backtop


Print 66 comment(s) - last by JimmyC.. on Aug 4 at 6:58 PM


VIA Nano

Intel Atom
Intel's Atom gets some competition

Intel's Atom processor has become the darling of netbook manufacturers. The 45nm processor is used in a number of netbook designs ranging from the Eee PC 901/1000h to the MSI Wind.

Intel, however, is not alone in its quest to provide low-power, low-cost processor for mobile and desktop use. VIA is also throwing its hat into the ring with its 65nm Nano processor. VIA officially announced its Nano in late May of this year and promised parts ranging in speed from 800MHz on up to 1.8GHz.

Preliminary benchmarks provided by VIA showed the Nano outpacing Intel's Atom in many benchmarks -- today, PC Perspective and Hot Hardware put those claims to the test.

Both sites took an Intel Atom 230 (1.6GHz) processor paired with an Intel BOXD945GCLF motherboard and pitted it against a VIA Nano L2100 (1.8GHz) processor with a VIA Nano Reference Motherboard.

Although the VIA Nano was clocked 12.5% higher than the Intel Atom used in both site's testing, Hot Hardware found that the Nano was anywhere from 15% to 20% faster than the Atom. In many cases, the delta between the two was much greater in the Nano's favor. PC Perspective also found that under load, however, the difference in power consumption between the two processors was 17 watts with the Atom having the upper hand. Both sites came away impressed with VIA's Nano platform and its openness compared to the relative closed Intel Atom platform.

"Vendors are not allowed to build Atom motherboards with PCI Express, digital video outputs or more than one memory slot," said PC Perspective. "VIA on the other hand is openly courting board manufacturers to put as much technology on a mini-ITX design as they can - as long as they DO build one."

VIA sees the Nano processor as its ticket to success while Intel appears to fear Atom will cannibalize its Celeron sales."

"It's clear from our limited testing of the VIA Nano that the company has a strong product at the ready," added Hot Hardware. "Virtually across the board, the Nano showed good performance versus its primary competition, and it has a number of other things going for it as well, like pin-compatibility with the established C7 core, an open platform, and hardware acceleration for certain types of encryption."

It will be interesting to see how much support VIA gets from OEMs with its Nano processor. OEMs which choose to use VIA's OpenBook reference design will no doubt take advantage of the new chip and HP might just throw out the VIA C7-M currently residing in the 2133 Mini-Note PC in favor of the Nano.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Long time Via Fan
By masher2 (blog) on 7/29/2008 3:00:07 PM , Rating: 4
As a long time fan of Via's Mini-ITX motherboards (I have two of them built into the walls of my house), I'm glad to see they're continuing to refine and improve the platform.




RE: Long time Via Fan
By FITCamaro on 7/29/2008 3:15:41 PM , Rating: 3
What are they used for in your house Mike?


RE: Long time Via Fan
By nosfe on 7/29/2008 3:26:47 PM , Rating: 5
HTPC for the termites


RE: Long time Via Fan
By JasonMick (blog) on 7/29/2008 3:27:45 PM , Rating: 5
I suspect they control the security systems to protect against over-zealous AGW supporters should they ever assail his stronghold, I suspect ;)


RE: Long time Via Fan
By masher2 (blog) on 7/29/2008 3:42:07 PM , Rating: 5
The dead baby seals I have staked up around the perimeter should repel those.


RE: Long time Via Fan
By psychobriggsy on 7/30/2008 5:03:11 AM , Rating: 4
They might use the blubber as bio-fuel, thus enabling them to get to your front door!

But then all they'd do is wave a stick of celery at you.


RE: Long time Via Fan
By masher2 (blog) on 7/29/2008 3:29:23 PM , Rating: 2
One controls lighting and my zoned heating and a/c ducts; the other acts as a house web/file server.

I've had plans to add some more automation functions and a tie to the security and phone system, but I never seem to get around to it.


RE: Long time Via Fan
By teldar on 7/29/2008 4:26:13 PM , Rating: 2
What do you use as an interface? Web controlled or touch panels?


RE: Long time Via Fan
By masher2 (blog) on 7/29/2008 5:31:52 PM , Rating: 2
Web interface, plus some led-lit keypads for the lighting (not touch screens...I was on a budget).


RE: Long time Via Fan
By FITCamaro on 7/29/2008 4:31:30 PM , Rating: 2
Sounds pretty awesome Mike. Write the software yourself?


RE: Long time Via Fan
By masher2 (blog) on 7/29/2008 5:32:53 PM , Rating: 2
The low-level software I did...it's written in Atmel machine language. The higher-level code is just a microlinux distro.


RE: Long time Via Fan
By Rodney McNaggerton on 7/30/2008 1:26:35 AM , Rating: 4
Would you write a blog about a newbiews guide to such a thing? From A-Z?


RE: Long time Via Fan
By masher2 (blog) on 7/30/2008 3:54:31 AM , Rating: 3
I might at some point, if there's enough interest.


RE: Long time Via Fan
By aBott on 7/30/2008 8:54:36 AM , Rating: 3
I'm interested! As are a few friends of mine.


RE: Long time Via Fan
By PAPutzback on 7/30/2008 9:54:47 AM , Rating: 3
I am interested also. Do It!


RE: Long time Via Fan
By Denigrate on 7/30/2008 12:25:46 PM , Rating: 2
Count me in as a very interested party.


RE: Long time Via Fan
By DM0407 on 7/30/2008 4:05:35 PM , Rating: 2
Ditto


RE: Long time Via Fan
By dude on 7/31/2008 5:23:20 PM , Rating: 2
I would as well, as I may want to implement this in my house in the future.

It's pretty interesting


RE: Long time Via Fan
By JimmyC on 8/4/2008 6:58:26 PM , Rating: 2
Word!! +1 for the writeup.


RE: Long time Via Fan
By Ringold on 7/30/2008 1:32:34 PM , Rating: 3
I'm interested, too. Sounds like there is potential for serious geek-points without spending several thousand bucks. I couldn't code my way out of a paper bag, but where there is a will there is a way.


RE: Long time Via Fan
By RyanM on 7/30/2008 1:36:55 PM , Rating: 3
Color me intrigued. When I put structured wiring into my house, I always wondered if I could ever get around to any home automation. I'd love to see a step-by-step on this.


RE: Long time Via Fan
By deeznuts on 7/30/2008 2:55:35 PM , Rating: 2
Just do it already!


RE: Long time Via Fan
By maverick85wd on 7/31/2008 4:02:31 PM , Rating: 2
a home automation tutorial blog post in the near future would warmly regarded, by myself as well as apparently quite a few others


RE: Long time Via Fan
By 4wardtristan on 7/31/2008 8:35:00 PM , Rating: 2
+1 for keen-ness


RE: Long time Via Fan
By Oobu on 7/30/2008 3:24:48 PM , Rating: 2
If you wrote a guide for this you would pave your way into Internet geek heaven. Your name would never be forgotten.


RE: Long time Via Fan
By jhb116 on 7/31/2008 12:46:20 AM , Rating: 2
This sounds like a worthy article on a tech site - maybe like this one. :)


Sounds good
By FITCamaro on 7/29/2008 2:43:23 PM , Rating: 4
Lets see those Nanobooks start rolling out. Nanobook. Has a nice ring to it. :)




RE: Sounds good
By Smartless on 7/29/2008 2:49:48 PM , Rating: 2
Hmm as long Nano isn't synonymous with Apple. Otherwise it might be a Macbook Nano or iMac Nano or a iNannynannybooboo.


RE: Sounds good
By Denithor on 7/29/2008 2:50:40 PM , Rating: 5
Certainly sounds better than Atombook doesn't it?


RE: Sounds good
By nosfe on 7/29/2008 2:58:12 PM , Rating: 5
Up and At 'em, Atom Ant!


RE: Sounds good
By HakonPCA on 7/29/2008 3:30:27 PM , Rating: 5
no its:

"Up and At them"

(no Mr. wolfcastle its up and atom...)

"Up and At them"

"...my eyes...the goggles do nothing"


RE: Sounds good
By nosfe on 7/29/2008 3:34:02 PM , Rating: 3
RE: Sounds good
By HakonPCA on 7/29/2008 3:48:24 PM , Rating: 2
none needed, thanks ;) I've picked the reference I prefer, you picked one you prefer


RE: Sounds good
By Etsp on 7/29/2008 4:04:10 PM , Rating: 3
There is more than one path to enlightenment my child...
http://uk.imdb.com/title/tt0701201/quotes


RE: Sounds good
By DM0407 on 7/30/2008 4:08:29 PM , Rating: 2
Random link that won't enlighten anyone:
http://www.hinduonnet.com/seta/2002/09/19/images/2...


Power Consumption Correction
By Khato on 7/29/2008 3:02:55 PM , Rating: 2
The PC Perspective article is stating that the atom only uses approximately 4 more watts under load compared to idle. The consumption difference between the nano and atom and load, meanwhile, is around 17 watts.

They do go into analysis of how the nano actually apparently uses a bit less energy to actually compute a task, though that methodology is somewhat flawed as it doesn't appear to take into account the energy consumed while idling afterwards. That's the thing... Atom consumes basically the same whether under load or idling.




By Brandon Hill (blog) on 7/29/2008 3:04:57 PM , Rating: 2
Thanks for pointing that out -- it has been corrected in the article.


RE: Power Consumption Correction
By kkwst2 on 7/29/2008 3:33:05 PM , Rating: 2
I think you're being kind. I understand what they were trying to do, but it seems misguided.

If you look at the power graphs, the Via spends almost the whole time at or above the Atom in power consumption. So for a given period of time to get the energy consumption you integrate this over that time. Clearly the Via used more power during a given time period. Sure in theory you could go right into another task, but that's not reality. As you said, the idle time must be factored in. I guess it shows the complexity in trying to compare power consumption between products of differing performance.

Now, whether the trade off of (significantly) more performance for a small hit in power is worth it probably depends on your needs. It certainly looks like a competitive product to me, and it's great to see the competition. I would think this would force Intel to allow more flexibility in motherboard design. However, to try to spin it as the Via consuming less power seems very misleading to me.


RE: Power Consumption Correction
By omnicronx on 7/29/2008 4:24:39 PM , Rating: 3
Am I the only one to notice that right now the atom must be paired with a dated i945 chipset that consumes 22w. until they fix this, it really does not matter that the nanos load is 10w higher than the atom, because there are via chipsets out there that consume much less power.


RE: Power Consumption Correction
By Khato on 7/29/2008 4:49:40 PM , Rating: 2
Except that these measurements are total system power consumption, and hence already include Via's rather large chipset advantage.

Really, I was expecting -far- more performance from the nano. As it stands, it's not beating out the atom by much once the clock speed advantage is factored in, and it's consuming far more power. Only thing that can be considered as holding back atom at current is the platform - no question that the current chipset is not well suited to the task.


By masher2 (blog) on 7/29/2008 5:37:32 PM , Rating: 3
> "The Nano [is] consuming far more power..."

From the article:

quote:
Using the same method to gauge the results of our CineBench 10 test, we find that the VIA Nano used 63,434 watt-seconds (Joules) of energy to render the scene while the Intel Atom used 65,893 watt-seconds (Joules) of energy - an advantage of 3.8% to the VIA CPU.
A secondary advantage of the Nano is the VIA mobos tend to have far more functionality for this market, like being able to directly drive LVDS displays, use 12v power without an external power supply, etc.

Still, I think Intel had made an admirable entry, and once the chipsets catch up to the Atom, its going to wind up dominating Via.


RE: Power Consumption Correction
By omnicronx on 7/29/2008 8:39:41 PM , Rating: 2
quote:
Really, I was expecting -far- more performance from the nano. As it stands, it's not beating out the atom by much once the clock speed advantage is factored in, and it's consuming far more power.
Actually its right on par, if you take into accout how much extra time it takes the atom to process the exact same tasks. For example in one of the benchmarks, the nano took around 30% less time to encode an mp3 which means it gets to idle for that much longer, in this case the nano actually consumed less power when taking this into account. Considering that processors like these are not going to be used even close to 100% of the time, this makes a huge difference.

I would also like to add that the nano is an out of order processor (atom is in order), so any task that the atom can not perform its version of hyperthreading on, we see a big difference in performance.


RE: Power Consumption Correction
By Khato on 7/30/2008 12:00:56 PM , Rating: 2
quote:
Actually its right on par, if you take into accout how much extra time it takes the atom to process the exact same tasks. For example in one of the benchmarks, the nano took around 30% less time to encode an mp3 which means it gets to idle for that much longer, in this case the nano actually consumed less power when taking this into account. Considering that processors like these are not going to be used even close to 100% of the time, this makes a huge difference.


Actually, this is incorrect. Why? Basically because the atom platform under load only draws maybe 1 watt more than the nano platform idling, which means that the atom platform would have to take 17 times longer to complete the task for energy equality.

I know the PC Perspective article said that the two platforms are basically equal power draw to complete a task, and that is true, but after the task is complete the nano is still going to be drawing its idle power, which they are not taking into account. All they're doing for power consumption per task is power draw * length of task, so for the MP3 encoding, basically 77.5 * 465 for the nano, and 60.1 * 605 for the atom - those don't get their exact numbers, but it's close enough to show that that's what they're doing. Now the issue is that after the nano's done, if there's nothing else to do it's just going to be idling for 140 seconds, which suddenly puts it at consumping around 20% more power.

The correct way to do what PC Perspective was attempting is to calculate the load energy usage while completing the task - 17 * 465 for nano, 4 * 605 for atom. This tells you that nano used 7905 joules more than idling to complete the task, while atom only needed 2420. So yeah, nano is by -no- means using less power to complete tasks.

As for the OOO versus in order, trust me, I know. That's why I'm surprised that the atom is keeping up so well. With a 12.5% clock speed advantage, nano is only managing to outpace atom by 30%. Considering the difference in power consumption, that's pretty pathetic. Sure, there's a market for them, especially until the second generation atom comes out, but that market is only there because of price - low voltage core processors consume less power and easily beat out the nano in performance.


RE: Power Consumption Correction
By jconan on 7/29/2008 9:27:20 PM , Rating: 2
but it's 65nm vs 45nm. can't wait 2 see how a 45nm nano will compare to atom...


RE: Power Consumption Correction
By Mojo the Monkey on 7/30/2008 1:04:26 PM , Rating: 2
Phew! I almost grew tired of reading your whole sentence, but you saved that single keystroke and one whole character by using the number two, to indicate the actual word "to"; all the while sacrificing any shred of credibility that anyone with an IQ higher than 50 might have otherwise credited to you. The sheer genius and cunning of your grammatical shortcut is astounding. Bravo.


RE: Power Consumption Correction
By Oobu on 7/30/2008 3:28:43 PM , Rating: 2
This is the Internet pal. There's no way around that. Oh, and thanks for wasting 15 seconds of my life.


RE: Power Consumption Correction
By mindless1 on 7/30/2008 5:44:14 PM , Rating: 2
Yes, the internet, not a text message on a phone with half it's buttons missing.


RE: Power Consumption Correction
By flydian on 7/31/2008 1:03:25 AM , Rating: 2
Bravo! Bravo! Thank you for saying what I've been thinking for years!

P.S. And you too mindless1


miss something?
By skiboysteve on 7/29/2008 3:50:28 PM , Rating: 4
Somehow both reviews and this article seem to miss the huge difference between them. Size. Physical size.

Atom is smaller than a penny, the via chip is absolutely huge in comparison.

It uses more power and its bigger, and it has higher performance. How is this at all a surprise? People think this is a huge achievement.... intel could do the same thing. They are targeted at different size/power envelopes... they arent really 'competitors'




RE: miss something?
By tfranzese on 7/29/2008 4:03:16 PM , Rating: 2
Did you even bother to look at the cited articles or did you just look at the pretty pictures at the top of this page? They did compare size and they both have approximately the same package size. Die sizes are also pretty similar, with Intel having the advantage especially due to the 45nm process.


RE: miss something?
By skiboysteve on 7/29/2008 4:07:39 PM , Rating: 2
hot hardware didnt say anything about the size.
pc perspective showed it slightly smaller than a quarter..


RE: miss something?
By defter on 7/30/2008 2:47:26 AM , Rating: 2
Die sizes are NOT similar as can be seen from this picture: http://www.pcper.com/images/reviews/597/chip_compa...

According to official information Atom has a die size of 25mm^2 while Nano is 63.3mm^2. A 45nm shrink of Nano would be at least 35mm^2, still significantly bigger than Atom.


RE: miss something?
By Aloonatic on 7/30/2008 4:46:27 AM , Rating: 2
Is that extra size mostly down to the larger cache on the Nano?

How likely is it that VIA will produce a 45nm version?

And if they do, will Intel have 32nm Atom ready by then?

Is there any more news on the Atom Duo (or whatever it's called) yet, by the way?


RE: miss something?
By psychobriggsy on 7/30/2008 5:20:15 AM , Rating: 2
A 45nm version of nano is clearly on VIA's plan, probably for next year when TSMC will offer it, otherwise it will be when Fujitsu offer it (as it is made on a 65nm Fujitsu process currently, and that is one of the possible reasons it gets so hot).

Hopefully it would reduce power consumption, or allow increased clocks, thus making nano even more competitive against Atom. In addition, Intel would have to prioritise Atom on their 32nm process when it is ready, yet Intel don't appear to be that bothered with Atom overall, they don't want it to eat away at their low-end Celeron market for one.

I would be interested to see how a 1.4GHz nano performed, as this speed would appear to be equivalent to the Atom, yet it could consume a lot less power than at 1.8GHz.

Clearly the mini-ITX boards using nano (at any speed or wattage) are going to be far better than any atom variant because of Intel's restrictions.


RE: miss something?
By ajfink on 7/29/2008 4:12:38 PM , Rating: 2
Yes and no. In the Nettop arena, Nano is the clear winner. In the UMPC sector, they overlap with pros and cons, and in the really small form factors, Atom wins.


Different markets with that power consumption
By strikeback03 on 7/29/2008 3:20:18 PM , Rating: 2
Sounds like the Nano power consumption is more in line with Intel's C2D notebook processors, so for any device where battery size is a factor the Atom will have a large advantage.

Also compatibility remains to be seen, VIA has not had the best record of compatibility with Linux for example.




By sliderule on 7/29/2008 3:34:52 PM , Rating: 3
These are my concerns too. Battery life, and linux drivers.

I like HP's mini note, but they're lacking in those two areas. It's a shame to because recently someone from HP announced they were thinking of lowering the quality of the case(I assume they're going to a plastic case)to lower the cost of the machine. As if the price is the only reason they aren't flying off the shelves.

For me, the main appeal of the HP mini note is its durable aluminum case, and nice keyboard...if only it had intel inside. :(


Pin Compatable?
By littleprince on 7/29/2008 3:09:58 PM , Rating: 2
So its the same size footprint wise as the C7-M now?
Or did I misread? Or DT misread?




RE: Pin Compatable?
By Brandon Hill (blog) on 7/29/2008 3:12:51 PM , Rating: 2
Yes, it is pin compatible with the C7-M.


Looking at it all wrong.
By pauldovi on 7/29/2008 3:15:07 PM , Rating: 2
The Atom is good not because it is fast, but because it is cheap to manufature and it is doesn't use a lot of power. I cannot see how the 65nm product can compete with the 45nm Atom.




RE: Looking at it all wrong.
By wordsworm on 7/29/2008 9:33:43 PM , Rating: 1
I think you're looking at it all wrong. I wouldn't be surprised if you could soon go out and buy a Nano. Try doing that with an Atom. I thought I was in love with the Atom until I found out it only did 32 bit. Then I started to dislike it when I found out it was basically impossible to buy without a machine.

The Atom looks like its marketing department is afraid of it. I think there's a good DT article that discusses this issue. It's the little CPU that could that's the black sheep of the family because of its economy and performance.

What I would like to see is Via get together with AMD to make a good ITX board with 'good' graphics.

Via is the dark horse in this race. I always like the underdogs, and right now, this underdog just grew a nice new set of teeth.


Comeback for VIA?
By subhajit on 7/29/2008 2:54:37 PM , Rating: 2
They certainly need it. I Hope their chipset department also comes up with something good.




Also
By sprockkets on 7/29/2008 2:57:28 PM , Rating: 2
The nano is a 65nm processor and the atom is 45nm. While the atom will be great for those phones, nano will be much better for actual computers.

WTB one with an nVidia chipset or at least a PCI-e slot. Although there are pics of such atom boards with advanced features, Intel just doesn't get the fact that I would buy a mini-itx Intel board if it didn't cost $225 instead of $90.

And let's not get started on AMD's DOA DTX.




Dual-core Nano
By Suomynona on 7/29/2008 3:49:22 PM , Rating: 2
I can't wait to see them slap two of these on a die. The Nano might not compete too well with Atom in the mobile sector, but it should make a nice mini-desktop platform.




Some observations
By tfranzese on 7/29/2008 4:08:02 PM , Rating: 2
I'm baffled why VIA is using 16-way set associativity, seems like overkill based on traditional computer architecture studies. Between that and the enormous caches for what aims to be a low power, small processor they could easily come close to Intel for die size and bring down that power consumption without going to 45nm. Not sure how dependent on a large cache the Nano is, anyone?




Entirely Disappointing
By mindless1 on 7/30/2008 6:18:08 PM , Rating: 2
IMO, both companies have really missed the mark. Intel's development of the Atom without a more power conservative northbridge, then adding a southbridge too, is a bit of a sad joke against the target of small size and efficiency.

Via has a similar problem, thinking only about a tiny processor instead of on-die board chipset. Both companies really need to look hard at the intended applications and that nothing makes sense except a CPU-Chipset hybrid.

Further, the 20 pin regular ATX power requirement, and even worse the addt'l 4 pin 12V on intel's platform is an unneeded and excessive compatibility compromise. Both should use a 12V DC barrel power connector for input from a power brick and onboard conversion to any voltages required.

In the end, we have platforms that aren't very tiny, aren't all that inexpensive. They do very little to displace their predecessors, you don't even get digital output onboard what has to be seen as a complete product since the CPU is soldered on. At least there's hopes for the Via platforms but looking at the past Via platform boards, most were rather ridiculously priced for what they are.

At least there were a couple positive notes for Via. Onboard CF slot and open platform. The Nano's power consumption is too high though, even on this front they fail.

<Yawn> I kept hoping via was looking to do something more interesting than just the cheapest CPU they felt they could sell at all but it looks like the same thing, different name, different day.




"We don't know how to make a $500 computer that's not a piece of junk." -- Apple CEO Steve Jobs














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki