHZ celeron

[Oinquer @ May 21st 2008 6:22AM]

and wth is that HDD score?? seems too unrealistic...

[gad get @ May 21st 2008 6:29AM]

Yeah, what kind of a hard drive did they put in that thing?

[Ender17 @ May 21st 2008 6:42AM]

The Wind has a standard 2.5" hard drive, so of course it beats the crappy SSD in the EEE and the pathetic 1.8" drive in the TZ.

[d00b @ May 21st 2008 6:56AM]

The hard drive may well be a 3.5", seeing that this is a ultra-compact desktop type and not a subnotebook.

Narrowly extrapolating from the CPU Score, a 1.6 GHz Atom would be roughly equivalent to a Core 2 Duo running at 600 MHz. (1159 / 2832 / 1.33 * 1.6) * 100% = 49.2%.

[George @ May 21st 2008 7:15AM]

Intel have already stated a while back that the 1.6ghz atom will perform around the same mark as a 800mhz celeron ulv in single threaded performance. Its approximately half as fast at the same clock speed. This is because the die of the chip is considerably simpler and executes in order.

But the Atom processor supports 2 threads (via hyperthreading) which should offset this limitation somewhat.

Remember, atom works at a 2.2 watts tdp (with hyperthreading) while the existing celeron 353 in the eee 900 works at 5 watts. If hypertheading does boost performance then atom will give better performance per watt... which is the point.

[d00b @ May 21st 2008 7:31AM]

The low-power argument for the Atom is more relevant for smaller (hand-held) form-factors than it is for subnotebooks, let alone for ultracompact PCs (which are AC-powered). CPU power consumption is only of many factors for a x86 general-purpose device.

Even assuming that web browsing will be the dominant use for the LCUPs, low-power will need to be offset by substantially longer battery life for the trade-off to be worthwhile. Web page rendering is getting more CPU-bound with more use of Flash & animation.

[notYou @ May 21st 2008 11:11AM]

That's a good analysis, but it won't help the Atom when it comes to consumer products since the cpu is just one piece.

In other words, if the Atom has a good perf/power ratio, but it takes longer to perform the task, you've probably just wasted the Atom's thrifty power savings keeping the rest of the system functioning during the wait (lcd, backlighting, spinning drive, fans, etc.). That's why Intel is focused on the entire platforrm (so they can similarly optimize the other subsystems that they have dominion over). But, unless they're going to actually build the unit themselves (not sure, are they?), they won't have control over every other power draining subsystem in there.

[Rob @ May 21st 2008 2:09PM]

Meh, the Atom isn't significantly slower than the Celeron in the 900 Eee, and is faster than the Celeron in the 701, which wasn't too bad. More importantly, the 701 got burning hot even with its tiny buzzing fan. If the Atom is going to have significantly reduced heat production, I'm all for it.

[Sascha @ May 21st 2008 7:34AM]

Pretty interesting that they are using the Silverthorne for the Wind PC and not the Diamondville, which should be way cheaper.
I would like to see these benchmarks on the Diamondville, too just to see how it compares to the Silverthorne.
Compared to these crystalmark benchmarks:

http://www.eeepcnews.de/2008/04/18/intel-atom-benchmarks-via-isaiah-vergleich/

i am sure we won't see a big difference between these 2 Atom versions

[CosterMonger @ May 21st 2008 10:32AM]

because Diamondville isn't out yet.

[warlun @ May 21st 2008 7:37AM]

isn't the HDD score for the eeepc so low becos of it's low capacity?

[Ender17 @ May 21st 2008 7:56AM]

No

[m-p{3} @ May 21st 2008 8:00AM]

I don't know about them, but if I had to do a benchmark, I'd make sure there is as less as possible in configurations, to avoid some variables. Evaluating some PC under Windows XP and some under Windows Vista might cause some differences, so this could be biased.

[bobgorila @ May 21st 2008 8:06AM]

But what about the battery life benchmarks, hmmm?

(where's my all-day computer, dammit)

[slarity @ May 21st 2008 8:27AM]

As already stated I think having vista on there is a terrible idea...

[TRAFFICBLOWS @ May 21st 2008 8:35AM]

why can't i get a ULV processor for my desktop!?!?!

[K @ May 21st 2008 9:00AM]

This is a pointless benchmarking test that should be ignored.

Atom in the MSI Wind Laptop(not desktop) will be 1.6ghz which will be slightly better than the Celeron in the currently available EEE 900 series. All of the machines above should of had XP as the OS too.

When the MSI Wind notebook is tested, then we will see.

[fgpx78 @ May 21st 2008 9:04AM]

"first and second generation Eee PC owners shouldn't feel any compulsion to rush out and upgrade to Atom on day 1."

...except for battery life I would say.

[Patriks7 @ May 21st 2008 9:10AM]

i have a question about this atom
is this the "celeron atom"?
because i read an article couple days ago and saw something about a celeron atom

[JAmerican @ May 21st 2008 9:19AM]

Still getting the Wind Laptop. You can't stop me Engadget :P

[average_joe @ May 21st 2008 9:31AM]

Battery times for the portables will be so much more interesting, as it is the single most important factor for me.

[Howard @ May 21st 2008 9:38AM]

I was really psyched about the Atom until recently. It will be great for battery life, but you have to go back a couple generations to find a CPU that it can beat clock-for-clock. I know it's all about low power, but I would at least expect them to beat a four-year-old Pentium M. Apparently that's too much to ask.

I was all set to buy an MSI Wind at launch, but I just picked up a used Thinkpad X-series notebook on eBay for the same price instead. It might weight half a pound more, but Core Duo still destroys Atom.

[d00b @ May 21st 2008 9:45AM]

@Howard: I was thinking the same thing, picking up a X61. Lenovo had a clearance on the older version for around $800 last year (free shipping). This was with a 2.0 GHz Core2Duo, 2 gigs, and an 6-hr battery, all in a 4-lb package. Should've went for it, but didn't. Am hoping the deal will come around again.

[coffee @ May 21st 2008 9:50AM]

I'm not looking to harness massive power here, I just want a light email/browser/office machine.

[Che @ May 21st 2008 10:01AM]

The table is missing two important rows: CPU power consumption and system price.

[digitallysick @ May 21st 2008 11:00AM]

These results are bogus, how can you compare products if they are not running the same OS, put xp on all of them, or vista on all of them, and then run the tests.

[Sascha @ May 21st 2008 11:24AM]

@CosterMoster

Diamondville is available on the Intel Little Falls Mini-ITX board, so you can buy it ;)