Via intros QuadCore processor with guess how many cores

Early this year, Via introduced its Nano X2 processor, a dual-core implementation of its Isaiah architecture built on TSMC’s 40-nm chip fabrication process. Today, Via is announcing a new product, the QuadCore processor, that combines a pair of Nano X2 chips on a single package to deliver a low-cost, low-power CPU whose position in the market is fairly distinctive.

We visited Via-Centaur’s Austin, Texas offices yesterday, where we chatted with Centaur President and Founder Glenn Henry and Via VP of Marketing Richard Brown. We came away with some fresh details on the QuadCore processor and a better sense of Via’s future plans as an intriguing third-place supplier of x86-compatible PC processors.

Initially, Via is introducing a single model of the QuadCore, the L4700, rated for 1.2GHz operation and a maximum thermal and power envelope (TDP) of 27.5W. This CPU has a front-side bus speed of 1333MHz and a total of 4MB of L2 cache. However, as with other recent PC processors, those specifications only tell a part of the story.

New on the QuadCore’s feature list is something called Adaptive Overclocking, which isn’t really overclocking in the void-your-warranty sense we’ve come to know and love. Instead, Adaptive Overclocking enables the CPU to range dynamically up to frequencies as high as 1.46GHz, provided there’s sufficient thermal and power headroom available, much like Intel’s Turbo Boost. Via indicates this possibility by listing the QuadCore’s clock speed as “1.2+ GHz,” just to keep folks guessing. Much like Intel’s most recent incarnation of Turbo Boost in its Sandy Bridge chips, Adaptive Overclocking will allow the QuadCore push beyond the chip’s specified TDP peak for short periods, provided system temperatures will allow it.


The QuadCore ranges up to 1.46GHz under load

The QuadCore’s dynamic frequency scheme is complicated by the fact that it involves two separate chips, each with dual cores. Dynamic power and frequency management happens on the chips themselves, using power consumption estimates provided by the VRMs, in concert with the operating system’s P-state requests. The chips then coordinate, Henry told us, by talking over a “side channel” of wiring between them, built into the CPU’s package. Each chip can range up in frequency independently of the other, provided there’s headroom available between the two.

When the cores are idle—the vast majority of the time, in most systems—they will drop to lower frequencies and voltages, of course, and clock gating will kick in to save power. Via doesn’t yet have power gating and separate voltage islands like we’ve seen in some Intel and AMD processors, however. Henry hinted those are likely to come in Via’s next chip.

The rest of the inter-chip communication in a QuadCore processor happens via the front-side bus, just as it did in Intel’s quad-core, Penryn-based Yorkfield CPUs, for instance. Because the QuadCore uses Via’s “V4” front-side bus technology—which is very similar to Intel’s FSB—and the same pinouts as prior Via CPUs, this new processor is instantly compatible with Via’s existing chipset and motherboard infrastructure. Still, there’s work yet to be done. Although today is the CPU’s official unveiling and the building-block Nano X2 parts are already shipping, Via doesn’t expect the QuadCore to enter production until the third quarter of this year, because the quad-core part requires additional qualification.


The Nano X2 (left) and QuadCore (middle)

Once it arrives, Via expects the QuadCore to find its way into a range of products, from netbooks to low-cost desktops, small-form-factor PCs, and mini servers. However, we here in North America can probably expect to see the QuadCore arrive first in motherboards, probably of the Mini-ITX variety, aimed primarily at embedded systems and such. Outside of a few notable exceptions like the Samsung NC20, Via has had precious little success getting the Nano and its derivatives into consumer products here. The Nano has gained more traction in emerging markets like China and India—markets that are more price-sensitive and less brand-sensitive than ours. That may be where the QuadCore finds its footing, as well. Brown couldn’t yet point to any upcoming notebooks or low-cost desktops destined for these shores with a QuadCore CPU inside.

If the QuadCore can’t crack the consumer market here, that may be a shame, because it stands to occupy a spot in the x86 firmament not really served by any other CPU. The “Isaiah” core, you may recall, has a true out-of-order execution engine tuned to deliver decent per-clock performance at relatively low-power operation. In many ways, it may be most similar to the Bobcat core in AMD’s “Brazos” APU, though with somewhat higher per-clock performance. The major difference, of course, is that the QuadCore has four cores, while Brazos and Intel’s Atom platforms top out at two. With no higher-end CPU business to threaten, Via can offer a low-cost quad-core option with a promising performance story for a range of needs, from light-duty computing to multimedia playback and even some gaming.

The firm showed us selective results from its own internal benchmark bake-offs versus an AMD E-350 APU, and the QuadCore won by margins ranging from 5% in SysMark 2007 to 28% in CPUMark 99 and 100% in Cinebench R10. (The margin in Everest’s AES encryption bench was 1871% in favor of the QuadCore, thanks to Isaiah’s built-in encryption engine.) When paired up with a discrete GeForce graphics card, the QuadCore even took on one of the Internet’s most enduring memes by churning out acceptable frame rates in a live demo of Crysis 2.

With a TDP of 27.5W, the first QuadCore isn’t positioned exactly opposite Brazos (9-18W TDPs) or the netbook and desktop Atom variants (~13W) in terms of power consumption, but its power draw is still sufficiently low to fit into small notebooks (say, in the 12-13″ range), all-in-one PCs, and other small form factors. Brazos’ mark of 18W is really a more comfortable fit for netbooks and such, and Via plans to reach that TDP level with another, not-yet-announced variant of the QuadCore, likely running at 1GHz. That version will, obviously, be a more direct threat to AMD’s baby Fusion APU.

Total platform power consumption is really a more important issue than CPU TDP alone, and the QuadCore doesn’t have as many traditional chipset functions integrated into it—not a memory controller, graphics engine, or PCI Express—as the latest Atom or AMD’s APUs. However, Henry was quick to point out that all of that functionality is there in a Via-based system, potentially in just two chips like the competition—or, in the case of the QuadCore, one chip and one dual-chip package—thanks to Via’s single-chip core logic solution. That I/O chip’s contribution to total power draw is relatively small, at around 4W. Also, crucially, chipset silicon typically is built on older process nodes, said Henry, so it’s cheaper to manufacture. Locating graphics and the rest there keeps costs down, which is clearly a key to Via’s strategy.


Glenn Henry explains what’s where in an expanded version of the Nano die

Chipset support may be something of an issue for the QuadCore. Via has two basic options right now, as Brown explained. There’s a single-chip core logic offering, the VX900, with DirectX 9-class graphics and robust acceleration for H.264 video decoding, and there’s the VN1000 dual-chip set with DX10.1 graphics and only partial H.264 decode acceleration. Neither is an ideal choice, obviously. A better answer is in the works in the form of new chipset with DX11-class graphics, but it’s not ready yet. Furthermore, Adobe doesn’t take advantage of Via’s H.264 decode acceleration in its Flash video software, and that isn’t likely to change any time soon.

On the plus side, H.264 playback is something of a binary it-will-or-it-won’t type of performance question, and the QuadCore should have enough muscle to decode and play H.264 video streams without additional assistance, albeit at the likely expense of battery life versus dedicated hardware. Henry expressed confidence the Nano X2 would do fine, as well.

Our next steps, obviously, will have to include getting our hands on a QuadCore-equipped laptop or motherboard for some testing. We’re intrigued to see how it stacks up against Atom and Brazos, and how well it handles a range of common computing tasks.

Via’s next steps seem to be fairly evident, as well. Henry told us they’re working on their next chip now, which will involve a true refresh of the Isaiah architecture, with improvements for efficiency and performance. He let slip that this next chip will be manufactured on a newer process node, 28 nanometers, with “more than two cores on a single die,” but beyond that, he wasn’t ready to divulge details just yet. The logical thing would be to put four cores on a chip, perhaps with an integrated memory controller and maybe even graphics, but we’ll have to wait for official confirmation of those plans down the road.

Comments closed
    • Michael REMY
    • 8 years ago

    i don’t know whenever the screen cap of taskmanager is real or not but it shows with 4 cores at 1.4Ghz, cpu usage is already at 50% even with 4GB of memory !!!!!!

    it feels like this chip is really really less powerful than a single atom which do better with only 1 core !

    ps : maybe another software runs on the same time but we don’t know…

    • CBHvi7t
    • 9 years ago

    “rated for 1.2GHz operation … This CPU has a front-side bus speed of 1333MHz”
    The FSB ist clocked higher than the core?!
    I find that odd.
    I hope to see this in netbooks and laptops.

      • tfp
      • 8 years ago

      To me what is interesting based on the picture is that the cpu clock speed is 5.5 x 266 mhz or 1.46 so the base clock is 266. The old intel and I believe via by extension, FSB was always 4 x clk. To me the picture shows one of a few things.

      1) VIA multi for FSB is 5 x times clk to reach 1333mhz. This is kind of odd because FSB was always “quad pumped” so I don’t think this is reasonable…
      2) Eng sample and they are only running the FSB at 4 x 266 or 1066mhz at this time.
      3) There are 2 separate clk generators one for the CPU core and one for the FSB.

        • CBHvi7t
        • 8 years ago

        Na, the FSB clock is 333MHz. You can’t have anything else than 1,2,4.
        the CPU has it’s own clock-domain.
        likely the core+L1+L2 are m *266MHz with m={3.5, 4, 4.5, 5, 5.5, 6}
        The oddity of the high FSB-clock is due to the marketing simply giving the transfer-rate, not the actual clock. That is like saying it is a 3 issue CPU so the rate is 3.6GHz.

          • tfp
          • 8 years ago

          So you think they have 2 different clock generators on the chip

            • CBHvi7t
            • 8 years ago

            They could generate the clocks from one external 66.6MHz oscillator.
            One PLL would be on each die and one in the northbridge.
            why not? of cause just speculation.

    • FuturePastNow
    • 9 years ago

    I think this would be great for a general-purpose home server.

      • swaaye
      • 9 years ago

      I think leftovers would be better because they are free. 😉

        • FuturePastNow
        • 9 years ago

        Maybe, but they don’t satisfy my need for new toys 🙂

    • Buzzard44
    • 9 years ago

    4! Yay, I was right!

    • clone
    • 9 years ago

    I’ve never had any reservations regarding buying a Via processor, the bigger problem has always been finding one to buy.

    they come up with some interesting product that must be selling somewhere but I’ve rarely noticed it in the market.

    • jensend
    • 9 years ago

    I don’t know how engineers at VIA/S3/Centaur avoid getting fed up with life, much less how the company stays solvent. It seems like for quite a while they’ve come up with intriguing work and then their ideas never really make it to market. Somewhere along the road from decent-performance engineering samples to mass availability and market impact these things just seem to disappear.

    Nano was supposed to be the Atom killer [i<]three years[/i<] ago. How many people here have actually seen any Nano products in person? How about DX10 or greater S3 graphics cards (Chrome 4xx,5xx)? (Did these even really see mass production?) Whatever happened to the folks who developed the Envy24? That used to be a Big Deal- it powered all kinds of good audio interfaces from "higher-end" integrated audio to entry-level professional cards like the M-Audio Delta series. But I don't think VIA has made any significant advances here in about a decade. I think the only possible explanation for VIA managing to stay solvent is that they have a machine that turns the tears of frustrated engineers into pure gold.

    • maxxcool
    • 9 years ago

    I’d like to see a quad core bobcat take this thing on at the same frequency, and then compare it to a dual core c50….

    • BiffStroganoffsky
    • 9 years ago

    The 1.2GHz + rating brings back memories of Cyrix/Via days. Images on Nova of the Japanese nuclear reactor building exploding from the Hydrogen buildup after the cooling system failure reminds me why I don’t buy Cyrix/Via processors any longer. Thankfully, I was able to shutdown my boxes before there was significant build up and detonation but the cat still looks funny without its hair.

    • Mr Bill
    • 9 years ago

    I want a wall in my room covered with a blown up CPU! That would be awesome.

      • burntham77
      • 9 years ago

      I know! I saw that and now I want one for my office.

    • swaaye
    • 9 years ago

    So who is using Nano CPUs? They come up on news sites periodically but they don’t seem to end up in any consumer products other than ITX boards? Are they used for industrial stuff?

      • mnecaise
      • 9 years ago

      They’re in a lot of thin clients and x86 embedded controllers. Most thin clients that I’ve seen are Via chips. Think call centers and office environments.

    • ronch
    • 9 years ago

    I believe VIA released their new line of Eden X2 dual core processors not too long ago, which are just two Isaiah cores on a single die. Surprised to see how quickly they put together an MCM and have a quad core. Not exactly the hardest thing in the world, but at least they’re moving forward.

    I hope that their next generation product after this will be very competitive. Even reaching Core 2-like performance would give them a big boost in terms of mind share, if not market share, and that ought to earn them more respect.

    • sweatshopking
    • 9 years ago

    you know, I would like to put in a request to my review coworkers: if we could include the Windows Scores for gpu’s, cpu’s, SSD’s, etc. I know that some of us are huge epeen fans, and I have constant competitions to make sure that my scores are higher than my brothers. Just because it upsets them. Knowing what score it’ll give will provide little in terms of REAL numbers, but it would provide a bit of peen, and might be hand for the guy trying to build the 7.9 system.

    • mutarasector
    • 9 years ago

    “With a TDP of 27.5W, the first QuadCore isn’t positioned exactly opposite Brazos (9-18W TDPs) or the netbook and desktop Atom variants (~13W) in terms of power consumption”

    This pretty much sums it up, doesn’t it?

      • NeelyCam
      • 9 years ago

      I would say this is perfect for those who were left wanting more performance from Atom/Brazos. This will kill Zacates in nettops (Atom has been nettop roadkill for a long time already – maybe it shouldn’t even be mentioned here).

      It can’t beat low-power SB variants, obviously, but for a cheapo nettop this is perfect.

        • Anonymous Coward
        • 9 years ago

        [quote<]I would say this is perfect for those who were left wanting more performance from Atom/Brazos.[/quote<] Eh, I thought Intel had that covered with a flavor of Sandy Bridge.

          • NeelyCam
          • 9 years ago

          Surely SB is faster, but it’s in a different price class… I was thinking cheap nettops.

        • raddude9
        • 9 years ago

        I doubt not going to beat Zacate at all.

        Even in VIA’s own cherry-picked benchmarks from:
        [url<]http://hardocp.com/article/2011/05/11/via_quadcore_preview_centaur_tour[/url<] it's frequently only 10% faster than the dual-core 1.6Ghz Zacate. Some independent benchmarking will surely show up many more circumstances where having 4 weak core is a disadvangate compared to 2 more powerful cores. In balance, I think Zacate will win.

          • Rza79
          • 9 years ago

          [url<]http://images.anandtech.com/graphs/intelpinetrail_122009194423/30338.png[/url<] The Nano X2 is already faster than Zacate in real life applications. I don't see a 'QuadCore' with 300-400Mhz lower frequency change that. All it does is bring us back to that age old question of cores vs frequency. But from the looks of modern software, it seems that four core are the way to go these days.

            • raddude9
            • 9 years ago

            Try 600Mhz slower and not 300-400! The Nano in your graph is the top-of-the-range 1.8Ghz Nano, comparing the 1.2Ghz quad core against the 1.6Ghz Zacate and I see all of the single-threaded tasks going the Zacate’s way.

            Best case, in an ideal scenario, the 1.2Ghz Quad is only going to beat the 1.8Ghz dual by 33%. So unless you want to do some 3d rendering or video transcoding, the faster dual core will be the way to go.

            • Rza79
            • 9 years ago

            “Adaptive Overclocking enables the CPU to range dynamically up to frequencies as high as 1.46GHz”

          • NeelyCam
          • 9 years ago

          [quote<]Even in VIA's own cherry-picked benchmarks from: [url<]http://hardocp.com/article/2011/05/11/via_quadcore_preview_centaur_tour[/url<] it's frequently only 10% faster than the dual-core 1.6Ghz Zacate. [/quote<] Did you even look at those graphs? In most cases, wherever the word "CPU" is included, it's 50-200% faster. Also, VIA has a turbo to help with single-threaded loads. In balance, I think VIA will win hands down.

            • raddude9
            • 9 years ago

            Huh, 50% to 200%, what are you talking about?

            Even, as if you say, we cherry-pick the benchmarks that mention “CPU” (out of the benchnmarks that VIA picked) we get:

            CPUMark 99 v1.0 Score = 42% faster
            CPUMark 2.1 Final Score = 68% faster
            3DMark 2006 v1.20 (1280×1024) CPU Score = 57% faster
            Internet Explorer 8.0.7600.16385 CPU score = 15% faster
            —————————————————————————-
            Average of those four = 42%

            If we average those 4 together we get an average of 42% faster. Great you might say, not exactly the 50%-200% you were talking about, but pretty cool yea? But that’s is the result of cherry-picking the best results out of the already cherry-picked results that VIA fed us…. so not exactly real-world results.
            I for one will be waiting until we’ve seen a few independent reviews before putting any money down! And being VIA I won’t be holding my breath for those results.

            p.s. my phrase for the day is “cherry picking” 😉

        • mutarasector
        • 9 years ago

        “I would say this is perfect for those who were left wanting more performance from Atom/Brazos. This will kill Zacates in nettops (Atom has been nettop roadkill for a long time already – maybe it shouldn’t even be mentioned here).”

        It might kill Zacate in embedded applications _now_, but not by much, and is comparing apples/oranges. 1> This is a rather new processor compared to Zacate, 2> and is comparing a 4 core proc to a dual core Zacate, 3> nor is it’s TDP competitive (the metric that really counts in embedded applications). Furthermore, VIA’s offering is FSB based – a dead end, and will not outperform Bobcat dual cores on the next die shrink, or quad cores when Krishna & Wichita arrive.

    • flip-mode
    • 9 years ago

    Most importantly, can it play Crysis?

      • Meadows
      • 9 years ago

      You didn’t read the article. Shame on you, I expected more.

        • Scrotos
        • 9 years ago

        Technically, his question is still unanswered since they didn’t demo it on Crysis, they demoed it on Crysis 2.

        • flip-mode
        • 9 years ago

        I am disappoint.

          • Silus
          • 9 years ago

          I always liked that response. It really makes me LOL 🙂

            • Meadows
            • 9 years ago

            Show minimal appreciation for internet memes, or else a nearby krogoth will catch the scent of it and ram us all with garbage.

            • Silus
            • 9 years ago

            And I did! I really like it 🙂

    • Silus
    • 9 years ago

    Scott,

    It would be interesting to see an actual review for these processors. There’s always so much debate on competition and everything around those lines, yet review sites only pay attention to products from the “bigger” companies. And this is also true for most of the comments in here. People whine about how more competition is always better, yet they are ignoring VIA completely and part of that blame lies on sites such as TR, that barely do anything to change that.

    More information only makes TR better.

      • Meadows
      • 9 years ago

      This processor is not out yet.

        • Silus
        • 9 years ago

        I know that….I was talking about the future…
        Just look at TR’s CPU reviews. There’s one for the Nano and that’s it. How can there be any knowledge about other processors (other than Intel and AMD), when there’s barely any review with them in it, while there is a review for every iteration of Core i3, i5, i7, Phenom II 980, 990 and so on ?

        The latest reviews for i7 and Phenom IIs are completely pointless, the only saving grace for the i7 review being the terrific text Scott wrote. Apart from that, they are absolutely worthless, because they have nothing new in them.

      • flip-mode
      • 9 years ago

      From the article: [quote<]Our next steps, obviously, will have to include getting our hands on a QuadCore-equipped laptop or motherboard for some testing. We're intrigued to see how it stacks up against Atom and Brazos, and how well it handles a range of common computing tasks.[/quote<] What now?

        • Silus
        • 9 years ago

        I’m talking about this:

        [url<]http://www.anandtech.com/show/4017/vias-dual-core-nano-vn1000-chipset-previewed[/url<] Included in the same review as this: [url<]https://techreport.com/articles.x/20401[/url<] It's what reviews are made for: comparisons. If the smallest player in the x86 market doesn't even have some sort of presence in these reviews, how is it that anyone will even know they exist ?

          • Scrotos
          • 9 years ago

          I suppose the other half of it is that Via doesn’t send TR many review samples and thus they don’t include them on low-end roundups. Perhaps Anand is getting more stuff sent. You’d find that odd since TR also had a nice sit-down with Via in their offices so I’d think Via would be sending out review samples often and trying to milk that good relationship.

            • Silus
            • 9 years ago

            Yeah that was my reasoning as well. Via has every interest in being nice with reviewers. Why wouldn’t they want attention ?

          • Damage
          • 9 years ago

          Besides saying that we plan to review them when we can, I don’t know what more we can do here. My time machine broke after the whole OBL fiasco–seriously, lost my stealth copter there, too, bummer–so I can’t go back and put a Nano X2 in the E-350 review.

            • Silus
            • 9 years ago

            Come on Scott, I clearly mentioned future 🙂
            What’s done is done, even though it would’ve been great to see those 3 in the same review. Anandtech’s review shows potential against PineTrail, which is what you used in the E-350 review. My point with those links, is that the product was available at the time of your E-350 review, so why not for TR ?

            It’s perfectly understandable that you can’t review everything and as Scrotos said, Via might not be playing nice with TR, but if that’s the case it’s odd, since they have everything to gain in providing review samples to you. Don’t know if you want (or can) clarify that.

            In graphics card reviews, even when not having the product currently at hand, you have used data from previous reviews, in the current one as a measure of comparison. That’s great to see and something very few do, which makes me thumbs up TR, since it provides a meaningful comparison between generations. But if you don’t review a certain product, you can’t even do that.

            • Damage
            • 9 years ago

            Wait, this is about the future? What?

            You’ve got a serious case of temporal confusion.

            So, to summarize:

            -We said in the article we’d like to review the QuadCore, and we’re working to make that happen.

            -We didn’t review the Nano X2 alongside the E-350, so that makes you worried about the future-past time continuum wormfold thingy. I get it. (Lying.)

            The simple truth is that it just wasn’t big on our radar because we couldn’t find a Nano X2 for sale anywhere (there’s still nothing on Newegg or Amazon right now, AFAICT), Via never offered a review sample, and we were very busy with other things. What more do you want me to say? We have to pick and choose what to review, and the Nano X2 was a ghost. It didn’t make the cut.

            If you’re worried about not having a Nano X2 to compare in the future, well, perhaps we can arrange for one still. You know, in those GPU reviews, we don’t just dust off numbers from old reviews. We dust off old cards and test them again alongside the new ones with the latest games and drivers. I’m not making an iron-clad commitment to do that with the Nano X2, but it’s certainly possible.

            • Silus
            • 9 years ago

            For the E-350 I just wanted to know (if possible) of why you didn’t use the Nano X2. You cleared that up, so thanks! It’s unfortunate that Via didn’t provide one. Seems that they don’t really want the “free” publicity…

            As for the future, you mentioned that you would review it and that was more than enough. My point was if from that point on, we would see more Via stuff alongside the other products being reviewed.

            And speaking of graphics cards (since you’ve been eager – or maybe not 🙂 – to answer my questions and thanks for that), I asked you a awhile ago, if you were going to pursue that power consumption methodology of different loads, different power measurements seen in the initial GTX 580 review. Did you scrap that, or are there still plans to shift your graphics cards testing methods (for power at least) with this ?

            • DancinJack
            • 9 years ago

            LOOK HERE MATE.

            [url<]http://www.anandtech.com/show/4332/vias-quadcore-nano-gets-bigger[/url<]

            • Silus
            • 9 years ago

            Thanks 🙂
            It’s looking good for VIA. Hopefully they can reduce their power consumption numbers a bit more and it will be even better.

            • mutarasector
            • 9 years ago

            Scott,

            I’d almost prefer to wait and see what a refreshed Isaiah quad core (as hinted at in the article) would look like and would stack up against Krishna/Wichita. Assuming the Isaiah goes true quad core on a single die w/an IMC, the comparison would be more meaningful. Even overlooking it not being a SoC, as it stands right now, a Nano and Zacate comparison is simply meaningless. [EDIT} Maybe a year from now this will change?

    • mako
    • 9 years ago

    I’m not personally interested in this product, but it’s good to see that VIA is still putting out products.

    • NIKOLAS
    • 9 years ago

    Seriously, who in the first world would bother?

      • Meadows
      • 9 years ago

      It does depend. Heat output aside, this is a processor, so (theoretically) VIA can’t screw up on the driver front – and if the price is low enough for the performance, people might just take a look at it. Even in your hometown, what your superiority complex calls the “first world”.

        • srg86
        • 9 years ago

        Trouble is that it’s pretty much tied to VIA’s chipset (as no one else will make a chipset for it) so VIA can still screw up the drivers. Still it should be fine with discreet graphics, but rarely used, so there’s the VIA graphics, with crap drivers.

          • Meadows
          • 9 years ago

          [i<]Dicreet[/i<] graphics, you say? What, are the graphics so bad too that they need to hide it?

            • destroy.all.monsters
            • 9 years ago

            You’re missing an “s” there Meadows. No Vanna White action for you.

            • Meadows
            • 9 years ago

            I can’t imagine where I’d put another ‘s’.

            • NeelyCam
            • 9 years ago

            “dicreet” doesn’t have an s.

            • Meadows
            • 9 years ago

            I can hardly believe I missed it. Thanks! I was looking all over at each and every word *except* the first.

            • NeelyCam
            • 9 years ago

            Gotta love it when a grammar nazi screws up.

            • Meadows
            • 9 years ago

            I screw up with dignity.

            • derFunkenstein
            • 9 years ago

            You owned up to it and didn’t have a hissy. That’s all we can ask.

            • Meadows
            • 9 years ago

            Who do you think I am, Krogoth or something?

            • srg86
            • 9 years ago

            oops Typo alert

            Still, your point still stands. I do have a Nano based machine and the Graphics are the main trouble spot. I really wanted to run it in 64-bit mode, but getting the Graphics to work in Linux (32 or 64) is a total pain, usually ending in total lockups. In Windows I had trouble getting it to boot because the versions of XP64 and Vista I was using had never heard of “CentuarHauls” and BSODed. Windows 7 booted, but locked up because of the crap graphics.

            It’s now running on 32-bit XP and running like a champ, a shame but at least I still have it working.

      • stdRaichu
      • 9 years ago

      It might be been nice to see something like these available for embedded NAS units. More grunt* than the ARM’s, PPC’s and Atoms that are currently prevalent but still a low power envelope. Most NAS units end up being CPU-limited because they use softraid, and everyone always picks RAID5 or 6, which eat a lot of CPU cycles doing parity calculations.

      * Last review I saw for the nano was ages ago, but I recall it being appreciably faster than the atom and didn’t eat a whole lot more power. And the atom hasn’t really gone through any major performance bumps since then.

        • swaaye
        • 9 years ago

        The single core Nano used more power and was slower than a Pentium M “Dothan”. It was faster than Atom but surely uses more than twice the power than one of the Atoms with the integrated IGP.

        • Anonymous Coward
        • 9 years ago

        I really doubt that most NAS units end up being CPU limited. If they are, someone is not putting the CPU to good use.

          • swaaye
          • 9 years ago

          I think they will because you need a good bit of CPU power just for the ethernet transfer let alone stuff like Samba overhead. My router’s NAS function can only manage about 12MB/s with its 400 MHz MIPS 24k CPU and it’s pegged when doing that.

          On PC even with some TCP offload you need something like a 1.5 GHz CPU to get to the point where the HDD can’t keep up with the transfer on a gigabit link.

          But I don’t know much about NAS options. Are there some with fancier hardware taking over tasks from the wimpy CPUs in these things?

          • cygnus1
          • 9 years ago

          Possibly not on a 100mbit, but on a 1gbit link they most certainly do end up CPU limited. As swaaye pointed out, the XOR parity calculations reading and writing RAID 5 and 6 are very CPU intensive. That’s why expensive NAS units have a dedicated RAID engine whereas cheaper (slower) ones use soft RAID.

          I don’t know enough about the math going on in the parity calculations, but I’d be curious if softraid drivers could off load calculations using CUDA, OpenCL or DirectCompute with one of the more modern integrated GPUs.

      • indeego
      • 9 years ago

      Africa makes a strong patriotic statement with a -10 vote on this post.

    • eitje
    • 9 years ago

    I’m really proud of Centaur – keep up the good work, guys!

    • grantmeaname
    • 9 years ago

    I really hoped they wouldn’t in the Nano years, but VIA got left behind hard. 1.46GHz Quad core is 27.5 watts? I have a 2.8GHz dual core for that, from three years ago.

      • eitje
      • 9 years ago

      I agree with you strictly from a product perspective, but considering the miniscule size of the VIA CPU team they’re actually doing quite well. 🙂

        • Bensam123
        • 9 years ago

        Yeah, I agree… They definitely are quite the underdog both in their graphics line up (which is inferior to even Intels) and CPUs themselves. I at one time recommended VIA chipsets to people waaaay back when, sucks to see someone that has fallen so far.

      • OneArmedScissor
      • 9 years ago

      TDP doesn’t tell you much about a low power CPU beyond whether or not it should go in a 9″ netbook.

      That’s like saying, “I had a 3.8 GHz single core for that 6 years ago!” I think we all know how that went.

        • grantmeaname
        • 9 years ago

        For 27.5 watts? No, you certainly did not.

          • OneArmedScissor
          • 9 years ago

          *WHOOOOOOOSH*

          • Damage
          • 9 years ago

          Yeah, folks seem to be missing this. TDP is peak power draw, not typical, AND the QuadCore’s TDP is way lower than the usual desktop power envelopes of 65W-125W. The fact that it fits into Via’s existing infrastructure and is being mentioned for small notebooks and such should be a big clue. This mix of potential performance and power draw is somewhat unusual, as I said in the article.

            • PrecambrianRabbit
            • 9 years ago

            Scott, do you have any insight into a better power/energy metric than TDP? Something I’ve been advocating is to use two numbers – task energy and idle power. I like energy for the task, because I don’t actually care if I use twice the power if the task gets done twice as quickly. But I like power for idle time, because that doesn’t depend on the performance of the processor.

            • derFunkenstein
            • 9 years ago

            You still need TDP when you’re determining cooling design. You can, in theory, run a CPU at max for infinity+1 hours consecutive and you need to keep it (relatively) cool.

            • PrecambrianRabbit
            • 9 years ago

            Sure, but I’m not saying take it off the data sheets. It’s an important design-time constraint, but it tells you little about how the part will behave in the consumers hands. (Other than perhaps not frying said consumer’s hands.)

            • swaaye
            • 9 years ago

            there are Core i3 and i5s that are in that power class and I’m gonna guess that things would go poorly for Nano there. However Nano would undoubtedly be a lot cheaper to compete.

      • mutarasector
      • 9 years ago

      Wasn’t part of the reason VIA got left behind hard not just because 3rd party chipset vendors became irrelevant, but because of the patent-infringement issues between the two (VIA/Centaur and Intel) that resulted in some cross-license agreement to settle it? IIRC, VIA only has what, 2-3 years left on their current 10yr. x86 license from Intel, no?

    • DeadOfKnight
    • 9 years ago

    I still don’t know anything about VIA or their target consumer base.

      • stmok
      • 9 years ago

      Don’t waste your time or money with VIA.

      * Not as good value compared to what their competitors offer.
      * After-sales support sucks balls. (Especially their industrial embedded solutions compared to Intel or AMD.)
      * Driver and application support (to their hardware acceleration features) is most often a joke in Windows and typically non-existent in Linux.

      AMD’s 2nd generation Bobcat-based APU, codenamed “Krishna” will be quad-core under 28nm. It comes in 1st half of 2012. You are more likely to see this come out on time compared to VIA’s QuadCore.

      VIA makes money when they sell their solutions to no-name OEMs in China/India. And its often their cheap ARM-based VIA 8505 processor. (Not their x86 lines.) …You can buy such netbook/tablet solutions here in Australia for about $130 to $170. They are usually equipped with either Windows CE 6.x or Android 2.x

      VIA should just switch over and focus on ARM. (Where MS said they’ll do an ARM-port of Windows 8)…Glen Henry and his team is being wasted on the x86 market.

        • DeadOfKnight
        • 9 years ago

        How can they possibly turn a profit in these times then?

        • NeelyCam
        • 9 years ago

        [quote<]VIA should just switch over and focus on ARM. (Where MS said they'll do an ARM-port of Windows 8)...Glen Henry and his team is being wasted on the x86 market.[/quote<] No. VIA should hold on to the one key advantage they have over others - the x86 license.

          • Corrado
          • 9 years ago

          Does VIA still have a bus license from Intel though? Making x86 CPUs is all well and good, but when you have to engineer the entire platform, it becomes less enticing to stick with x86. Especially if your market is primarily low power and embedded systems that could just as easily run on an ARM system.

            • axeman
            • 9 years ago

            I thought Intel pretty much made sure no one can license the bus, at least not at a price anyone will go for.

            • NeelyCam
            • 9 years ago

            VIA doesn’t need a QPI/DMI license from Intel. VIA could use Hypertransport between the CPU and the chipset if they wanted to.

            NVidia was in a different situation, since they didn’t have the license to build the x86 CPUs themselves. Their chipsets needed to interface with whatever bus the Intel/AMD CPU had; they didn’t have the QPI/DMI license, and AMD didn’t have the volumes to make it worth NVidia’s while, so – poof – NVidia was out of the mobo business.

          • mutarasector
          • 9 years ago

          I’m partially with stmok on this one. Although I’m not a big advocate of the idea of mixing ISAs or developing a hybrid ARM/x86 processor, I think VIA could be an exception in this case. Given that one of VIA’s strengths is AES encryption and their VIA Padlock/Advanced Cryptography Engine, an integrated ARM core might just make sense as the processor engine to offload this feature from the x86 side, and allow VIA to take Nano something more towards their CoreFusion processor. This would play to their strength in embedded, POS, and KIOSK applications that simply don’t require DirectX-11 capability, but solid 2D performance, and _security_, such as military grade hardware running QNX RTOS applications. It just doesn’t make sense for VIA to chase after Intel/AMD and ARM for consumer space however, as the competition is simply too entrenched.

    • UberGerbil
    • 9 years ago

    Is this question part of some new contest? Because if the prize is something from Via….

      • MadManOriginal
      • 9 years ago

      Hopefully this possible contest is only open to entrants from India and China because then I can have my cake and eat it too – complain about the contest not being available to everyone while also secretly being glad because I wouldn’t actually want to win.

        • paulWTAMU
        • 9 years ago

        The only way to win would be not to play?

Pin It on Pinterest

Share This