VIA Nano Bests Intel Atom In Netbook Benchmarks

Glib Piglet writes "ZDNet UK has a whole set of benchmarks comparing a 1.8 GHz Nano in VIA's Epia SN motherboard and a 1.6 GHz Atom in Intel's 'Little Falls' D945GCFL mobo. It's not good news for Chipzilla: 'As far as memory performance is concerned, the Nano is clearly superior in every test' and 'The VIA Nano emerges as the better processor for internet tasks. While the Atom needs 132.8 seconds to display simple HTML pages, the Nano does it in 70.1 seconds.' The Nano even outperforms Nehalem on one test. It's not all a win for VIA, though. The benchmark concludes that in some ways all netbooks, underpowered as they are, remain in the IT stone ages."

Comments

Hmm.

[Anonymous Coward]

Didn't we read similar tests already? The Nano should of course win, since it has out-of-order execution design. But, from what I hear the dual core Atom is at par of the single Nano. When the dual core Nano is released (should be late this year, early next year), it'll wipe the floor with the Atom.

Re:Hmm.

[cb88]

what about when the dual core nano comes out... which it is also via's design is clearly superior if only they had the factory to produce 32nm chips

All but the important test

[afidel]

Yeah, it's 1.5-2x faster but it also draws 50% more power so comparing them solely on the basis of CPU performance kind of misses the point of a netbook, if you want a fast CPU use a full laptop.

Re:All but the important test

[Anonymous Coward]

...50% more? since when? Sure, the Nano may use more power, but it's nowhere near 50% more. 60.1*1.5 is 90.15, and x2 it's 120.2. The Nano tops out at 77.5. Making up bullshit is not "interesting" or "insightful".

Re:All but the important test

[Anonymous Coward]

Link please?

Everything I've seen shows the Nano drawing more power than the Atom under full load, but drawing LESS power while idle. I don't own a netbook, but my guess is they would spend more time idle than under full load. Most people aren't doing number-crunching on a netbook.

Re:All but the important test

[bluefoxlucid]

Apparently making up bullshit is interesting.

Re:All but the important test

[Lord Ender]

I agree. This review misses the point entirely. Netbooks are about portability--size and battery life. An Intel Atom-powered netbook can do all your web/officy stuff (as well as full-screen Hulu) and run for eight hours on a charge. There is no benefit in bumping the speed up a touch if that means shortening battery life.

If you want video editing and gaming capabilities, netbooks aren't for you. The only netbook processors that might interest me would be those that give me more speed with the same or less power use as the Atom.

Re:All but the important test

[afidel]

Um, RTFA? 48W*1.5=72W, 68~=72.

Re:All but the important test

[afidel]

TFA has the Atom based unit running 48W full out and the Nano based unit at 43W idle (no numbers given for Atom system strangely). I think it's safe to assume the Atom unit is drawing significantly less than 43W at idle if it uses 48W running synthetic benchmarks.

Re:All but the important test

[the 99th penguin]

It also depends whether it would be worth using more power but finishing the task quicker, assuming that when idling they would be more or less comparable. One problem the Atom has had so far is that the chipsets they have been paired with draw a lot of power by themselves making the power savings less noticeable than they could have been.

Re:All but the important test

[CaptainPatent]

50% more power my foot.

The Intel Atom at 1.6Ghz is a 2-4 watt processor.

While they hide it in the details of their press release a bit, the VIA Nano processor running at 1Ghz is the 8 Watt processor I believe you're referring to while the 1.8Ghz processor tips the scales at around 25 Watts.

Re:All but the important test

[electrosoccertux]

Thank you for pointing that out, I couldn't tell from the GGP's moderation. ;)

Re:All but the important test

[afidel]

I'm talking about from TFA the power draw of the entire system. I don't really give a crud about how low power the CPU is, I care how much battery life I can get out of a give weight in batteries. The Atom based unit draws 48W full tilt vs 68W for the Nano based system, heck the Nano system draws 43W idle!

Re:All but the important test

[gEvil (beta)]

Apparently making up bullshit is interesting.

Yes, it is.

Re:All but the important test

[Shadow-isoHunt]

But it doesn't have a chipset drawing more power than the CPU, either.

Re:All but the important test

[AnarkiNet]

Maybe its time for Folding@on-the-go.

Re:All but the important test

[Anonymous Coward]

OH STFU! The VIA C7 was always compared against the Dothan even though the C7 was 1/2 the power. Nano beats Atom on performance with slightly higher power. Congrats VIA!

Re:All but the important test

[CaptainPatent]

The article is about the Nano processor VS. the Atom processor. While an entire system was put on the table for benchmarking, it's hardly representative of the problems the processor itself represents for netbook use nor is it the focus of the article.

I'm sorry I misinterpreted your response to a processor-related article to not be about the processors being analyzed.

As a side note, I'd like to add that it's funny that as I said the Nano benchmarked is a 25 watt processor and the Atom benchmarked is a 4 watt and the difference between the overall benchmarked system power at full load is 20 watts... Of course that couldn't be related to the processors being 21 watts apart... because you:

"don't really give a crud about how low power the CPU is"

Re:All but the important test

[GooberToo]

Yeah, it's 1.5-2x faster but it also draws 50% more power

That means if it is under clocked to draw the same power it's still likely 50% faster. That's still a win for VIA. Heck, even if you under clock it and its only 20% faster, faster is still faster. I'm not sure how you see that as a lose.

Re:All but the important test

[philipgar]

Actually, this implies that the difference in power of JUST the CPU is far more than 50%. This is because most of the system's power draw is NOT from the CPU. Lets assume for instance that at idle the Atom processor consumes 5 watts of power (the rest of the system consumes the rest). This means that at idle the VIA cpu consumes almost 8 watts of power (both these estimates are perfectly reasonable based on the class of processors that they are). This estimates means that when idling the VIA cpu is consuming 50% more power than the atom.

Now, when the cpus are at load, the Atom processor is consuming ~8.5 watts of power, and the VIA is consuming ~25 or 26 watts of power. This looks to me that the via processor is consuming 4 times the power of the Atom, not merely 50% more.

Of course, this estimate is assuming that the Atom processor's idle power is only 5 watts. In reality, the idle power it consumes is likely even lower, as it was designed to minimize power dissipation. Now, claiming that the VIA's system power is approximately 50% more than the Atom is not accurate, but that doesn't mean that the CPU is not consuming that much more power. Anyone doing a fair comparison between the processors would likely be focusing on the difference in power of the CPUs themselves. Otherwise in a full system, the difference between a CPU that requires 50 Watts of power, and one that requires 100Watts of power wouldn't be that significant.

Phil

Re:All but the important test

[afidel]

Well it would be interesting to see the 1GHz Nano vs the 1.6GHz Atom since they would appear to have about the same power envelope once chipset is considered (assuming all other parts between the two units were equivalent) but that's not what the article was comparing at all so claiming the the Nano bests the Atom was way off base.

Re:All but the important test

[Vu1turEMaN]

thank god, someone gets it!!!!

the only thing that is stopping via from being HUGE is their gpu. if they had a faster, more reliable solution we'd all be using via netbooks right now.

Re:All but the important test

[wirelessbuzzers]

The power consumption numbers in TFA are totally useless as applied to netbooks. My laptop (an ultraportable, not a netbook... it has a 1.6 GHz Core 2 Duo in it) uses maybe 25 watts at full load, and maybe 8-9 watts at idle, counting the screen and wireless (that's in Windows; Linux uses slightly more power for some reason).

TFA quotes one system at 48 watts and the other at 68 watts under load. You can't say that this is representative of their performance in a netbook.

The real question is, if you compare them with chipsets, clock speed, voltages, hardware, power management strategies etc that are suitable for a netbook, how is the power consumption and performance? TFA doesn't address that, so it's basically useless.

Also, remember that the two power figures of merit for a netbook are idle power and joules per operation. Suppose that procs A and B are the same at idle, and that B uses twice as much extra power under load than A, but is twice as fast. Then B wins almost no matter what you're doing (yadda cooling yadda voltage ramp yadda yadda).

Re:All but the important test

[CaptainPatent]

Well it would be interesting to see the 1GHz Nano vs the 1.6GHz Atom since they would appear to have about the same power envelope once chipset is considered (assuming all other parts between the two units were equivalent) but that's not what the article was comparing at all so claiming the the Nano bests the Atom was way off base.

And on that I completely agree.

Re:All but the important test

[Tenebrousedge]

That's not quite true. Most of the use of a netbook doesn't need a powerful CPU, it's true. Currently my Atom-powered netbook has clocked itself down to 1 Ghz; for posting on slashdot, nothing more is necessary. However, that's not to say that there is never a need for it to scale back up. It's like your car (come on, you knew that was coming): you don't usually drive it as fast as it will go, but it's nice to be able to go fast if you have to.

I would want a faster processor if only to reduce bootup time. Although I hear that others have made do with the existing processor limitations.

Also, eight hours on a charge? Where can I get me some of that?

Re:All but the important test

[afidel]

They did give the idle power for the Nano system though, 43W which is within 10% of the Atom system full out. I really wish they had given the idle numbers for the Atom system, that missing piece would most likely strongly back my argument. Gah, I just did a bit more reasearch and they used the freaking DESKTOP chipset for the Atom. That thing draws 17W more than the portable version which further exacerbates the difference between the platforms.

Re:All but the important test

[poot_rootbeer]

It's like your car (come on, you knew that was coming): you don't usually drive it as fast as it will go, but it's nice to be able to go fast if you have to.

If you buy a Toyota Echo, you have an affordable little car that will reliably get you from Point A to Point B. But you have to accept that your cargo capacity is going to be rather limited, and even if it's possible to get it above 85mph on the highway, the car wasn't really designed to do that.

Re:All but the important test

[Nethemas the Great]

The whole power usage thing as classically measured by what the processor draws under load doesn't exactly produce a fair and accurate picture. For instance, since AMD chips of recent years have tended to consume more power than Intel offerings. However, the north bridge for AMD chips consumes less power than for Intel in large part because the memory controller is bolted onto the AMD chip rather than the north bridge. Also, if a processor consumes say 50 watts of power and completes a given task in 15 seconds and another processor consumes 35 watts and completes the same task in 25 seconds the second processor has actually consumed more power.

Re:All but the important test

[BikeHelmet]

Old benchies from last august put them at equal battery life, which suggests to me that it's the size of the battery affecting the battery life of the netbook - not the Atom or Nano CPU.

It's long been known that Nanos are incredibly fast compared to Atoms.

Now if only I could buy one. "Released" a year ago, huh? Where? :P

Re:All but the important test

[hierophanta]

Also, eight hours on a charge? Where can I get me some of that?

you can get it here: http://www.amazon.com/Samsung-NC10-14GBK-10-2-Inch-Netbook-Processor/dp/B001O94FY8/ref=pd_bbs_sr_2?ie=UTF8&s=electronics&qid=1234307109&sr=8-2
its rates it at 6, but other places rate it at 8. to be completely fair - i do own this bit of kit and i regularly get a reading of 6hours batter left. and i run xp - linux would probably realize that 8 hours

Re:All but the important test

[evilbessie]

I haven't done this maths so this might all be wrong.

But if you spend more time waiting for the Atom to do stuff do you get more usable time out of the Atom or Nano, rather than waiting for it to computer something. So your 8 hours might be something like 6 with 2 hours waiting for the computer to catch up.

Re:All but the important test

[Anonymous Coward]

...50% more? since when? Sure, the Nano may use more power, but it's nowhere near 50% more. 60.1*1.5 is 90.15, and x2 it's 120.2. The Nano tops out at 77.5. Making up bullshit is not "interesting" or "insightful".

You should investigate further before you claim bullshit.

In fact, the OP was quite generous to Nano. The 1.8 GHz Nano is rated (by Via) at 25W TDP (Thermal Design Power). The 1.6 GHz Atom 230 (desktop version) is rated (by Intel) at 4W TDP, the N270 (1.6 GHz netbook version) at 2.5W, and the latest N280 (1.666 GHz netbook) is a mere 2.0W. I'm sure Intel and Via use slightly different methods for measuring TDP, but not that different, you know?

Nano doesn't seem to have separate desktop and netbook versions. If you want the highest speed grade Nano in your netbook, you're going to be putting in a CPU which draws 12.5x as much power as the N280.

Re: the PC Perspective review you linked, you need to be aware that (1) PCPER measured whole-system power consumption, not the processor by itself, and (2) Intel's Atom desktop boards ship with an extremely old integrated graphics chipset which uses so much power that it actually requires a fan (the Atom CPU itself doesn't). Go back to your link and scroll down to the graph of entire-system power consumption during MP3 encode. Notice how the Atom's graph is nearly flat, going up by only ~3W during activity? That's because in this case, even at full load, the CPU accounts for a vanishingly small percentage of the total system power.

The story's a bit different in netbook land, where Intel will sell you a different version of the CPU with a more modern chipset. The netbook version features a reduced power variant of Intel's standard front-side bus. (The 'desktop' version of the Atom uses the normal Intel FSB, which increases its TDP and makes it compatible with the ancient chipset mentioned above.) The total TDP of an Atom N270 plus the netbook chipset (north+south bridge) is about 12W, less than half the TDP of the top speed bin Nano CPU all by itself.

The Nano's problem in a nutshell: its ULV low clocked versions still uses much more power than even the fastest single core Atom variants (5W TDP @ 1.0 GHz is the wimpiest version Via lists), but then loses its performance advantage. The high clock rate versions which destroy Atom in performance benchmarks use about as much power as a low or midrange Core 2 Duo. But even if you disable one core, a C2D is going to completely destroy the Nano in any benchmark. The only advantage Nano has over C2D for a CPU in the ~15-25W TDP power class is really low idle power consumption (500mW for the top 1.8 GHz speed bin, 100mW for the rest).

Unfortunately for Via, that advantage has not proven compelling enough for any manufacturer of portable computers to jump on the Nano. And it has literally no advantages for netbooks, where you never want a CPU over 10W. Via's been trying to get design wins for a long time now, and nobody's biting.

Oh, and there's one other problem they face: they might try to buy design wins by pricing it really low to make up for its shortcomings, but that's a game they can't win. Nano's die is ~63mm^2, Atom's is ~26mm^2. They can probably sell cheaper than a 45nm Core 2 Duo (82mm^2 for 3MB L2 cache variants), but not enough to make up for the C2D's vastly better performance within the same thermal envelope.

Re:All but the important test

[muridae]

You've never driven an Echo, have you? It was not a third row of seats, be any measure, but they broke some laws of physics and reality to put that much trunk space in those cars.

Re:All but the important test

[MarkCollette]

It's actually pretty easy to go 136 km/h (85 mph) in a Toyota Echo. I'd say anything up to 160 km/h is trivial. Over that and buffeting winds or curved roads will probably make you uncomfortable. But still, it'll go all the way to the 180 km/h limit where the governor kicks in. And that's probably just because the stock tires are only rated to 180 km/h (like stock tires for almost every car).

Re:All but the important test

[Lord Ender]

I don't use my Atom-powered netbook for physics simulations, so I spend zero time waiting for it to "do stuff." The network speed is pretty much always the bottleneck.

Re:All but the important test

[Lord Ender]

Also, eight hours on a charge? Where can I get me some of that?

The Samsung NC10, as well as several Asus modles do 8 hours. MSI has a version of the Wind with an 8 hour battery, too.

Personally, I bought the NC10. It seems to be slightly superior to its peers in every category (except price--markets at work, I guess).

Re:All but the important test

[Anonymous Coward]

Lets assume for instance that at idle the Atom processor consumes 5 watts of power (the rest of the system consumes the rest). ... (both these estimates are perfectly reasonable based on the class of processors that they are).

You're way off. The Atom is 2.5 watts PEAK (with typical dissipated power a little over 1 watt). In standby it's around a hundred milliwatts or less.

Re:All but the important test

[Anonymous Coward]

But, if the Nano can do the job 6 to 8 times faster, it consumes less power to than the atom to complete the task, and therefore much less power.

yeah but ...

[Anonymous Coward]

does it run crysis?

Re:yeah but ...

[Dishevel]

does it run crysis?

2.6 FPS :)

Re:yeah but ...

[stfvon007]

don't you mean FPM?

Re:yeah but ...

[Kumiorava]

Yea, but you didn't set the details right, I got it going close to 60 SPF.

Beats?

[hicks107]

Beats? Bests?

Re:Beats?

[Anonymous Coward]

It's not a typo. I refrain from bitching you out only because English may not be your first language. In any case, you have learned a new word.

http://www.merriam-webster.com/dictionary/best%5B4%5D

I'd be nice to see it in some devices...

[fuzzyfuzzyfungus]

Unfortunately, all of that is largely theoretical until VIA can score some design wins, which is a pity because the present state of things doesn't exactly motivate chipzilla to drop margins or loosen restrictions on Atom.

Re:I'd be nice to see it in some devices...

[fuzzyfuzzyfungus]

I'm fairly sure that they are referring to some sort of multi-page benchmark, or a rather meaty benchmark page, the number is just too absurd otherwise. I've loaded slashdot on the XO-1's screaming 450mhz Geode plenty of times, and it turns in better numbers than that. I've also tried the original 600MHz celeron based EEEs(with el-cheapo 2gig flash disk, no less) and the browsing experience for anything but heavy flash was wholly unobjectionable.

Poor tests

[chill]

The VIA chip has built-in crypto accelerators and the idiots running the test pick something that doesn't use it! How about a with and without for comparison?

Re:Poor tests

[Chordonblue]

No doubt for all those wi-fi access points you'll be cracking on your Acer Netbook, right?

Re:Poor tests

[eddy]

Or maybe for free (as in performance) whole-system disk encryption?

Unfortunately, 'reviewers' think it beneath them to actually do any work beyond running their standardized tests. I've tried to reason with some of them before. They'll just continue running their LAME-MT and non-padlock enabled truecrypt or whatever. I tell you though, with Intel finally having crypto primitives in their new instruction set, they'll have to adapt sooner or later. Just as soon as Intel provide the how-to and/or software for them to blindly follow.

Re:Poor tests

[ofc]

The Nano has been available for a while now, but the (64-bit) Linux kernel still doesn't support the crypto accelerator and random number generator.
I recently bought a Nano based Jetway board and had add crypto and rng support myself. It's mostly working now, but I'm waiting for VIA documentation to do things the right way.

This just in...

[stephentyrone]

Processor with markedly higher power draw achieves superior benchmark results. News at 11.

Re:This just in...

[Anonymous Coward]

HUGE miscalculation considering I have a netbook (asus 900a) with the Atom N270 processor and I'm running Virtualbox with Windows XP, along with viewing webpages rather easily on Firefox in Linux.

I'm going to guess microseconds or even milliseconds was meant... but by far it's laughable that anything would take that many seconds for simple html.

And that while benchmarks are skewed against it

[Eukariote]

PCMark 2005 has been shown to yield wildly varying results for the nano depending on which CPU ID (CentaurHauls, GenuineAMD, AuthenticIntel) it is being presented with: http://arstechnica.com/hardware/reviews/2008/07/atom-nano-review.ars/6. Not surprisingly, if PCMark is made to think it is an Intel CPU, the benchmarks suddenly jump up across the board. Intel money buys good benchmarks.

Re:And that while benchmarks are skewed against it

[Anonymous Coward]

That's a really interesting link. Thanks!

Re:And that while benchmarks are skewed against it

[Leafheart]

Holy crap Batman, I wished I had mod points to mod you up

Is that really a win?

[Anonymous Coward]

It takes over a full minute to render a simple HTML file, and that's considered a win?

Re:Is that really a win?

[afidel]

Not A HTML page, HTML pages . i-bench is a browser torture test discontinued in 2003 and the HTML dates back to 2001 so it's not too relevant to today's web where CSS and DOM dominate, not table based layouts.

Re:Is that really a win?

[Colonel Korn]

Not A HTML page, HTML pages . i-bench is a browser torture test discontinued in 2003 and the HTML dates back to 2001 so it's not too relevant to today's web where CSS and DOM dominate, not table based layouts.

You, sir, need to be modded informative before another dozen "my netbook renders web pages in 2 seconds - they must be using Vista lolz" posts go up.

Now find one...

[Chordonblue]

The real news here is that even with these numbers, VIA will manage to blow whatever opportunity they have to gain advantage on netbooks.

It'll either be overpriced, hard to obtain in quantity or both. VIA seems to have a bad habit of showing stuff that, while it isn't vaporware, it's not something you'll actually SEE short of a consumer electronics show somewhere.

Re:Now find one...

[Eukariote]

Such are the joys of competing against a monopolist.

Re:Now find one...

[Zebedeu]

Are crappy support and drivers also part of those joys?

I've written off VIA a long time ago.

Intel Atom 330 turns the tables though

[Vigile]

http://www.pcper.com/article.php?aid=664

The benchmarks for the new Atom 330, dual-core HyperThreaded CPU seem to turn the tides though.

The Nano has ALWAYS been a better CPU than the Atom but that doesn't seem to matter when it comes to the push that Intel has...

Re:Intel Atom 330 turns the tables though

[Anonymous Coward]

The benchmarks for the new Atom 330, dual-core HyperThreaded CPU seem to turn the tides though.

Atom 330 benchmarks have been out for months, and Intel is limiting it to desktops.

Re:Intel Atom 330 turns the tables though

[Svartalf]

That's more because of TDP, I suspect.

Re:Intel Atom 330 turns the tables though

[BikeHelmet]

And the dual-core Nano will turn the tides on it - except that, you can't buy dual-core Nanos.

Intel won with the Atom because they actually sold Atoms. I can't find Nanos anywhere! I've been looking for a whole year.

Re:Intel Atom 330 turns the tables though

[PayPaI]

It should be noted that they are testing the Atom 330 with a nVidia GeForce 9400M chipset, which you cannot purchase today. The BOXD945GCLF2 motherboard with the Atom 330 (dual core) has the same 945GC chipset as the BOXD945GCLF motherboard that has the Atom 230 (single core), and that is the only dual core Atom cpu+chipset that you can purchase today.

Re:Intel Atom 330 turns the tables though

[jaavaaguru]

The Nano has ALWAYS been a better CPU than the Atom

Really? I always thought the Atom used less power, making it a better processor for use in netbooks (where the Atom is aimed at).

Re:Intel Atom 330 turns the tables though

[Mike McTernan]

Yeah, well Via's driver support for Linux is horrible, while the Intel Atom stuff just works perfectly with Ubuntu out of the box. That may have something to do with why Intel is preferred.

Re:Intel Atom 330 turns the tables though

[Fred_A]

Yeah, well Via's driver support for Linux is horrible, while the Intel Atom stuff just works perfectly with Ubuntu out of the box. That may have something to do with why Intel is preferred.

While that would potentially make me (and possibly a number of others here) prefer the Atom, since in Real Life almost nobody runs Linux on their desktop it doesn't have much impact on anything.

My laptop does run Linux but it's a regular Core2. It's 11" though so I can actually carry it around.

132 seconds to display simple HTML page?

[rolfwind]

"While the Atom needs 132.8 seconds to display simple HTML pages, the Nano does it in 70.1 seconds."

I was thinking of getting a netbook, but damn, not with that performance. Over 2 minutes? Is this a big miscalculation somehow?

Re:132 seconds to display simple HTML page?

[_avs_007]

I don't know what the hell kind of webpage they were trying to display... I have an Acer Netbook with the Atom in it, running Windows 7. It renders slashdot, ars, and even facebook, within 3 seconds or so...

Re:132 seconds to display simple HTML page?

[gEvil (beta)]

No idea what the hell they're trying to render, but my netbook renders just about every page I visit at about the same speed as my desktop machine. If there's any difference, I don't notice it.

Re:132 seconds to display simple HTML page?

[CannonballHead]

I wondered that, too. even 70 seconds is way too long. And they are referring to "simple" HTML pages. Maybe they meant milliseconds.

Re:132 seconds to display simple HTML page?

[Clueless Moron]

Ditto here (Acer Aspire One with XP, using a 1.6GHz Atom N270). About 3 seconds for slashdot.

If firefox is already cached, it starts up in only 3 seconds.

Re:132 seconds to display simple HTML page?

[Abreu]

I'll also join in and say those numbers are just wrong. My Acer AspireOne renders all kinds of webpages with no noticeable difference from my desktop machine

Re:132 seconds to display simple HTML page?

[tool462]

With the word 'pages' in there, I'm inclined to think they have a set of a pages they throw at it, probably picked to test a bunch of different types of elements to render or control for some other variable. The time is likely the aggregate time it takes to render all of the pages in total.

Re:132 seconds to display simple HTML page?

[_avs_007]

If that's the case, then what's the point?

Doesn't matter if the atom did it in 130 seconds, because 70 seconds still sucks.

Whose going to be doing that in real life? That's like comparing which is faster at encoding h264. Doesn't matter what the results are, becuase it'll suck for both, and nobody will be doing this on a netbook in the first place.

Re:132 seconds to display simple HTML page?

[Atti K.]

The benchmark was obviously rendering lots of simple HTML pages, not just one.

"While the Atom needs 132.8 seconds to display simple HTML pages, the Nano does it in 70.1 seconds."

Whoosh?

Re:132 seconds to display simple HTML page?

[tool462]

The point is that there are error bars on the measurements. Maybe the atom renders a single page in 500ms and the via in 300ms. That's all well and good, but what if the processor was handling some other background task for one or the other during that time? The extra few ms makes a difference. So, instead they have each one render some 100+ pages in a row. Little anomalies like that get averaged out, so you get a better comparison between the two.

Re:132 seconds to display simple HTML page?

[GweeDo]

Better context of this statement from the article would help. Because obviously single page renders are darn snappy (atleast on my Dell Mini 9 + Ubuntu 8.10). Do we have exact details on what pages they were trying to render in sequence to back your claim?

Re:132 seconds to display simple HTML page?

[corychristison]

I'd also like to know what the hell they were testing it with.

I'm running Gentoo on my MSI Wind U90. It boots pretty damn quick (about as quick as my desktop, actually) runs smooth for all tasks.

Re:132 seconds to display simple HTML page?

[Atario]

Obviously, my Netflix queue.

Re:132 seconds to display simple HTML page?

[Anonymous Coward]

Needless to say: Porn website with 132 animated pictures.

Re:132 seconds to display simple HTML page?

[Tenebrousedge]

I do it in real life, on my netbook. I save Firefox sessions and open about thirty pages at once whenever I load the application. The main point of a netbook is as an internet appliance, yes? Why are you surprised to find that one of the benchmarks involves opening web pages?

For encoding video, you are probably correct: anyone who is doing that is unlikely to be using a netbook. However, your post reads like an argument against benchmarking these computers at all. Perhaps you should reconsider your statements.

Re:132 seconds to display simple HTML page?

[Anonymous Coward]

They were running the benchmark on Vista

WTF Test?

[Crashspeeder]

There's data missing for the Atom in the wattage test to the 132 second HTML rendering, I'm not sure this test is anywhere near correct...for anything.

What kind of MONSTER HTML file are they throwing at these systems? Why put the Cinebench multi CPU benchmark up if it doesn't show any data at all except for the Pentium E5200 (the Atom is a single core CPU, why even run it?). And how is a Cinebench 64 bit test running on Vista 32 bit?

Re:WTF Test?

[dreemernj]

They used iBench 5.0 for the HTML rendering test. Its an out of date benchmark for testing HTML render speeds that doesn't really represent internet browsing as it is today.

I love VIA processors because I've had great experiences with them, but this particular "win" seems to have very little actual value to it.

Wait, what?

[Kokuyo]

Okay, I didn't RTFA... 70 seconds to display a html page? As in one minute and ten point one seconds?

Err... Please tell me that I'm missing something here.

Re:Wait, what?

[Anonymous Coward]

They are apparently trying to render an 80MB HTML file.

Re:Wait, what?

[gparent]

Please tell me you're kidding and you aren't _that_ clueless regarding iBench.

Re:Wait, what?

[Ant P.]

Err... Please tell me that I'm missing something here.

How about the s? On the end of "pages"?

Re:First Post

[Ethanol-fueled]

I don't believe you.

While the Atom needs 132.8 seconds to display simple HTML pages, the Nano does it in 70.1 seconds.

With those speeds, why do they call these things "netbooks?" :)

Clearly...

[Anonymous Coward]

They were using Windows Vista.

Atom has almost enough compute for me

[Francis]

*Disclaimer: I work for Intel
*Disclaimer 2: I actually do software research for Intel, and I haven't a clue about anything to do with hardware or business

I have a little EEE pc with an Atom 1.6GHz - I'm actually find it does have enough compute for most of what I do.

I did a stopwatch test on my computer - it takes less than 45 seconds from pushing the power button to getting on the network and rendering a web page. I'm running WinXP, but people have reported significantly better numbers with Linux.

The only time where I find I'm wishing for more compute power is when I'm watching HD flash video. (like Hulu or Youtube in HD mode - I get dropped frames)

I believe this is because Flash is written really quite poorly, and the video rendering code isn't very good. If I download the video and play it with VLC or something, it plays smoothly.

This is really the only reason I want more compute power on my eee pc. I'm actually hoping silverlight takes off so I don't have this problem anymore.

Re:Atom has almost enough compute for me

[Anonymous Coward]

Same disclaimers for me, I'm at Intel, have nothing to do with hardware design or anything like that.

But this is a nonsensical comparison. They compare Intel's not-top-of-the-line DESKTOP Atom board to their NETBOOK board.

The desktop Atom is actually crippled with a pretty lousy chipset. Based on the old 945G, as opposed to the mobile Atom getting a new chipset based on some third-party's chipset. (Sorry, this shows how little I deal with Atom; I just don't remember what graphics chipset it is.)

And, of course, they don't compare it to the dual-core Atom.

Re:Atom has almost enough compute for me

[Anonymous Coward]

Dr. Who, is that you? I just saw an interesting 32 nm demo at the EETimes video. It would be nice to see the game section of the demo, as well...

Re:Atom has almost enough compute for me

[Ant P.]

Just for fun: Power button to login, typing the URL to stackoverflow.com and letting it load = 1m10

This is on a stock eee701, with Debian+Xfce+Firefox. There's two big slowdowns - it takes several seconds to react to pressing the power button unless the AC is connected, and having an SD card in adds about 5 seconds. The SSD access LED isn't on constantly so I think there's room for improvement.

My real dislike with it is that resuming from standby takes almost as long as a cold boot.

Re:First Post

[electrosoccertux]

I don't believe you.

While the Atom needs 132.8 seconds to display simple HTML pages, the Nano does it in 70.1 seconds.

With those speeds, why do they call these things "netbooks?" :)

Very large web page. 17 seconds on an e5200 (That's a 2.5Ghz Core2Duo).

I had a feeling the second I learned the Atom was an In-Order processor that it was going to suck. Sure enough, it feels rather sluggish. Getting a dual core + dual threaded Atom might be better.

VIA's documentation is a nightmare to trudge through. Their hardware is usually great, but designing a product around it tends to be very difficult. With Intel, OTOH, we usually have no trouble getting a hold of an engineer if we have questions. Poor VIA...we'd love to use their chip but their support just isn't dependable when we have deadlines to meet.

I hope the netbook crowd (Acer esp) can muscle some legit documentation from them-- I'd take the Nano over the Atom any day.

Re:First Post

[Anonymous Coward]

Because you'll get a chance to catch up on reading a book while your browsing the net.

"outperforms nehalem"

[Anonymous Coward]

eh? Nowhere in that article does it come even close to the E5200 gimp cpu...

nehalem?

[Anonymous Coward]

It's not in the review at all.

No Intel Idle Power usage??

[Piranhaa]

http://www.zdnet.co.uk/i/z5/rv/2009/01/netbooks_pwr.jpg

Why doesn't Intel get scored on IDLE power consumption? Who cares about MAXIMUM when idle is the state that most of these netbooks will be in. wtf?

Re:No Intel Idle Power usage??

[Anonymous Coward]

http://www.zdnet.co.uk/i/z5/rv/2009/01/netbooks_pwr.jpg

Why doesn't Intel get scored on IDLE power consumption? Who cares about MAXIMUM when idle is the state that most of these netbooks will be in. wtf?

I can tell you what the missing bar would look like: about 45 or 46W, 2-3W less than the full load measurement. ZDNet reviewed the D945GCFL Intel Atom desktop board, not an actual netbook. This is important: the D945GCFL is how Intel has found a way to sell an ancient chipset which uses far more power than the Atom CPU it's paired with. (If you've seen a picture of the board, you'll see one small fanless heatsink and one large one with a fan. The large one with a fan is for the chipset, not the CPU.)

Since that chipset's power draw doesn't vary much (it doesn't really have a low power mode), the CPU's swing of 4W between idle and full activity (the desktop version of the Atom is rated at just 4W TDP) doesn't make much difference in the total system power.

Nobody builds netbooks with that chipset because Intel has a much lower power chipset designed specifically for Atom netbooks. Any review which claims to compare the Atom and Nano for suitability as netbook processors and uses Intel's desktop Atom board as a basis for comparison is extremely suspect.

Re:First Post

[c7fanboy]

Wow 100 guys from Austin, TX (www.centtech.com) can make an x86 bests Chipzilla?!?! Centaur Technology FTW!

price, performance, low power, pick two

[Anonymous Coward]

price, performance, low power

pick two

the nano might have better performance, but it's way more expensive.

Re:price, performance, low power, pick two

[c7fanboy]

Naa it is a cheap chip... The 1.6Ghz Desktop boards are much less than $100... The Samsung NC20 which uses the Nano is sub $500

Re:price, performance, low power, pick two

[Anonymous Coward]

I'm pretty sure you can classify sub 500$ as quite expensive compared to sub 100$, so the point still stands.

left in the stone age?

[cadu]

"The benchmark concludes that in some ways all netbooks, underpowered as they are, remain in the IT stone ages."

i don't know what kind of netbooks they're talking about, all newer netbooks (with decent resolution like 1024x600+ and 1gb of ram with a intel atom or via nano) perform VERY well, you can play quake3 in those using the onboard intel chip at the netbook lcd's native resolution, you can install windows xp and use that normally or go the [better] linux way and have a fully capable machine for programming, fun , studies.....

i used to listen to mp3s while programming on my first linux box , and that was a pentium 166mhz with 64mb of ram.....kernel 2.2.dontknow, can you guys tell me where 1.6ghz of processor with usb/wifi/bluetooth/1gb of ram/3d accelerated graphics is stone age? i wonder why they allow this kind of bullshit to reach slashdot's front page T__T

Re:left in the stone age?

[owlman17]

i used to listen to mp3s while programming on my first linux box , and that was a pentium 166mhz with 64mb of ram.....kernel 2.2.dontknow, can you guys tell me where 1.6ghz of processor with usb/wifi/bluetooth/1gb of ram/3d accelerated graphics is stone age? i wonder why they allow this kind of bullshit to reach slashdot's front page T__T

That's true. I did pretty much the same thing on a 128 MB Celeron 366. I programmed on it, surfed, listened to my mp3s. (It got a bit sluggish then but got the job done.) I played Q3 on it at 1024x768 - well I had a Riva TNT2 card. Sure, you won't be playing Crysis on a typical netbook anytime soon (except maybe on an Asus N10, albeit slowly) but calling it stone age tech is a bit of a stretch.

Re:First Post

[altek]

Agreed, same here. But I'm still waiting for the Ion platform to come out, which should best both of them, giving a good performance mark between the Atom/Nano and the Core 2 Duo.

power usage vs cpu performance

[j1m 5n0w]

Thanks for the info, that is the real story that was completely missed by the article. The main benefits of atom are obscured when attached to a lot of power-hungry peripherals.

Is it just me, or is this a really dumb article?

[0123456]

So they compare the power consumption of _Netbook_ CPUs by comparing the power consumption of _Desktop_ motherboards running _Vista_, when every netbook I'm aware of runs XP or Linux and Intel is world-renowned for having tied the desktop Atom to an appallingly crappy, backward and power-hungry chipset?

Then we should be surprised that they discover that a CPU which takes far more power than the Atom gives better performance in tests which are probably single-threaded and hence not even using the Atom's CPU to its full capacity?

You are an Impostor!

[Anonymous Coward]

Currently my Atom-powered netbook has clocked itself down to 1 Ghz; for posting on slashdot, nothing more is necessary.

You need at least a 2.5+ GHz Quadcore to fire off all those AC-Post on slashdot - I should know!

What browser are they using!!

[vosester]

"While the Atom needs 132.8 seconds to display simple HTML pages"

Good to see IE8 coming out of beta

one correction the E5200 is NOT nehalem

[Anonymous Coward]

the E5200 is wolfdale, the i7's are nahalem, get it right.... (nehalem stomps wolfdale). My i7-920 scores 93851 in zlib.

Netbooks might not be all that important.

[zigmeister]

It's not all a win for VIA, though. The benchmark concludes that in some ways all netbooks, underpowered as they are, remain in the IT stone ages."

Even if netbooks (whatever that comes to mean, netbooks keep evolving into more powerful machines, people start saying, "Ya email and surfing is great but what about some modern games and Matlab?", but I digress) don't turn out to be huge (I disagree, who doesn't want a machine with insane battery life?) the whole overpowered phones with intertubes will be huge. So if VIA plays their cards right, (and ARM for that matter) they could have a really huge untapped market on their hands.

God! I love the smell of statistics in the evening

[Anonymous Coward]

It gives me the warm fuzzy feeling in the tummy! On the more serious side of the news, I did not know they had IT in the stoneage, though maybe the Flintstones didn't show it all? Could be!

Nano's Crypto Smokes

[meehawl]

It's interesting they didn't run any real crypto tests that actually, you know, *used* the Nano properly. The Nano comes with the Padlock engine built in, for hardware crypto. With Padlock-aware software running crypto, the Nano "spanks" Core 2 Quads with lots of welly and gives even Intel's i7 a run for its money.

And Vista

[DingerX]

Plus benchmarking these things running Windows Vista? Are you nuts? I mean, XP is bad enough.

Never believe articles with marketspeak

[Kupfernigk]

Describing something as in the "IT stone age" is simply wild exaggeration. I was in the IT stone age, and I don't notice the tape punch on the side of my netbook. My evil minded suspicion is that the people who write the reviews are (a) quite detached from reality - they have no idea of real world user needs and (b) it's bad news for them if the industry suddenly says "OK, performance is good enough - let's just focus on cost, reliability and power consumption", because they won't have anything to write about.

If anything, the real issue with netbooks is that they are too good for the industry model. They make the majority of laptops look just too big, expensive and heavy. How many CIOs are wondering how many of their sales force really need Thinkpads and could get by quite happily with a netbook, a bluetooth mouse and a small projector? With the same technology in thin clients or low power desktops, you can do everything real users need, give the execs their Asus Lamborghinis or their Macbook Airs, and still save a lot of money at the next refresh.

As noted above, the answer to performance issues is to start fixing bloatware. Microsoft shareholders are rightly asking why so much money goes on very, very speculative R&D, but perhaps they should be asking what would happen if the same effort went into redesigning Office so that (a) it worked really fast on small low power computers and (b) the user interface was properly tailored for small screens. It can be done - Chrome makes more efficient use of small screen real estate than Firefox does, but some real investment could, I'm sure, make the whole thing so much better.

Out there, the real world is suddenly discovering that small cars are more economical, easier to drive and park than big ones, that big houses cost a lot to heat and cool, and that in general despite what marketing tries to tell you bigger is just bigger, not automatically better.

AES speed was tested - 32x faster than Atom

[Anonymous Coward]

Take a look at the AES encryption speeds for the Nano though:
http://www.zdnet.co.uk/i/z5/rv/2009/01/netbooks_ev_2.jpg

• 58,285 for the 1.8GHz Via Nano !!
• 9,294 for the 2.5GHz dual core Pentium
• 1,836 for the 1.6GHz Atom

From http://reviews.zdnet.co.uk/hardware/components/0,1000001694,39613937-3,00.htm:

In the AES test, the Nano even outperforms the dual-core Pentium, and substantially so. In fact, with a score of 58,285, the Nano is faster than a Nehalem processor.

Specialized hardware helps!

given my own experience...

[pjr.cc]

What surprises me is that its only twice as fast.

A Looong time ago, when htpc's were becoming something interesting in AU (not long before digital tv was broadcast on all stations in AU - around 2002). I bought an epia nehemiah m10k (there weren't many to choose from at the time, the m8k was the other one and the big diff between the two was one was passive cooled and the other active - i got the active one for the extra cpu grunt).

Anyways, a short time ago I noticed that intel were shipping mini-itx boards and I was initially interested in the conroe-capable (dual and quad core) mini-itx boards, but also happened to get a netbook (the original acer ones a110 i think?). What I wanted was a small box i could shove into the small space somewhere in our IDC for taking mail relay traffic and doing a small amount of web serving (php gallery).

So I installed fedora 10 onto the atom on an external 320gb usb hd. At some point I remembered the epia I had lying around (still had a 256m stick of ram in it - versus the atom's 1.5gb) and when it died originally I never bothered to figure out whether the power supply had bit the dust or the board itself.

Long story short, I plugged it into a new psu, and it managed to boot quite cleanly the exact same OS I had installed onto the ATOM.

So they were running an identical OS, the results were quite astounding. Running Super pi at 8m digits took 70s on the epia and 39s on the atom. I also ran a few others but they all came out roughly the same, the atom was about twice as quick.

Now remember the atom came out last year or late 2007 (cant remember) while the epia is from 2002.

So if the new nano is only twice as quick, i'd be surprised. But it is also somewhat more expensive (in AU at least). I'll probably still end up with atom 330 (dual-core) based itx board, just cause its easier to get my hands on.

2 minutes

[Anonymous Coward]

> While the Atom needs 132.8 seconds to display simple HTML pages

That's more than 2 minutes! Did they mean milliseconds or something?!?

Story and headline is WRONG, it should test 1.3ghz

[Anonymous Coward]

Story and headline is WRONG! It should test the 1.3 ghz Via Nano against the 1.6 or 1.66 GHz Intel Atom for the correct comparison for Netbooks. You will never see a 1.8 GHz Nano in a Netbook because it's too power hungry. To get down to 8W TDP, the Via Nano has to drop to 1.3 GHz. Furthermore, they failed to test multi-processor performance correctly which clearly has an advantage for Atom because of its hyperthreading capability.

Bullshit!

[Hurricane78]

And still, your whole dissertation -- which apparently comes straight out of your uninformed ass -- is completely useless,
since the Atom can only be so low in power usage, because all the power-draining stuff is in the north-bridge!

Have you ever looked at a board with an Atom CPU? The thing with the big fat cooler is the north-bridge. That what looks like the north-bridge is the actual Atom CPU!

And if you take the sum of the power those two chips need, they lose to every "netbook CPU"! By far...

But as long as there are parrots and retards who think because they can pull it out of their ass, it must be true, Intel will do just fine, selling its fraud chips. And you are directly responsible for that fraud too. So thank you... really... (NOT!)

Typo, question, reaction

[kenh]

First the typo - the board is D945GCLF

Second, why didn' they use the dual-core, hyper-threading Atom MB, the D945GCLF2? The latest board, the D945GCLF2 includes Gigabit Ethernet, not Fast Ethernet. Link

Finally, I've built systems with each of the "Little Falls" MBs from Intel, and all nice (considering cost) and very-capable MB/CPU combos. If the VIA CPUs are "better" that's great, but they tend to be very pricey by comparison ($85 for Intel vs. $285 for the VIA EPIA SN 1.8 GHz board referenced)