Comments Locked

98 Comments

Back to Article

  • tynopik - Saturday, December 3, 2011 - link

    Without further due -> further ado
  • Kristian Vättö - Saturday, December 3, 2011 - link

    Fixed, thanks!
  • gwolfman - Saturday, December 3, 2011 - link

    Btw, another typo, but not you're fault. The x's for the CPU features. I'm assuming an x means the feature is present, but then why do the i7 and i5 'K' variants lack AES-NI and VT-d (for example)? That doesn't make sense. I can understand not including vPro, but AES-NI!?! That's gotta be wrong.
  • ericloewe - Saturday, December 3, 2011 - link

    If I recall correctly, they did the same thing with SNB. They're probably afraid of overclocked "low-end" parts cutting into their higher-margin stuff, in this case SNB-E or IVB-E-
  • Kristian Vättö - Sunday, December 4, 2011 - link

    The K-series chips have AES-NI, and it's in the table too :-) VT-d is missing, and that's correct.
  • kishorshack - Saturday, December 3, 2011 - link

    IGP is something that im looking forward to.
    Lets see how powerful these are
    N it will decide fate of several graphic cards present in market
    I hope it should be as powerful as GT 555m, but well might not be
    Another interesting fact is quick sync are we going to see support for more codecs
    n i still expect a huge leap over the previous gen :)
    At first i thought it was anand who wrote it
    But kudos to kristian
    He brought such important information to our notice :)
  • LtGoonRush - Saturday, December 3, 2011 - link

    It's just too bad you only get HD 4000 on the high-end and unlocked chips, the places where you're least likely to use it. It makes a lot more sense on the low-power processors where you're more likely to want to run without a videocard.
  • therealnickdanger - Saturday, December 3, 2011 - link

    It seems counterintuitive from a consumer prospective. The weakest (least expensive) CPUs should have the better IGP. If Intel follows suit with SNB, there will be low end SKUs with the HD4000. The i3-2105 comes to mind.
  • bupkus - Saturday, December 3, 2011 - link

    The i3-2105 comes to mind

    Unfortunately, I bought the 2100 before the 2105 was released.
    So much for early adopters.
  • futurepastnow - Sunday, December 4, 2011 - link

    I think it's probably safe to assume there will be at least one i3 with HD4000.
  • retrospooty - Monday, December 5, 2011 - link

    "It seems counterintuitive from a consumer prospective. The weakest (least expensive) CPUs should have the better IGP"

    Not reallly... People paying for the high end systems arent very licekly to be satisfied with internal graphics. It would just go unused in most systems.
  • retrospooty - Monday, December 5, 2011 - link

    n/m... Early morning word based dyslexia. LOL On to coffee.
  • JarredWalton - Saturday, December 3, 2011 - link

    Whoa there, tiger! HD 4000 should be a nice improvement relative to the HD 3000, but to even mention the GT 555M in the same sentence is getting out into fantasy land. The successor to the HD 4000 might actually get close to that level, but Intel is saying HD 4000 will be up to 60% faster than the HD 3000. Here's a comparison of Intel's quad-core HD 3000 against a similar laptop with a GT 555M:
    http://www.anandtech.com/bench/Product/327?vs=417

    To summarize (just focusing on Medium detail 768p results), the GT 555M ends up being anywhere from 100% to 257% faster than the HD 3000. On average across ten different titles the 555M is 170% faster. This isn't even getting into the question of rendering quality, where the HD 3000 in some games seems to render at lower quality or with a few glitches. If HD 4000 does end up being 60% faster on average, the GT 555M would still be another 70% faster than the HD 4000. If you'd like the specific numbers from bench, here's how the GT 555M compares to the HD 3000:

    BFBC2 Med: 147% faster
    Civ5 Med: 201% faster
    DiRT2 Med: 187% faster
    L4D2 Med: 106% faster
    Mafia2 Med: 197% faster
    ME2 Med: 113% faster
    Metro Med: 143% faster
    CoP Med: 156% faster
    SC2 Med: 257% faster
    TWS2 Med: 197% faster

    Anyway, I'm hopeful that HD 4000 will be a nice step up and will push AMD and NVIDIA into improving their entry level GPUs. If GT 555M becomes the new entry level part (that would make sense as the GT 525M/HD 6630M are roughly at the level where HD 4000 is supposed to launch), you could at least plan on running most games at medium detail and still get >30 FPS. I'm not sure how much further Intel can take their GPU though -- HD 4000 will probably be around 1/3 the total die area; will Intel really use half of the die for Haswell's IGP? I doubt it.
  • Matias - Saturday, December 3, 2011 - link

    Actually acording to this leaked slide, 3dMark Vantage Performance Preset for graphics is almost trippled, so it kind of does match 555M according to your link. :-)
  • Matias - Saturday, December 3, 2011 - link

    Slide:
    http://imup.se/i/JQo6oNv5Wf.jpg
  • KaarlisK - Saturday, December 3, 2011 - link

    That's vs the HD 2000.
    The comparison with the GT 555M is with a HD 3000.
  • iwod - Sunday, December 4, 2011 - link

    I was about to point that out. HD 3000 is double the speed of HD2000, so if HD4000 is 300% faster then HD2000,this is about 50% faster then HD3000.

    And we are talking about a Benchmark here, other test according to the slide show we should only expect on average 20-30% improvement.

    One thing that is not mention is the Graphics Quality, Since the new HD4000 natively support Direct X 11, we should see graphics quality improvement.
  • Death666Angel - Sunday, December 4, 2011 - link

    "One thing that is not mention is the Graphics Quality, Since the new HD4000 natively support Direct X 11, we should see graphics quality improvement."

    How much of a role does that play though, if you can only have the better image quality if you sacrifice FPS, thus playability? :-)
  • jesh462 - Sunday, December 4, 2011 - link

    DX11 supports performance enhancing features, not just higher image quality. Some games out right now use only DX11 for performance enhancment.
  • JarredWalton - Monday, December 5, 2011 - link

    The only game I've seen so far where DX11 improved performance over DX9/DX10 modes is DiRT 2, and that was only at Ultra Low to Medium settings. If you use High or Ultra High, DX11 results in a performance hit. I'm curious though: do you have other games where DX11 specifically improves performance? (I suppose DiRT 3 and F1 2010/2011 might also have the same pattern as DiRT 2, so other than those?)
  • mechBgon - Saturday, December 3, 2011 - link

    Good writeup, thanks Kristian. I'll get an early start on rationalizing the i7-3770 :)
  • chaos215bar2 - Saturday, December 3, 2011 - link

    Can anyone explain why the versions with unlocked multipliers are missing most of the instruction set additions? The most logical explanation I can think of is that Intel is worried someone will clock their CPU high enough that all commonly used instructions work fine, but something from the additions might not, but that's pure speculation. It just doesn't make much sense to me that the instructions would be disabled in the CPU and not the BIOS. Shouldn't it be up to the user to make sure they clock their CPU such that everything works, especially on what's supposed to be the top of the line (aside from Sandy Bridge E).
  • chrone - Saturday, December 3, 2011 - link

    yeah it's too bad the unlocked version didn't get the full feature. i hope they do include vt-d though. :/
  • KaarlisK - Saturday, December 3, 2011 - link

    IMHO, it's simple.
    If you're someone using VT-d, Intel does not want you overclocking your CPU to achieve your required performance, Intel wants you to buy two CPUs.
  • darckhart - Saturday, December 3, 2011 - link

    if that is indeed intel's rationale, then it's very unfortunate. i'm not so concerned with missing out on the others, but denying VT-d on K chips will lose my sale.
  • name99 - Saturday, December 3, 2011 - link

    Really. So tell us what you'll do instead.
    Buy an AMD chip? --- good luck with that.
    Stick with an SNB that's slower, just to spite Intel? Kinda defeats the point of wanting an unlocked chip in the first place, doesn't it.

    At the end of the day, it is Intel's ability to charge enterprise a lot which allows us to get incredibly powerful chips at reasonably low prices. Ranting about the "unfairness" of it all is dim-witted --- the unfairness pretty much all goes in the direction of enterprise being screwed, not you and me.
    And if Intel occasionally has to distort the product line to make this work, well that's the breaks. There ARE dumb things Intel has done that are worth complaining about; this does not strike me as one of them.
  • Kalessian - Sunday, December 4, 2011 - link

    Maybe you replied to the wrong guy, but darckhart didn't rant or say it was unfair. He said without the VT-d feature he wouldn't buy it. What's wrong with that? If someone doesn't like a product, they are allowed to voice why. How else will the company know what customers want?

    You didn't have to explain how tiered pricing works, either, everyone knows this is intel taking advantage of the lack of x86 competition.

    And VT-d is important to some people, so some of us ARE getting screwed out of something we used to have.
  • Kristian Vättö - Saturday, December 3, 2011 - link

    It might be a differentiation thing. If you want to overclock, you can't have the extensions. Or you can pay more for SNB-E which has VT-d (though not TXT or others).
  • DanNeely - Saturday, December 3, 2011 - link

    Aside from VT-D for people playing around with lots of VMs I don't think any of the other features have any prosumer use at present.
  • MartinT - Saturday, December 3, 2011 - link

    "The TDP is down from 90W to 77W, which is a 26% decline."

    Actually, that's a 14.444% decline.
  • Kristian Vättö - Saturday, December 3, 2011 - link

    Fixed, thanks!
  • arbitmax - Saturday, December 3, 2011 - link

    Actually the TDP of SNB is 95W. So, that is a decline of 18W ~ 18.94 %
  • Kristian Vättö - Saturday, December 3, 2011 - link

    True, LOL. Silly me, I have to start double-checking these numbers from now on, this is embarrassing!
  • MrSpadge - Saturday, December 3, 2011 - link

    In FF8 the 2nd table is too large.

    MrS
  • Kristian Vättö - Saturday, December 3, 2011 - link

    It was definitely a tight fit, I had to edit it several times to make it fit. Unfortunately, I can't really do anything because switching the layout would would still make it too big, and I would prefer not to take any info away.

    It's fine on Chrome for me though
  • IlllI - Saturday, December 3, 2011 - link

    i dont know what any of these letters mean:

    VT-d
    TXT
    AES-NI
    SIPP
    vPro

    or why some chips dont have them, yet some do
  • Kristian Vättö - Saturday, December 3, 2011 - link

    No offense but if you don't know what they mean, that usually means you don't have to worry about them :-) They are mainly instruction set additions, which are helpful in some tasks. E.g. VT-d enables direct I/O access for virtual machine, which is useful if you do a lot (heavy) virtualization.

    In general, I would love to explain all the abbreviations in every article but in case there are many, it might add too much clutter to the article (could easily take paragraph/each to explain them in detail).

    Would you be interested if we did a page that had all the abbreviations we use, and we could update it when new ones come? I ask because people ask for the abbreviation explanations so many times, and Wiki is sometimes too technical to get the gist of things. I can suggest this to Anand if you feel like it would be helpful :-)
  • Ushio01 - Saturday, December 3, 2011 - link

    That would be awesome, do it.

    Thanks in advance.

    (I'll forget otherwise cause of tonights drinking)
  • Death666Angel - Saturday, December 3, 2011 - link

    Some websites have a glossary which is very handy for this sort of stuff. Sometimes they also have mouse-over explanations of the words in the glossary when they show up in normal text. :-)
  • ravisurdhar - Saturday, December 3, 2011 - link

    +1 for mouse-over pop ups with definitions. :)
  • Matias - Saturday, December 3, 2011 - link

    +1! Good idea.
  • Paedric - Saturday, December 3, 2011 - link

    It would be a pretty good idea, IMHO.
  • IlllI - Saturday, December 3, 2011 - link

    some kind of glossary or something would be useful :)
  • Etern205 - Wednesday, December 7, 2011 - link

    From what I know (may be incorrect)

    VT-d= something to do with VM, but uses direct hardware?
    TXT= Trusted Execution Technology: something to do with buffer overflow attacks
    AES-NI= Programs that uses it, shows great performance improvment
    SIPP= Stable Image Platform Program. I guess this is something to do with long term supply availability?
    Like Asus SIPP, motherboards will have 1-2 year stable supply
    VPro= Intel's marketing to business market with features like remote management.
    Just like Intel Centrino and Intel VIIV.
  • semo - Saturday, December 3, 2011 - link

    Why do only the high end chips come with 2MB of L3 cache per core?

    I thought Intel's engineers insisted that 2MB is the sweet spot in terms of L3 cache. Or is that a moot point when it comes to shared L3 cache (where presumably one core can use 2MB or more if the other cores are not loaded)?
  • Kristian Vättö - Saturday, December 3, 2011 - link

    Intel wants you to pay more for more cache ;-) It's the same situation as in Sandy Bridge - i7 has 8MB, i5 has 6MB and i3 has 3MB.
  • retrospooty - Saturday, December 3, 2011 - link

    " Ivy Bridge has clearly been designed with power efficiency in mind"

    I woyuld say, its more like they are so far in front of AMD, they can push it that way . If AMD had an even remotely competitive product you would see IB at alot higher speeds. True with SB too. As it is most SB overclock to 4-4.5ghz on air. I bet these do 5ghz easily on air.
  • Kristian Vättö - Saturday, December 3, 2011 - link

    Definitely. This is a point that I will be bringing up in the "Why IVB is still quad core" article, which is in the works as we speak :-) However, I'm sure Intel is afraid of ARM and its efficiency, as it's not only AMD that they compete with nowadays.
  • Ushio01 - Saturday, December 3, 2011 - link

    I thought Intel announced all consumer chips would max out at 4c/8t at least to Broadwell with only the GPU seeing an increase in size. Since if you need more cores there's the enthusiast cut down xeon chips.
  • Kristian Vättö - Saturday, December 3, 2011 - link

    Do you have a link? I would really appreciate it, would be very helpful in the article :-)
  • Taft12 - Saturday, December 3, 2011 - link

    Uhhh, you can buy a 6-core / 12-thread consumer i7 chip TODAY (just need a socket 2011 motherboard to go with it)
  • Kristian Vättö - Saturday, December 3, 2011 - link

    The LGA 2011 platform isn't exactly meant for consumers. First, there isn't a single mobile LGA 2011 CPU, it's all desktop. Consumers buy mainly laptops. When we look at the actual die, it's a harvested SNB EP die, meant for servers.

    I think using the word mainstream (like Intel does) is better than "consumer", because the chips are available for consumers too after all. Most of them just don't buy them.
  • euler007 - Saturday, December 3, 2011 - link

    Buy your logic, any high-end product is not meant for consumers.

    The LGA 2011 platform is aimed at the "enthusiast" consumer willing to spend more money on their computer.

    The Porsche 911 is aimed at enthusiast car owners willing to spend more money on a performance car.

    The guy buying the Porsche is just as much a consumer as the guy buying a Fiat 500.
  • Taft12 - Saturday, December 3, 2011 - link

    Sorry Kristian, your articles are fantastic, but if it says Core on the package, it's a consumer product no matter how unreasonable the price.
  • Kristian Vättö - Sunday, December 4, 2011 - link

    That's why I suggested the use of word mainstream instead. You are right, any product with the brand Core in it is a consumer product. However, I think the original point was that why is there no mainstream (~$300) 6-core CPU.
  • DanNeely - Sunday, December 4, 2011 - link

    Most office laptops and desktops using intel CPUs have core series parts. Mainstream; not consumer is the proper word.
  • DanNeely - Saturday, December 3, 2011 - link

    I thinking DDR3 bandwidth bottlenecks are still a major factor as well. If DDR3-1333 would have bottlenecked Intels last generation hex core parts, it's reasonable to assume that the faster SB/IB CPUs would still be bottlenecked on DDR3-1600, and that Haswell will be bottlenecked on DDR3-2000 when in launches next year.

    Adding more channels drives up mobo costs; and for laptops volume also becomes a problem so mainstream 3/4 channel designs aren't commercially feasable. Broadwell's launch coresponds to when the current rumormongering has DDR4 becoming widely available so Ushio01's claim about it being when Intel goes beyond quads in the mainstream seems reasonable.

    I've also seen rumors that Apple's threatened to switch to one of Intels rivals if they don't cut TDPs in their mainstream laptop CPUs significantly. I'm not sure how credible I find this claim, vs them just wanting the performance gap for LV/ULV parts narrowed. Apple's far more able to pay a premium for low power CPUs than most of its PC rivals are giving them a competitive advantage. Forcing power levels down globally benefits their rivals as much as it does them.
  • Meaker10 - Sunday, December 4, 2011 - link

    1333 to 1600 makes almost 0 difference with these CPUs at dual channel.

    We have laptop ram that can run at 1866mhz already so bandwidth as always is not an issue.

    Intel is not pushing things in certain directions because they dont have to.

    Who exactly would Apple move to? AMD? Nope. Arm? Nope. IBM? Nope.
  • Ushio01 - Saturday, December 3, 2011 - link

    With the i5 2500K offering so much performance with low power usage why would anyone buy an Intel dual core for there own build. Most websites seem to prefer recommending AMD quad cores for there own system build guides afterall.
  • Taft12 - Saturday, December 3, 2011 - link

    *raises hand*

    The i3-2100 is about half the price of the i5-2500K. It's not exactly insignificant. 1/3 less TDP too.
  • dusteater - Saturday, December 3, 2011 - link

    I think it is funny saying 7 series chipsets will support USB 3.0, when we are only talking 4 ports. I really do not understand why it is so hard to make all the ports USB 3.0? Intel fails again, and I will not be replacing all of my computers/servers in 2012 like I hoped.

    Again, it's been 4 years since USB 3.0 has been ratified and all Intel can come up with is putting 4 ports in their chipset. Pathetic.
  • dagamer34 - Saturday, December 3, 2011 - link

    Because the required bandwidth to supply all of those ports quite a bit, unless you want Intel to put in more ports than the bus can handle. Then you run into the situation where you can saturate the bus if you push through enough data.

    Forcably adding more ports when you don't increase bandwidth just means you've added a hub.
  • DanNeely - Saturday, December 3, 2011 - link

    USB2 took several years to fully displace USB1, for the same reasons. In addition to the overall bandwidth problem; most devices only used the lower bandwidth connection and saving a dollar/chipset by only putting enough of the bigger high speed controllers on for the number of high speed ports a typical consumer might actually need adds up to real money over the entire production run.
  • 3DoubleD - Saturday, December 3, 2011 - link

    How many USB 3.0 devices are you planning to connect? 4 ports seems like it would cover the majority of possible usage cases. If you need more you could thrown in a USB 3.0 PCIe card for <$50. Your mouse, keyboard, and USB 2.0 devices can use the plentiful regular 2.0 ports.

    On a separate note, I haven't even seen any USB 3.0 devices that looked very attractive to buy. There are a few cheap flash drives, but their performance hardly exceeds USB 2.0 speeds. USB 3.0 isn't ubiquitous enough to warrant buying an expensive flash drive in most cases. External HDDs could use the extra bandwidth for sequential read/writes. More complex attached storage devices could use the extra bandwidth, but you would only need one port per device, but how many devices are you going to have? Probably not more than 4. The USB 3.0 spec looks great for mobile device charging, but I haven't seen any phones with microUSB 3.0 ports to take advantage of the spec.

    Lastly, don't forget that mobo manufacturers can add additional USB 3.0 ports through 3rd party controllers, just like they have been doing on Z68 boards. So there might be many boards with 6 or even 8 USB 3.0 ports.
  • Death666Angel - Saturday, December 3, 2011 - link

    Hey!
    Thanks for the article!
    But I do have one problem with the "Specifications of Low Voltage Ivy Bridge CPUs" chart. The last 2 columns overlap with the Twitter feed. Is it just me? I use FF8 with 1920x1200 resolution. :-(
  • Death666Angel - Saturday, December 3, 2011 - link

    Nevermind, someone else already pointed it out.... that's what I get for searching for "chart" not "table". :D
  • Kristian Vättö - Saturday, December 3, 2011 - link

    Someone else also said that the table doesn't work properly in FF8 (see first page of comments). I'm using Chrome and the table works fine, although it's a very tight fit.
  • X-Files - Saturday, December 3, 2011 - link

    You may also want to take a look at this:
    http://hwbot.org/forum/showthread.php?t=34548.
  • ClagMaster - Saturday, December 3, 2011 - link

    This is the CPU and chipset I will be upgrading from my 5 year old G965 Chipset / Q6600 CPU / 4GB DDR2-800 PC.

    Double the performance for slighly less thermal load and for about the same cost as the original components.

    I will even throw in a SSD now that Harddrive prices have skyrocketed from Thailand flooding.
  • randomlinh - Saturday, December 3, 2011 - link

    same here, but only coming from 3yrs ago, heh. Though, it really depends on what price point. And if sandy bridge chips/boards go on a fire sale, I may just do that instead. Though, I was hoping these would be out earlier. My new build will likely get pushed to summer then.
  • gramboh - Monday, December 5, 2011 - link

    Same here. I`ll be upgrading from a Q6700 @ 3.3GHz (formerly E6600 @ 3.2), P965 mobo and 4GB DDR2-800 which I built in April 2007. It has seen 3 video cards (8800 GTS 640MB -> GTX 280 -> GTX 560 Ti) and is still going strong. BF3 runs pretty nicely at 1920x1200 on high detail/2X AA/16X AF, but the load times are brutal, I really need an SSD.

    I'm happy with what Ivy Bridge is offering. 77W TDP on the 3770K means 4.5GHz should be a breeze, maybe even closer to 5GHz, not that it's needed for my use (desktop and gaming).

    Somewhat disappointed the Z77 chipset is only bringing 2 SATA6 ports, although I doubt I will ever have more than 2 SSDs installed anyway. Integrated gigabit ethernet is nice (Intel I assume).
  • frozentundra123456 - Saturday, December 3, 2011 - link

    It is about what we expected, but this is kind of a disappointing line up to me. I would have liked to see more of a decrease in power consumption at least, if they are not going to increase clockspeed. Going from 95 to 77 watts on the desktop parts is no big deal to me. If you have a system with a high end graphics card or even SLI, saving less than 20 watts is a minimal decrease. And again, increasing IGP performance on the desktop isnt a big deal either. Maybe for OEM systems, the increase in IGP performance will impress, but these chips are so fast that it seems a waste not to have a discrete card in a desktop.

    Maybe this lineup for the mobile sector will be more of an improvement since power usage and IGP preformance mean more there. I also wonder if all the mobile parts will get the HD4000 like all SB mobile chips got the "high end" IGP. I also wonder how the HD4000 will compete vs mobile Llano.
  • ltcommanderdata - Saturday, December 3, 2011 - link

    You mention the launch timeframe for desktop Ivy Bridge is April 2012. Has there been any indication for mobile parts? Are they expected around the same time or will they come afterword, perhaps when dual core parts are available?
  • vhx - Saturday, December 3, 2011 - link

    Not that special of a release, as evident by keeping the i7 and i5 lineup names. Still no 6 core processors, more focus on integrated GPU's. Good direction for laptops and maybe HTPC's, boring for everything else including desktops.
  • biassj - Saturday, December 3, 2011 - link

    So reading around, in order to use Intel VT-d the motherboard must support it? I can still run VMware without Intel VT-d?

    Note: coming from AMD Phenom II here
  • Kjella - Monday, December 5, 2011 - link

    Sure, it's optional. It's not even the most important virtualization feature, it's only important for heavy disk I/O. If most of your VM work depends on the CPU you'll probably not even notice the difference.
  • Denithor - Saturday, December 3, 2011 - link

    First, 'omision' is spelled 'omission.'

    Second, will the new more powerful GPU accelerate Quick Sync significantly beyond what it already does on SB? I'm curious, DVDFab now supports QS on SB and I have yet to see this reviewed anywhere. Would love to see a legit review here of the technology!
  • Hrel - Saturday, December 3, 2011 - link

    Correct me if I'm wrong, but doesn't "enabling hyper threading" not make producing the chip any more expensive AT ALL?!

    So assuming that, WHY DON'T ALL THESE CHIPS HAVE HYPER THREADING?!!!!

    My laptop has a quad core with hyperthreading and I'll be damned if my dekstop's cpu has less threads. With that said I'm not spending 300+ dollars on a desktop CPU. Right about 200 is what I spend on desktop CPU's. GIVE ME 8 THREADS INTEL! WTF! i've been waiting for a 200 dollar quad core with hyper threading from Intel since the Conroe days when they released the Q6600. I'm getting seriously pissed.

    Or am I just mistaken and it makes the chip significanlty more expensive thereby justifying the 100 dollar price tag? Oh wait i3's have HT but i5's DON'T!!!!! GaHHHHHH!
  • Iketh - Saturday, December 3, 2011 - link

    it's all marketing...
  • MonkeyPaw - Saturday, December 3, 2011 - link

    It's the intel business model. There's what the market wants, what intel can do, and what they actually provide. Rarely do all 3 match. Intel sits on good ideas in case they need to break them out for the sake of competition.
  • toyotabedzrock - Saturday, December 3, 2011 - link

    I hope manufactures throw a fit with Intel soon.

    There are too many parts with differentiation that seems to have been done by someone on crack.

    And whats with disabling of hyper threading. I thought they wanted to push programmers to learn to use more cores.

    And why does one chip have no AES-NI?

    And only 3 have the HD 4000? They are acting like their graphics are actually desirable, which they are not.

    HD 4000 = Slow
    HD 2500 = Slower
  • A5 - Saturday, December 3, 2011 - link

    You're right, Intel has no idea what they're doing. Come on, man.

    They differentiate their products like this because their customers (PC makers, not enthusiasts like us) want them to.
  • danjdanjdanj - Saturday, December 3, 2011 - link

    That's 1600 MT/s not MHz. 1600MHz would inply 3200MT/s, which it cannot support
  • stephenbrooks - Saturday, December 3, 2011 - link

    Looks like because Intel has such big a technical lead, speed of development is being dictated by how fast AMD can catch up (or not). Perhaps they'll add 0.1GHz per year for a few years?
  • Phylyp - Sunday, December 4, 2011 - link

    With these new (and lowered) TDP figures, would that imply we can overclock an IB a little more than an SB?

    Regarding the truncated feature set for the K-series (no VT-d, TXT, etc.), I've read that VT-x is the important once to have for virtualization for consumers. VT-d is more for hypervisors used in the enterprise.

    Losing AES-NI is a bit of a shame - while most of the software that uses it (http://en.wikipedia.org/wiki/AES-NI) is enterprise related, there are enough consumer products there (BitLocker and 7-zip).
  • Death666Angel - Sunday, December 4, 2011 - link

    "Losing AES-NI is a bit of a shame"<< The K-versions do have AES-NI enabled. :-)
  • Phylyp - Sunday, December 4, 2011 - link

    My bad, thanks for the correction :)
  • iwod - Sunday, December 4, 2011 - link

    May be questions for next time the Anand crew meet up with Intel,

    More SATA 6Gps Port? Ideas on SATA 12Gbps? We need that soon for SSD
    Graphics Drivers, Any news on improvement or Train Release Schedule like AMD and Nvidia?

    OpenGL 4.2 Support?

    My guess is that Software is where the development cost. Making GPU is much simpler then getting the drivers done and constantly update.

    USB 3.0 - Will Intel again include a USB 3.0 controller that requires huge CPU usage?

    More QuickSync Details? - 4K H.264 Profile 10 decode? There are already work on FFDShow from an Intel Engineers for QuickSync on SB.

    More Encoder Improvement? Speed and Quality? Its speed should at least surpass x264 fastest option encoding.
  • iwod - Sunday, December 4, 2011 - link

    Some of my questions answered here..

    http://hwbot.org/forum/showthread.php?t=34548.

    QuickSync is 50% faster, No mention of quality, if it is 50% faster and much better quality then this is a win.

    Only two 2 SATA 6Gbps Port
  • iwod - Sunday, December 4, 2011 - link

    I am still hoping we see another IPC Jump from Haswell. ( FMA )
  • beginner99 - Monday, December 5, 2011 - link

    ... you get away with crap like less features on k versions or putting the fast GPU in the expensive chips. Did not make sense with SNB and still doesn't now.

    But probably GPU is more likely to have errors and hence binning reasons? Nothing else makes sense.
  • adrianajones - Monday, December 5, 2011 - link

    There's a lot of buzz about 2012 laptops. Computers are getting so thin these days and the tablet market seems to be picking up quite considerably. If I'm not getting a tablet, I'm going to need a laptop table to go with this so I can work mobile. Here's what I'm thinking of http://bit.ly/vy3hNS
  • DanNeely - Monday, December 5, 2011 - link

    ***SPAM***
  • Hector2 - Tuesday, December 6, 2011 - link

    I suspect that there aren't any i3's listed yet simply because the higher volume Sandy Bridge chips are still cheaper than IVB chips is right now. Until the 22nm volume matches 32nm volume, they'll prefer to ship 22nm into segments that have higher margin and need the lower power it brings
  • Morg. - Wednesday, December 14, 2011 - link

    From intel, show the supposed performance of those Ivy Bridges ...

    It doesn't look like anything interesting for gamers right now, but ...

    Process shrink 22 nm : +30% powa
    Trigate : +x% powa

    Intel seems to have taken 20% of that for power consumption (very likely to get back at AMD who's been leading the perf/watt race for a few generations in serverland already)
    and 15% in performance.

    Let's hope the rest (x-5%) will be dedicated to overclocking headroom :p ).

    Overall the new chips would be 40% more efficient - one could hope that they will be Overclockable to death (I'd expect 6 Ghz under water following past trends.. but it could be fragile like sandy bridge . we'll see.) and they will incorporate a fully useless but yet improved Intel Graphics .

    When those Ivy Bridges go out, the Bulldozers will be past their early quirks including bios patches, driver patches, platform integration patches and windows 8 will be almost out and the performance picture will look a bit like this :

    Sandy Bridge : 100 (good OC headroom)
    Bulldozer : 115 (better OC headroom)
    Ivy Bridge : 115 (with more thermal headroom and potentially higher clocks -- )

    I can only say I'm happy, this is going to be a perfect time to buy anything as Intel will be fighting instead of stomping like these past 5 years.
  • Morg. - Wednesday, December 14, 2011 - link

    My bad. To get back at AMD in serverland, and ARM in tabletland.
  • Christian Ney - Wednesday, December 14, 2011 - link

    Source ?
  • JedLuna - Monday, January 16, 2012 - link

    are all these for lga1155 platform?

Log in

Don't have an account? Sign up now