Sandy Bridge CPU and graphics performance tested

Although the first retail processors based on Intel’s new Sandy Bridge architecture aren’t due until the first quarter of next year, AnandTech surprisingly managed to score an early sample for testing. The CPU in question is a Core i5-2400 that has four cores, 6MB of L3 cache, and a 3.1GHz clock speed. Turbo Boost wasn’t enabled in engineering sample, but it’ll reportedly scale the final product up to 3.4GHz when thermals permit.

Even without Turbo Boost, the i5-2400 fared well in the battery of benchmarks AnandTech used to test the chip. Intel’s new hotness averaged out to 23% faster than a Core i5-760—the CPU it may ultimately replace at the same price point. Power consumption looks good, too. Sandy Bridge doesn’t appear to have an advantage at idle, but under load, it drew 10W less than the i5-760.

Perhaps most impressive is the performance of the new CPU’s "Gen 6" integrated graphics core. In low-resolution tests with in-game details at their lowest settings, the Sandy Bridge graphics processor kept pace with a discrete Radeon HD 5450. That’s not the most impressive GPU in AMD’s lineup, but it’s a whole lot faster than the Intel HD Graphics option available in current CPUs. AnandTech didn’t observe any image quality issues with Intel’s early graphics drivers, which is encouraging news, as well.

Comments closed
    • d0g_p00p
    • 9 years ago

    I think it’s funny that people think they are entitled to overclock their CPU’s

      • A_Pickle
      • 9 years ago

      Yes, bastards trying to get the most out of the technology they paid for!

    • maroon1
    • 9 years ago

    i5 2400 (without turbo) performs about 23% better than i5 760

    It aslo performs on par ~ with i7 880, and better overall than Phenom II X6 1090T

    The final product should perform even better because it is going to support turbo.

    Based on the roadmaps and pricing in one of the anadtech pages, the i5 2400 should price same as i5 650 (which is cheaper than i5 760)

    So, who really dares to say that sandy-bridge is a minor update ?

    • swaaye
    • 9 years ago

    For IGPs, I thought that even GMA 950 was good enough for almost everything an IGP would be used for. It does Aero pretty well. That’s about all the 3D hardware needs to do for about 95% of the people who will use it. Once H.264/VC-1 hardware acceleration showed up, it seemed to me that we had IGPs that were about perfect.

    Go game on a 5450 and get some perspective on how “exciting” that performance level is. I would rather dig out an old Radeon X800 and game on that because it would dust a 5450 in any game that still lets you use a SM2 card.

    I doubt Intel has much interest in taking this tech anywhere interesting, such as a discrete card that could be competitive with the midrange or high-end. That would be interesting though.

    • Trymor
    • 9 years ago

    This chip sounds like it could be a bad thing in a way, at least for certain system builders. They may not have as many buyers popping for that discrete video card, and have to replace that revenue from somewhere else. Perhaps the base model prices will have to rise?

    Try

      • Chrispy_
      • 9 years ago

      Why do you always sign your posts?

    • srg86
    • 9 years ago

    reply fail…

    • NeelyCam
    • 9 years ago

    You know, everybody’s so focused on the IGP performance. To me the fact that Intel’s offering a 32nm quad-core for main stream is more important.

    Clarkdale GPU was perfectly adequate for 2D/HTPC purposes, and if I were to build a gaming rig, it would most definitely have a discrete card at the level of 5850 or above… I wouldn’t be happy with IGP gaming for at least another two years or so.

    Meanwhile, to me it matters far more to have CPU horsepower to do all sorts of HTPC encoding at low power levels (for low noise).

    • MadManOriginal
    • 9 years ago

    btw all this IGP talk cracks me up. Everyone loved to complain about Intel’s IGPs and now that there will be pretty significant improvements in gaming performance up to lower-end discrete levels, and that’s with just a 6 EU/SP part, there are i[

      • NeelyCam
      • 9 years ago

      Fanbois are scared that the reasons behind their choices are getting invalidated… they have to find something wrong with the enemy. OpenCL is such an important feature for a HTPC/low-grade gaming rig.

        • yuhong
        • 9 years ago

        Well, there is the claim that it is important for Apple, but that is probably just a rumor.

          • NeelyCam
          • 9 years ago

          Apple is, what, 10% of the market?

      • srg86
      • 9 years ago

      I think it’s the current fashion at the moment. They’ll move onto something else eventually…

    • ronch
    • 9 years ago

    Are you reading this article, AMD? Hello…

      • ClickClick5
      • 9 years ago

      Oh they are. And they are smiling knowing what they have in the labs.

        • yuhong
        • 9 years ago

        Do you think they will respond?

        • Phr3dly
        • 9 years ago

        Everyone’s got something great ‘in the lab’. A processor in retail is worth 10 in the lab.

    • kroker
    • 9 years ago

    Uhm… Waiter, there’s an IGP in my processor!

    Seriously, what is the advantage of the IGP being intregrated on the processor instead of the chipset? Especially since it doesn’t support OpenCL/DirectCompute anyway? And even if not obsolete from the start, the IGP will become so long before the CPU will. It also adds more heat to the CPU.

    Will there be versions of Sandy Bridge without an IGP for the 1155 socket?

      • OneArmedScissor
      • 9 years ago

      It has a direct connection and can share resources, so it’s faster and uses less power.

      There is no way in hell a tiny, low power GPU is going to cause heat issues for 32nm CPUs.

      Once it’s completely universal to have graphics built into the CPU, they can be used to switch off graphics cards when they’re not in use.

      Why would you want one without it? It’s not like they’re going to charge you less to build a few completely differently.

        • kroker
        • 9 years ago

        l[

          • NeelyCam
          • 9 years ago

          IGP/chipset being done on an older process is a cost-cutting measure; generally, the performance of the latest-gen process isn’t needed, and keeping the previous-gen fabs running and producing extends their life and improves profitability.

          But it seems integrating everything to the CPU die/package may turn out to be the most cost-effective option for the final product. Every computer needs a GPU of some sort, and putting it into the CPU die/package makes perfect sense. PCIe is already integrated, SATA and Ethernet/WiFi are probably next…

          And the GPU would most certainly benefit from the best silicon process; lower power, higher performance, more functionality in the same area… There is a reason why discrete GPUs are always riding the cutting edge of foundry process technology, even if the yields aren’t quite there yet.

          I’m just guessing, but I think OAS’s point was that when IGPs are integrated on every CPU, some sort of a pass-through protocol/spec will get defined to allow disabling pretty much 99% of the discrete GPU while allowing the IGP to drive the display through the DVI/HDMI/DisplayPort connector on the discrete graphics card. Or, alternatively, have the IGP drive the display at all times, and use the discrete GPU just to take care of the computational heavy-lifting.

            • Voldenuit
            • 9 years ago

            There is already a separate platform that has IGP pass-through (H55/57) and one that doesn’t (P55).

            If you read Anand’s article, he explicitly states that the same breakdown will apply for Sandy Bridge (H67 vs P67), right down to the crippled nature of the IGP-capable H67 chipset – no SLI/CFX and locked memory multipliers. This is a market segmentation trick, not a hardware limitation.

            Also, GPU transcoding is generally faster /[

            • MadManOriginal
            • 9 years ago

            Don’t forget that GPU transcoding is still lower quality, at least as far as I know. Sure, some people won’t care or say it doesn’t matter for portable use but to me transcoding multiple times for different targets isn’t so smart either. Not to mention that if GPU transcoding could be as high quality as CPU transcoding the speed advantage wouldn’t be nearly as large so comparing speed and power use is only 2/3 of the ‘speed-power-quality’ trifecta – btw, you may choose 2 of those things 😉

            Also you seem to misunderstand what people are calling ‘IGP passthrough.’ Basically we want technology like NV Optimus on the desktop. Maybe I’m mistaken but I’ve never read or heard of anything like that on a desktop chipset.

            • Voldenuit
            • 9 years ago

            They were rough at the start (especially ATI), but are a lot better now. I’d say they are on par with most commonly used CPU-only apps.

            You can always make a better encode with avisynth and mencoder if you have the skills (and time) to hand-tune every stream you get, but for drag and drop “dirty encoding” for temporary residence on an iAppliance, GPU encoding is as good as mainstream tools and much, much faster.

            And on the high end/prosumer segment, we have people like Adobe Premiere Pro CS5 supporting GPU acceleration of nonlinear editing and encoding with full quality controls. That’s big, but also out of most Average Joe’s software budgets.

            PS: wrt ‘Desktop Optimus’, I believe that the biggest obstacle is the sheer number of hardware configurations possible (compared to fixed laptop platforms). ATI tried that many years ago with Hybrid Power, and they couldn’t even get it to work reliably /[

    • NeelyCam
    • 9 years ago

    l[<'Dozer versus sandy bridge will hopefully be an interesting fight. But since you're going to need to upgrade from AM3 to AM3+ to use bulldozer the advantage is stacked in Intel's favor.<]l It's funny how it's become expected that Intel changes sockets every time, but if AMD changes sockets, it's somehow considered a disadvantage.. (I.e., SB requires a new socket too)

      • chuckula
      • 9 years ago

      l[

        • OneArmedScissor
        • 9 years ago

        Not very many people change out CPUs in the same motherboard. If the CPU is cheaper, the computer is cheaper, as far as most sales go.

        The way new CPUs are being made should also lower the cost of motherboards a little, and they can be had for $50 or so as it is. That’s not exactly a deal breaker,

    • Rza79
    • 9 years ago

    The SB IGP is as fast as the HD5450 (a 80 shader part).
    If the SB IGP in the sample is the 6 EU part and it didn’t Turbo, then we can expect something that’s around 3x faster in the final top of the line SB. That’s an if ofcourse.
    Llano’s IGP is rumoured to be a 400 shader part. So SB will still not cut it against AMD in the IGP departement.

      • NeelyCam
      • 9 years ago

      Rumor != reality.

        • Manabu
        • 9 years ago

        Bobcat is confirmed to be an 80 shader part (low clocked of course), so Llano being 400 shadder seems plausible, especially based on the dieshot we have.

      • OneArmedScissor
      • 9 years ago

      Your 3x faster figure based on the “dual-core” GPU and higher clock speeds is a bit unrealistic. It’s incredibly unlikely there’s enough bandwidth to go around, which is what has always hurt low end graphics, whether it be an IGP or card.

      If AMD really do start piling on hundreds of SPs to higher end “Fusion” chips, they’ll probably use sideport memory with a dedicated HT link as they have been doing, but with GDDR5. It doesn’t look like Intel has anything like that planned.

        • Anonymous Coward
        • 9 years ago

        I can’t see AMD doing anything with IGP that would require a new socket, so that sideport memory better be on-package.

        Or maybe its time to go to slot processors again!

      • Kurotetsu
      • 9 years ago

      400/5 = 80?

      So dead even with Sandy Bridge’s IGP without compiler support (and ‘without’ seems to be much more likely than ‘with’)? Or is AMD not using 5-way shaders with Llano?

        • MadManOriginal
        • 9 years ago

        They may be but I think you’re not doing the right comparison, or maybe it’s a hard comparison to make. The ‘divide by 5’ rule applies to NV versus ATi shader comparisons. So a ‘400/5=80’ shader part would be pretty good and nearing lower midrange parts like the 4650/70 and 5650/70.

        Intel’s IGPs, while sufficient for simple or older gaming (the latter of which is what interests me…I’m currently on a stock clocked i3-540 and using the IGP even at 1920×1200 don’t have a problem with most 5+ year old non-FPS or low req games,) don’t get directly comparded to higher shader parts very often if at all so I don’t know how to compare shader counts there. The Clarkdale IGP has 12 ‘shaders’ and is not that far behind NV’s 16-shader DX10 IGP or AMD’s 80-shader one. If Intel continues to do rather obvious graphics performance improvements that have been around forever (z culling) like they did with Clarkdale I will probably be perfectly happy with my limited gaming time to use the SB IGP and couldn’t care less how it runs the latest FPS or action game at 1024×768.

    • bogbox
    • 9 years ago

    It’s not fast ,compering a 2-3 years old graphics with the newest and greatest and at *[

      • raddude9
      • 9 years ago

      The fact that all of the benchmarks were taken at the lowest resolution and detail level is a bit suspicious to me, at least give us a look at what happens when you crank things up a notch, does the performance fall over completely?
      I have a feeling that all of these CPU-integrated-GPUs (AMD’s forthcoming chips included) will have serious problems with gaming at anything over lowest resolution/detail, the memory bandwidth just isn’t going to be there.

      • Voldenuit
      • 9 years ago

      More importantly, Fusion is OpenCL and Directcompute capable, so even if the CPU is not the fastest, applications such as media encoding, transcoding, rendering etc (you know, the things that take the longest time on a CPU) will fly compared to any CPU.

      If Fusion becomes ubiquitous, expect to see the IGP/APU being used for physics, pathfinding, etc in the near future, even with a discrete GPU installed (compared to SB, where the IGP is completely useless if you’re not using it as the main display).

      Intel is being very backwards-looking by releasing a dumbed down IGP for Sandy Bridge. Even the 2-year old R700 (Radeon 4250) has beta OpenCL support. Worse, it will hurt software development and support. Mac users can count on 100% of Macs sold within the past 2 years to have OpenCL acceleration, but PC users will have a fractured install base, which is going to slow down adoption considerably.

    • wira020
    • 9 years ago

    At LOW RESOLUTION and LOWEST SETTING… I dont know what people are getting excited about… if it cant at least max out graphic at that resolution, I dont think it’s much of an improvement at all…

    • esterhasz
    • 9 years ago

    wow, that IS impressive – so now almost every computer will be able to do somewhat decent graphics. but is there going to a software side innovation that will actually translate the HW potential into a bigger market for PC gaming? why is there no Microsoft managed market application for games that uses the built-in benchmark and monitor resolution to automatically adjust settings for good framerate? people actually pay money to game with a virtual joystick on their phone! granted, there is steam but to create a real mass market there has to be a big “get games” icon on the desktop when you plug in the machine for the first time.

    games like torchlight show that you can build great PC gaming that runs on a net book so just think about what you can do in two or three years when current 5450 power will have a 500 million PC install base (based on 300M PCs sold per year)!

    whatever your stance on distribution through closed markets, 5450 level graphics is way better than the WII and there is a huge market opportunity here. if Microsoft squanders is – and they probably will – I really hope that Valve can pick up the pieces…

      • Voldenuit
      • 9 years ago

      Don’t forget that all the IGP tests at Anand were run at the lowest grafx settings and 1024×768. In other words, the games would all look like sh!t.

      In a best case scenario, SB should challenge nv and ATI to step up the quality of their low end discrete and integrated parts, but it’s not a discrete GPU replacement by a long run.

      And since the IGP is not GPGPU/OpenCL/CUDA/Directcompute-capable*, that means half the silicon is completely useless if you’re running discrete grafx.

      *wtf, in this day and age.

        • esterhasz
        • 9 years ago

        sure, the horsepower is nowhere near current best of class games but the market I’m thinking of would be halfway between the casual iPhone/Wii/flash game market and classic PC gaming. Again, torchlight is a good example gameplay wise and if people pay 7 bucks to play classic doom on their phones couldn’t there be a market for fun shooters à la borderlands with torchlight graphics (which are IMO totally awesome when you look at what they need to run)? The PC as a platform for casual gaming would also be an opportunity for the more demanding classic games…

    • Voldenuit
    • 9 years ago

    Locking BCLK (ie killing overclocking) may be enough for me to switch teams. I don’t even OC my CPUs these days, but I can’t say I’m too thrilled by intel continuing to restrict its users’ options.

    If they price the K line competitively, it may change my tune, but I don’t have my hopes up.

    I also notice that there is no mention in the AT article about whether the IGP is OpenCL/DirectCompute capabe, unless I missed something.

      • stmok
      • 9 years ago

      Intel’s IGP does NOT do OpenCL or DirectCompute.

        • Voldenuit
        • 9 years ago

        Ouch. At least they’re consistent about being dinosaurs :(.

      • Krogoth
      • 9 years ago

      Why does any of it matter?

      Sandy Bridge was designed first and foremost as an OEM product (Average Joe). The IGP is quite adequate for this market. The users who would need OpenCL/DirectCompute will most likely opt for a discrete solution of some kind.

        • Meadows
        • 9 years ago

        “OEM market” is not a valid excuse for putting crap on the table.

          • Krogoth
          • 9 years ago

          Transistor budget and TDP puts a severe limitation what engineers can do with an integrated GPU on a CPU package.

          Implementing hardware-level OpenCL and Directcompute support costs too much for Intel to consider. It isn’t like AMD’s implementation is going to be any better.

            • Game_boy
            • 9 years ago

            Llano won’t have OpenCL or Directcompute?

            • Krogoth
            • 9 years ago

            Their OpenCL/Directcompute performance will likely be not good enough for gamers to consider. Making it more of a marketing bullet point if anything else.

            • Game_boy
            • 9 years ago

            480 shaders should be enough to accelerate some consumer apps at least.

    • Ryhadar
    • 9 years ago

    Llano’s GPU better be /[http://www.xbitlabs.com/news/cpu/display/20100826225852_Desktop_Bulldozer_Processors_Will_Require_New_Platforms_AMD.html<]§ ) The only clear win for AMD seems to be with bobcat. However, I just wanted to say that the very first thing I thought when I saw SB's GPU performance was: HOLY CRAP! Maybe now PC gaming will get taken seriously again.

      • shank15217
      • 9 years ago

      last I checked SB needed a whole new socket and a new chipset and it’s not socket compatible with older cpus.

    • pedro
    • 9 years ago

    reply fail

    • TaBoVilla
    • 9 years ago

    nice timing on this one Intel, you’re starting to sound like nvidia

      • NeelyCam
      • 9 years ago

      “Our bumps are not your bumps”

    • jdaven
    • 9 years ago

    So the 3.1 GHz Sandy Bridge is 23% faster than a 2.8 GHz Nehalem. So roughly 10-15% faster clock for clock as Anand says in the teaser intro on the main page. Also, you get 10% less power. Not bad. However, that means that Bulldozer only has to slightly perform better than Nehalem to match Sandy Bridge clock-for-clock and live up to the 10-100W power envelope.

    I change my vote on the weekly poll to match Sandy Bridge performance.

      • Althernai
      • 9 years ago

      Be careful when comparing clock-for-clock. In the comments to the article, there was this exchange:

      ——————————
      “I take it turbo was also disabled on the rest of the parts used to compare, right?”

      “Turbo was enabled on everything else – SB performance should be higher for final parts.

      Take care,
      Anand”
      ——————————

      The Nehalems are tricky: given half-way decent cooling (which I’m sure AnandTech has), they will Turbo even when all 4 cores are active. Thus, the supposedly 2.8 GHz Nehalem actually runs at a minimum of 2.93 GHz and goes up to 3.33 GHz in single or dual threaded tasks. A lot depends on what Turbo scheme Intel has chosen for Sandy Bridge.

        • jdaven
        • 9 years ago

        Very good point! I did not think about that.

        • shank15217
        • 9 years ago

        Bulldozer also brings these dynamic frequency scaling features into AMDs arsenal.

          • NeelyCam
          • 9 years ago

          Great. AMD can’t innovate, so they have to copy Intel’s brilliant ideas such as CPU Turbo and power gating…? Are they gonna copy Intel’s GPU turbo, too?

            • sweatshopking
            • 9 years ago

            can’t innovate? I think you’d better check the history books bub.

            • Meadows
            • 9 years ago

            Truer words have not been spoken here.

            • NeelyCam
            • 9 years ago

            Your sense of sarcasm is lacking today. People (mostly AMD fanbois) are always whining how AMD is so great and innovative and Intel is just copying them. I was pointing out that it’s a two-way street – this time it’s AMD that’s doing all the copying.

            • MadManOriginal
            • 9 years ago

            It’s a classic nerd-fanboi flaw. It’s like MS-Apple or more recently MS-Google. All we should really care about as users is whether things are getting better and not worry so much about who did what first.

            • Voldenuit
            • 9 years ago

            Yeah, also, IBM did most things first: multicore, SMT, SOI, 64-bit, superscalar, microops, OOE.

            I was going to add IMC as well, but it seems the DEC Alpha (whose designers went on to the K8) was the first there.

            • TaBoVilla
            • 9 years ago

            Next up on Intel’s list for future plagiarisms I suspect could be copying IBM’s method for four way SMT, similar to what it does on POWER7, to counter AMD’s innovation on doubling integer pipelines on each core for multithreaded performance, before AMD copies Intel’s scheduler to enable hyperthreading, in time for Intel to redesign branch prediction following AMD upping caches to avoid misses, etc, etc its a cycle…

      • OneArmedScissor
      • 9 years ago

      I have to question the notion that it’s 10% lower power. It’s running roughly the same clock speed that the i5 760 would “turbo” up to, and only 10w less in what is undoubtedly a synthetic stress test.

      That’s less than motherboard vs. motherboard territory. It’s probably no difference, whatsoever, in actual use.

        • NeelyCam
        • 9 years ago

        Agreed. But you can’t expect huge power savings without 1) a massiive architecture overhaul, 2) new process technology, or 3) heavily cherry-picked/binned silicon.

        And I’m talking about active power. I know you like to point out mobile systems are (and rightfully so), but that’s a result of tweaked and binned silicon that runs at low supply voltages but is unable to reach desktop-level clocks.

        … I’m now realizing that I’m stating the obvious, eh?

      • swaaye
      • 9 years ago

      On the topic of performance claims by AMD, refresh your memory here. 😀
      §[<http://www.youtube.com/watch?v=G_n3wvsfq4Y<]§ Sure it probably does that in some case, but we know how things looked overall in the end.

    • HisDivineShadow
    • 9 years ago

    If Intel, nVidia, and AMD could work out an interoperability standard for Optimus-like GPU switching (so that every dedicated GPU could work with every CPU/GPU APU) and see the state of each other, these onboard GPU’s could be very useful to everyone’s computers.

    Better still if said onboard GPU’s could be used for dedicated physics processing while leaving 3d to the dedicated GPU.

    Unfortunately, given that all the parties involved hate one another for various reasons, it seems incredibly unlikely. Perhaps Microsoft will create some kind of standard for a new DirectX/Windows technology to help facilitate this kind of feature.

      • Stargazer
      • 9 years ago

      q[

      • NeelyCam
      • 9 years ago

      Corporations don’t have feelings – they don’t “hate”. They make/don’t_make money.

        • JustAnEngineer
        • 9 years ago

        Jen-Hsun Huang can hate. He hates with the white-hot intensity of a thousand suns.

          • NeelyCam
          • 9 years ago

          Yeah, JHH is a bit of an enigma.. He built a pretty powerful company from ground up, and isn’t a crazy marketing/business guy with no sense of engineering (he was a hands-on design engineer back in the day).

          I refuse to believe that he’s dumb and oblivious to the reality in this business.I keep wondering if his “hate” is a rallying cry for morale, a PR trick or what… Then again, maybe he just lost his mind. It happens.

          • esterhasz
          • 9 years ago

          ah, come on, he’s playing the passionate CEO with a vision to impress the shareholders… that’s probably why he’s still around, despite the numbers – just as the Jobs has well understood that controversiality is good for business – every apple hater is feeding the feeling of individuality of their user base…

        • Umbragen
        • 9 years ago

        Hey, corporations are people too! The US Supreme Court says so.

    • marvelous
    • 9 years ago

    So what’s the deal with overclocking on the sandy bridge? Anand doesn’t seem to have a definite answer.

      • OneArmedScissor
      • 9 years ago

      It will just depend on which model you buy and how much Intel wants to bend people over for what they’re turning into a paid priviledge.

      • Krogoth
      • 9 years ago

      The clock generator is on the CPU, unlike previous schemes where it was on the motherboard.

      This is a problem for QPI/FSB overclocking. Any change to the FSB/QPI will affect the clock speed of your memory, PCIe lanes, PCI bus etc. PCIe and PCI devices hate operating beyond spec speed.

      The only hope is that some motherboard manufacturers might be able to circumvent this by forcing the CPU to work with external clock generators (like it always has).

      Otherwise, you are stuck with getting factory unlocked “K and Extreme Edition” series and have to do multiplier-level OC’ing.

      This is Intel’s way of getting a slice of the overclocking market, while killing off budget overclocking at the same time.

      • stmok
      • 9 years ago

      l[

    • OneArmedScissor
    • 9 years ago

    I like that it’s running 3.1 GHz at about 1.0v. Hopefully the trend of voltage for desktop CPUs creeping upward is over and they’ll finally fall more in line with their laptop counterparts.

    But other than that…meh.

      • NeelyCam
      • 9 years ago

      Undervolting FTW!

    • flip-mode
    • 9 years ago

    I’m impressed with the graphics performance. It’s not that I would want to play games with that. I would not like that at all, but it is an impressive improvement, still.

      • Krogoth
      • 9 years ago

      It is no surprise. Intel is getting more serious with their graphics, because Sandy Bridge is their answer to AMD’s Fusion.

      • ltcommander.data
      • 9 years ago

      What’s more, what was shown was only the 6 EU version, probably without Turbo Boost. It’ll be interesting to see what a 12 EU version can do. If it can keep up with a HD 5570 that would definitely satisfy many low end gamers and will no doubt be very popular with OEMs not having to integrate a seperate GPU anymore.

      • jdaven
      • 9 years ago

      I agree. The integrated GPU is quite impressive. If the parts are not running in turbo (1350 MHz) and are only 6 EU, then the full shipping product might match the performance of a radeon 5670 (12 EU, 1350 MHz). Quite impressive. I will never knock Intel IGP again.

        • DaveJB
        • 9 years ago

        Until it turns out that the graphics core of Fusion performs like, say, a Radeon 5830. Then “Intel graphics sux” will become the norm once again! :p

        • indeego
        • 9 years ago

        Sure you will. Their IGP drivers and support are horrid crap, and they only support them for 1-2 years, at best. Q35 chipset? Horrible support, and crashes on Windows 7 (and Vista) regularly. Intel has abandonedg{<.<}g

          • bdwilcox
          • 9 years ago

          Saying Intel’s new graphics are decent compared to their old graphics is like saying something tastes decent compared to dog-poo.

            • NeelyCam
            • 9 years ago

            Looks like this new IGP is about 2x better than 790GX/890GX…

            • flip-mode
            • 9 years ago

            Yep. I’d call it bare minimum performance for gaming. I don’t think any other integrated solution qualifies for gaming, really.

        • NeelyCam
        • 9 years ago

        This is great news. I was concerned that Llano would whack Sandy and AMD would hike the prices, but with this IGP Intel dodged a bullet, and we get to continue enjoying exceedingly cheap CPUs on both sides.

        Everybody wins. Except AMD.

          • shank15217
          • 9 years ago

          What crap, you don’t even know llano’s graphics performance.

        • MadManOriginal
        • 9 years ago

        Don’t you need to wait until His Appleness Jobs deems Intel’s graphics ‘good’ in order to decide? ;D

          • NeelyCam
          • 9 years ago

          Ah, I forgot about Apple.

          Could this mean that Apple won’t switch to AMD after all…? You know, considering that SandyBridge is here in Q4, backed with Intel’s reliability, while Llano in serious volume is dependent on GF’s ability to get their 32nm process act together?

          Methinks SB just scored.

            • Voldenuit
            • 9 years ago

            No OpenCL = no deal for Apple.

            Edit: Not to mention that Apple just refreshed their product line a couple months ago, so they’re in no rush right now.

            • Game_boy
            • 9 years ago

            When they do though, Llano/Bobcat is a possibility. They can’t keep Core 2+Ion for much longer, and they can’t fit a discrete card into everything.

            • Anonymous Coward
            • 9 years ago

            I don’t know so much about Intel vs AMD politics at Apple, but seems like Bobcat would work real nice in the mini and maybe the lowest end of the macbooks.

    • Krogoth
    • 9 years ago

    Sandy Bridge = Lynnfield replacement.

    Enough said

      • 5150
      • 9 years ago

      Yawn, call me when processors are less than $10 / core.

        • Krogoth
        • 9 years ago

        BUZZ!

        Wrong answer.

        Please try again. 😉

        • sweatshopking
        • 9 years ago

        I lol’ed at this one. well played

      • MadManOriginal
      • 9 years ago

      Within the product lineup, yes it is. In just about any other measurable way it’s not. I thought that Nehalem and Lynnfield were not worthy upgrades over a solidly overclocked Core 2 system unless you need >4 (or >2) threads for truly multithreaded uses. A 23% improvement over Lynnfield with only a 10% clock bump i[

Pin It on Pinterest

Share This