Back to Article

  • StevoLincolnite - Monday, May 28, 2012 - link

    Sad to see history repeating itself with the Intel Decelerators and drivers.
    Never again will I get a system that uses Intel graphics, AMD and Nvidia you can entrust they will update drivers frequently and gain performance over time...

    Still rocking an Atom 330 + nVidia Ion in my Mini-ITX rig; and other than Brazos... There is really no options available to upgrade it yet even after several years, but the machine does it job.
  • zeo - Wednesday, May 30, 2012 - link

    Cedar Trail ATOM isn't even using a Intel GPU! The 3600/3650 GMAs are using a Imagination PowerVR GPU! The SGX545 to be precise...

    So it's the 3rd party support that's lousy, and same problem Intel had with the GMA 500, based on the Imagination PowerVR SGX535, a few years ago... but Intel is working on improving the drivers, they're just focused on getting the drivers ready for Windows 8 release and so Windows 7 support has been put on the back burner till then.

    While Imagination has never supported Open Source drivers. So Linux users are on their own.

    However, Intel is going back to their own GPU with the next 22nm Silvermont update.

    While Intel isn't so bad in supporting their own GPU based GMAs, but they've never been known for great performance.

    Though, the HD4000 seems to have reached the okay for entry level mark and the next Haswell update promises to raise graphical performance by another 50%. While the 22nm Silvermont update may be using a GMA based on the HD4000.

    So while they still probably won't be breaking any performance records, they should be providing more respectful performance by the middle of next year.
  • ViperV990 - Tuesday, May 29, 2012 - link

    How are you measuring the power consumption? Does the 17W load figure at the input or output of the PSU? Reply
  • ViperV990 - Tuesday, May 29, 2012 - link

    s/Does/Is/ :p Reply
  • JarredWalton - Tuesday, May 29, 2012 - link

    Power is measured at the wall, so PSU efficiency is a factor. If it's a good PSU with 80% or higher efficiency, then the system is using a few watts less actual power. Reply
  • MrSpadge - Tuesday, May 29, 2012 - link

    "Load temperatures do break the 15-watt TDP that Logic Supply advertises, but only by two watts, certainly still reasonable."

    1st: it should read "power consumption" instead of "temperature".

    2nd: if you're measuring at the wall, at least 1 W is consumed in the PSU, probably a bit more. Including rounding errors this leaves us straight at the 15 W Logic Supply claims for the unit. I guess they were not counting the loss in the PSU.
  • JarredWalton - Tuesday, May 29, 2012 - link

    Thanks, edited for both items. Reply
  • DesertCat - Tuesday, May 29, 2012 - link

    Not that I'm sure that it will fix the reviewer's problems but Acer has a newer video driver from Intel on their site for their N2600-based netbook (Aspire One AOD-270). I think the reviewer was probably using the 1065 driver that is up on Intel's site (hence the comment about them being 4 months old), but Acer has the The latter one has a release date of March 20th. Might be worth a shot. Reply
  • funtasticguy - Wednesday, May 30, 2012 - link

    Hey, DesertCat, thanks for that tip. My new Gateway LT4004u had the same original 1065 driver. So I went to Acer's website and downloaded the 1075 driver and installed it on my Gateway. Now, finally XBMC Eden runs well (before I had this annoying flickering that rendered XBMC useless). In addition, my 720p and 1080p videos now work 100% perfectly within XBMC. So, Intel has made some progress and I'm a happy camper. It still won't run my PSX1 emulator well and some other games, but I suspect that as Intel updates it's driver again, all my emulators and games should work well.

    By the way, my Gateway has a 10+ battery life. It is also rather speedy and I'm happy with my purchase now -- especially since XBMC works well now and I can use it as a portable media player when I travel out of town.

    Again, thanks for the tip!
  • mfed3 - Tuesday, May 29, 2012 - link

    Ridiculously overpriced. I like the form factor for possibly using it as a hypervisor for a pfSense router and Windows Media Center tuner pool but the price tag is just stupid. Reply
  • JarredWalton - Tuesday, May 29, 2012 - link

    This is an industrial kiosk-style nettop, so it's not really intended for mainstream use. We were mostly interested in it to see how it would perform and handle Windows in practice, and it's been a disappointment for Dustin.

    I've got an ASUS 1025C netbook that's a bit better, but while local H.264 content seems to play back well enough, even at 1080p, the CPU is still a dog. MPC-HC can play a 1080p video fine...until you start trying to use the UI for something, at which point it becomes choppy. Internet video on the other hand is pretty much out: HD Netflix completely fails to keep A/V sync; Hulu is okay at SD resolutions, but the UI is still slow when used; and YouTube is pretty much dropping a few frames at 720p and dropping a lot of frames at 1080p, leaving only 480p or lower as "working properly".
  • Metaluna - Tuesday, May 29, 2012 - link

    For that I would take a look at Supermicro's Atom D525-based server board. Dual NICs and IPMI for around $220, IIRC. That's still kind of pricey, but coming in under $678 isn't too tough :). You do give up the nice form factor, but you can probably get close with some of the smaller mITX cases. Reply
  • rs2 - Tuesday, May 29, 2012 - link

    I agree 100%. I put together a similar Atom build (4 GB RAM, SSD, fanless) over a year ago for less than half of the "price as configured". Granted my case wasn't as small and fancy looking, but even allowing for that Logic Supply is charging a ridiculous markup, possibly approaching 100% of the actual build cost.

    A 20% markup would be more reasonable, and would see these devices selling in the $275 to $400 range. At that price point I might consider one. Otherwise forget it, it's more cost effective to just build a comparable machine myself.
  • name99 - Saturday, June 02, 2012 - link

    It's worth pointing out that a Mac mini ($599 for low-end model) is also 13W idle power.
    Of course it uses more when it's actually working, but it also does more.
    If your usage model has the machine mostly idle, it's probably competitive.

    I'm not saying this as a "rah rah Apple" comment --- add in the cost of Windows, and the fact that Windows probably won't do as good a job conserving power, and you're not that competitive anymore.
    My point was, rather, that this is just not that impressive. Crappy, supposedly low-power CPU, and industrial box vs computer with working graphics at much the same price and (for at least some usage models) much the same power usage.

    I mean seriously --- as others have said --- ridiculously overpriced.
  • Alyx - Thursday, June 07, 2012 - link

    I also dropped in here to see if this could be an upgrade to our current firewall. I think most of these things end up super expensive for some reason, but for their target market the price isn't too high really because its competing with desktops rather than netbooks.

    Sadly without multi-nic this unit isn't much use in pfsense. We have a number of these little guys and they are pretty rock solid and the 7 nics is rare for fan-less units like this (our company has multiple secured LANs).
  • Impulses - Tuesday, May 29, 2012 - link

    Intel entered the netbook market just to avoid ceding it to AMD/VIA, but they never wanted it to thrive. The whole ultrabook initiative was a way of driving costumers to spend more on the average system (admittedly for improved build quality at times). Atom cannibalized Intel's bottom offerings and they'd happily let it linger even longer if they could. They've clearly left a hole others should be able to exploit... Now, where's Trinity? Reply
  • JarredWalton - Tuesday, May 29, 2012 - link

    Right now? Trinity A10 is in $700+ laptops is all. I'm not sure if there's anything preventing someone from using mobile Trinity in a desktop/nettop, but I suspect it will be a couple months before we see anything. Reply
  • zeo - Wednesday, May 30, 2012 - link

    Intel did not enter the market to avoid ceding anything to AMD or VIA. Besides, which only VIA was ever anywhere near doing so at the time.

    AMD didn't have anything to compete in that range until they came out with the Fusion series in the beginning of 2011, well after the netbook market had been established.

    Netbooks were just a inevitable result of the desire for Mini-Laptops that goes back over thirty years now. While the OLPC project is primarily credited for accelerating this process and applying the actual pressure on Intel. Thus why Intel already had the ATOM ready for production when the netbook market established itself, but it wasn't their first try at making lower cost and more energy efficient processors.

    The very first netbook in fact was released with a Intel Celeron M 353 ULV processor.

    You are right that Intel didn't want the netbook market to spread, but it was because of the low profit margins of the netbook market and they accomplished that by setting restrictions and guidelines on netbook design. Though in large part many of those policies helped lower the cost of netbooks and made them even more affordable over time.

    However, they relaxed those limitations over the years and Cedar Trail is actually the last of the old 5 year product cycle of the ATOM. So starting with the 22nm Silvermont update the ATOM will go on the same 2 year product cycle as the Intel Core i-Series.

    Intel is even on record stating they will start developing the ATOM at faster than Moore's Law for at least the next two years to catch it up. Since Intel wants to seriously get into the mobile market and the ATOM is their best bet at doing so.

    While they may have waited too long because ARM has finally reached the point that they can offer solutions to rival the ATOM and so Intel faces challenges in both making its way into the Mobile market and holding on to the low end PC market as well.

    For AMD the focus is mainly on graphical performance, so that's where they shine in comparison to Intel's offerings, and they're sticking mainly to the traditional PC market. However, the rate of progress seems more or less lock step with Intel's and we shouldn't expect too much from either until next years updates.

    Like for example AMD's Tamesh will probably be their first offering that can go fan-less and compete in the tablet market but it won't be out till next year.

    While Trinity is basically just a update to Llano and more intended to compete with Intel's mid range offerings but should provide a alternative for Intel's Ultrabooks.
  • Mugur - Tuesday, May 29, 2012 - link

    ... afterall. Judging by the 2 serial ports. For a pfSense router a second NIC would have been nice.

    No home use for this one.
  • JarredWalton - Tuesday, May 29, 2012 - link

    Um, which part of the "industrial kiosk" target market did you not get? Such stations might very well need two COM ports. Granted, there are plenty of other markets that would not need this configuration (like pfSense routers, home users, etc.), but it's designed for a very specific niche. Now, is Cedar Trail a good fit for that niche -- better than say AMD's Brazos/Ontario? I guess Cedar Trail is lower power at 6.5W TDP, and it's totally fanless, but wow I have a hard time believing an extra 10W would matter that much for a kiosk location with a probably 50W display and various other lights and such that use another 100W+. Reply
  • hobbesmaster - Tuesday, May 29, 2012 - link

    I'm assuming he meant POS as "Point of sale" instead of "piece of s***" The two often go hand in hand, but the issues are usually software!

    A lot of those systems are moving to USB though from what I understand. Regardless, this should handle an XGA touch screen fine which is probably all its intended to do.

    They also list this under automotive, that wouldn't be a bad application either. You'd have a WVGA screen at max there.
  • beginner99 - Tuesday, May 29, 2012 - link

    Even though its obviously intels lack of creating useful drivers once more as the producer of a system requiring such a driver i would not buy it if it doesn't Logic IMHO is to blame too at least partially. But it seem to be the trend to release "faulty" products anyway. Reply
  • khimera2000 - Tuesday, May 29, 2012 - link

    This review has some nice information, but it leaves so many questions that should not of cropped up. How does it perform against the last generation on both sides? how does it compare to the current generation from its competitor?

    The last time I checked it was the E-series that AMD marked as there "power efficient line" why is this being compared to an A6? Why was it not compared to an E-450, or at least a E-350? at least these two would be in the same area, and so would be perfect for comparison against an Atom N2800.

    Without a point of comparison against other power efficient parts I look at the Data as a wash of useless. there's nothing in there that's relevant if its not seeing how it does against its competition.... and at least in my head that's the E-350, E-450, and VIA's cluster off product offerings.

    I don't care who comes out on top, I just want to know how heavy the trade off is at this level of power draw, I want to know if bad drivers can result in a system that has to push hotter then its competiters, I would like to see how far these chips have advanced... also a brief rundown on each companies secret sauce would be a nice touch... refreshers always help :)
  • Spunjji - Tuesday, May 29, 2012 - link

    Looks like you beat me to it! ;)

    I'm guessing the usual caveat of testing the systems they had to hand applies here, but you're right, the comparative data isn't of much use.

    Credit where it's due though; this information does at least tell us that Atom is somehow more useless than ever on the desktop.
  • JarredWalton - Tuesday, May 29, 2012 - link

    Bingo! Unfortunately, most of what we have on hand is limited, and I'm pretty sure none of us are interested in buying our own low-end Atom nettops. It's partly a problem of simply not getting enough of these types of systems for review, and when we do get systems they're often sent to different reviewers. Anand has played with a nettop at some point, so have I, and so has Ganesh. Most of those were a year ago or more, though, and so there's not a lot of overlap in the performance results. Reply
  • UrQuan3 - Monday, June 04, 2012 - link

    Wish we could arm-twist LogicSupply a bit for the other review units since they seem to stock AMD E-series and Via Nano systems as well as Atoms. Reply
  • Spunjji - Tuesday, May 29, 2012 - link

    For the curious:

    Pretty much what I'd imagined - Intel win for power consumption and thermals, AMD win for just about everything else. That 28nm shrink of Brazos needs to come sooner.
  • randinspace - Tuesday, May 29, 2012 - link

    I think at least part of the answer to your question lies in the fact that Logic Supply themselves seems strangely (tragically? misguidedly?) married to the Atom if Brazos' presence in only one of their fanless systems is any indication. Reply
  • Spunjji - Tuesday, May 29, 2012 - link

    A comparison to / mention of Brazos would have been nice, what with it being the prime competitor in this market! I'd really like to know how Intel's update changes the performance stakes, as Brazos was never that far ahead on the CPU side to begin with. I understand that you may not have had a system to hand but some passing comment would be lovely. Reply
  • silverblue - Tuesday, May 29, 2012 - link

    Brazos is a fine competitor, however it'd be an unfair comparison as you mentioned above because of its higher power consumption and temperatures. Brazos-T is on the way which should improve on that.

    Ironically for AMD, Brazos is much faster at single threaded workloads than Atom; it's only when Atom's HT is leveraged that the two get much closer.
  • KZ0 - Tuesday, May 29, 2012 - link

    Last page: "Atom wasn't integrated enough and small enough to actually make it into retail smartphones; no with Medfield it is, but at the same time that sort of design just isn't fast enough for Windows products."

    supposed to be " ... now, with Medfield, ... "?
  • Denithor - Tuesday, May 29, 2012 - link

    And another one, fourth paragraph on the last page:

    And that's the crus of the issue:

    The word should be 'crux' not 'crus' (which isn't even a word). Spell check should have caught this one...
  • JarredWalton - Tuesday, May 29, 2012 - link

    Unfortunately, this was some information I added to Dustin's review in our CMS, which doesn't really have spellcheck in place. Thanks for the catch; the offending typos have been fixed. Reply
  • eanazag - Tuesday, May 29, 2012 - link

    We should not be surprised that Intel cannot get a GPU to function correctly on a part that they make less money on than their main parts. When the last 3 generations of Intel IGPs have been additionally handicapped by their software. I have an Core i3 540, which processor side I am very happy with, but on the graphics side I feel like I should have went a different route for.

    In x86 this is where we are left. We can do the old desktop model and buy an Intel processor with a discrete graphics card and not compromise. Otherwise we are buying a compromise in either direction with Intel and AMD.

    I am still perturbed Intel did not license x86 to nVidia. I think things would look a lot different today if Intel didn't try to squeeze nVidia out of as much as they could. Think Ion and chipsets. I really think AMD would be doing better if they didn't do the same thing in chipsets to nVidia. I really don't think there would be serious competition to Apple in the tablet space if Intel would have granted an x86 license to nVidia. Would there be other ARM choices? Yes. But no one would be as successful as pushing a gaming platform as nVidia is with Tegra/Android ecosystem if Android was left to do it on their own.

    I really want to commend nVidia on staying relevant. I really thought their ship was going to sink when AMD and Intel kept finding ways to relegate nVidia to only discrete cards.
  • JarredWalton - Tuesday, May 29, 2012 - link

    I disagree on several points with your idea that NVIDIA would somehow make a better x86 CPU than AMD or Intel, and that they could also get all the chipset stuff right. NVIDIA chipsets ranged from decent with a few minor quirks to downright awful at times. I think NVIDIA pulled out of the AMD chipset business voluntarily when they decided they couldn't offer better quality for less money. On the Intel side, they could have continued to offer something, but I'm convinced we're better off with SLI on Intel chipsets rather than NVIDIA locking SLI use to their own chipsets. Reply
  • Dustin Sklavos - Wednesday, May 30, 2012 - link

    On that I actually disagree. It's true their chipset business was a little "eh," but nForce 2 was a stone cold killer back in the Socket A days. Reply
  • Lothsahn - Tuesday, May 29, 2012 - link

    I work with Logic Supply for the enterprise, and they're great. They have excellent communication, solid products (we've had little to no RMA's), and excellent support and customer service. When we did have an issue, they had a person in-house reproduce our issue and solve the problem within a day.

    Compared to working with other sellers (like Dell), it's a world of difference. They make solid products at slightly higher prices, but you get what you pay for.

    We've actually been using a very similar product for 2 years, based off the NF96FL mainboard. While our system isn't quite fanless, Logic supply recommended a specific set of fans that are high reliability, and we haven't had a single failure (>40 deployed systems).
  • bernstein - Tuesday, May 29, 2012 - link

    whats really interesting woud be a comparison with amds atom counterpart e-350, especially in those 3dmarks... Reply
  • aduncanvickery - Tuesday, May 29, 2012 - link

    I would be interested to see what the Linux performance and support is like for this system. I know a lot of retail outlets use slightly-customized small linux distros for their in-store browsing and internet catalog kiosks. Any thoughts? Reply
  • Lothsahn - Tuesday, May 29, 2012 - link

    The PowerVR driver is a mess in linux. Search for "poulsbo linux", and you'll see a lot of information about that. That means the graphics driver problem is likely a huge issue in this model.

    OTOH, we've been using Ubuntu 10.04 on the NF96FL mainboard (Atom D525) and we haven't had any stability issues at all. We've had ~40 systems deployed, and to my knowledge, haven't had a single crash.
  • zeo - Wednesday, May 30, 2012 - link

    Imagination never supported open source drivers for their GPUs, and 3rd party driver support tend to be messy to begin with... The Intel based GMA's though have typically provided Linux support.

    While Intel also supports initiatives like the Tizen OS.

    Fortunately, Intel will be going back to their own GMA with the next generation ATOMs.

    While Nvidia may be offering something with their Kepler series, as it's suppose to have versions that can go as far down as being included in Smart Phones. So a updated version of the Nvidia ION may not be out of the question.
  • randinspace - Tuesday, May 29, 2012 - link

    I was just looking at Logic Supply's website. I'm not surprised that they sent you guys a "new" platform for testing since they have no reason to do anything other than push their existing customers to upgrade, but I am surprised that they only offer a single AMD fanless solution when they're still offering Core 2 Duo ones. I can almost see a case for seemingly offering mobile C2D (I didn't double check but I was assuming these were the ones that had the onboard GPUs from nVidia) instead of Fusion in their Linux systems as well if only because Apple (indirectly) got away with it but... I simply couldn't find a product on their website that made sense to me from top to bottom, or which even more importantly actually made sense to buy from them instead of someone else. The msi products, for instance.

    Of course now that I've considered all that at least it finally makes sense why they approached you guys about doing a review that's not even aimed at enthusiasts let alone consumers in the first place.

    [My last sentence looks strange to me so just to clarify: I'm not trying to accuse you guys of anything aside from UNbiased reporting, I'm just saying that in the absence of compelling products Logic Supply looks like they could stand any kind of buzz related to their band. Since this review at least inspired me to check their website maybe that wasn't a bad idea on their part?]
  • ericgl21 - Wednesday, May 30, 2012 - link

    Latest graphics driver I found for Atom Nx00 is v8.14.8.1077, dated 2012/06/03.
    It is for the following devices:

    Download from here:
  • speculatrix - Wednesday, May 30, 2012 - link

    gma500 used the PowerVR SGX 535, which was a disaster; the driver was closed source and not great quality, developed by Tungsten Graphics

    Tungsten Graphics was bought by VMWare. Chances are there's noone left who helped develop the driver, and probably noone who has authority to open source it. I did try contacting vmware to ask about what happened but noone (wanted) responded.
  • name99 - Saturday, June 02, 2012 - link

    "Despite being the third generation of Atom processor from Intel, performance per core and per clock has essentially stood still since the very first Atom was introduced, and it continues to do so. Other than the single-core and dual-core models, Atom is about making a very small, inexpensive, low power x86 chip. The 32nm shrink that the N2800 represents is all about reducing power consumption further still, which is how we can get two x86-based cores with a combined TDP of just 6.5 watts.


    Unfortunately, there's a rub. ... Intel was having problems getting the GMA 3650 working properly in Windows. "

    All this, IMHO, simply reinforces my point all along regarding ARM vs Intel.
    Yes, Intel has astonishing logic and process engineers. But they cannot change the fact that the full x86 package (and that is what Intel has decided Atom will ship with) is insanely complex. It took 7 years to move Nehalem from idea to shipping product, and that's only going to get worse.

    Meanwhile ARM is much more agile, able to move from A8 to A9 to A15 micro-architecture during the time Intel has stood still, and ARM seems to be a lot more willing to let the past go. They've specifically stated that the ARM-64 instruction set is based on all the lessons they've learned over the past years about what works and what doesn't, and that basing it on the existing 32-bit instruction set was not a priority where that did not make sense. Of course the first few rounds of ARM-64 CPUs will carry 32-bit baggage, but I imagine that will be dropped as soon as those aggressive customers with substantial control over the code their devices run (*cough* Apple *cough*) are willing to do so.
  • powerarmour - Wednesday, June 06, 2012 - link

    There is also a rumour it seems that Intel won't have any GMA 3600/3650 drivers ready for Windows 8 either...

    Imagine buying a Cedar Trail tablet in the bizarre hope that you'll be able to run the latest and so called greatest on it from Redmond, and be faced with basically a buggy mess due to the current Windows 7 drivers (even the latest 1077 version will downright crash the system if you attempt to run a Metro apps)

    This is bad stuff, very very bad...

    Xbit labs did a review of their own also :-

    2D performance is simply dreadful and unacceptable, this is epic fail pure and simple. And they have the nerve to sell these to the public?

Log in

Don't have an account? Sign up now