Rumor: Core i9 processors will arrive for LGA 2066 in June

If you have a distaste for AMD rumors, well, don't worry—we have some new dirt about Intel's upcoming hardware today, too. The skinny is this: if the info is accurate, Intel will be branding its purported quad-core Kaby Lake-X CPUs as Core i7s, while the rumored six-, eight-, 10-, and 12-core Skylake-X CPUs on the same platform will be known as Core i9s. Watch your sodium level as we dive in.

This information comes courtesy of a few people on the Anandtech forums. Forum user Sweepr posted a list of upcoming processors and promised an image as proof, which was subsequently delivered by forum user dooon. The image appears to show an Intel presentation open in Powerpoint, displaying a list of LGA 2066 processors and their relevant specifications. We've reformatted the information in a table below.

  Cores/threads Base clock Turbo "Turbo Max" L3 PCIe 3.0 Memory TDP
Core i9-7920X 12/24 TBD TBD TBD 16.5 MB 44 lanes Quad-channel

DDR4-2666

140W
Core i9-7900X 10/20 3.3 GHz 4.3 GHz 4.5 GHz 13.75 MB 44 lanes Quad-channel

DDR4-2666

140W
Core i9-7820X 8/16 3.6 GHz 4.3 GHz 4.5 GHz 11 MB 28 lanes Quad-channel

DDR4-2666

140W
Core i9-7800X 6/12 3.5 GHz 4.0 GHz N/A 8.25 MB 28 lanes Quad-channel

DDR4-2666

140W
Core i7-7740K 4/8 4.3 GHz 4.5 GHz N/A 8 MB 16 lanes Dual-channel

DDR4-2666

112W
Core i7-7640K 4/4 4.0 GHz 4.2 GHz N/A 6 MB 16 lanes Dual-channel

DDR4-2666

112W

There's a lot to talk about here. Most profound is further fuel for the notion that Intel will once again be increasing the core count of its top-shelf Core-series processor, this time to 12 cores. The idea of having to buy a ten-core processor to run a pair of graphics cards with a full 16 lanes of PCIe 3.0 for each remains as annoying as ever.

A few discrepancies popped up between the information posted by Sweepr and the information in the image posted by dooon. Sweepr claims that the Skylake-X-based Core i9 processors will have TDP specs ranging up to 160W, but all of the chips in the image appear to say 140W. Sweepr also claims that the Skylake-X processors will have 1 MB of L2 cache. That would be four times as much as today's Skylake and Kaby Lake processors, and might help to explain the relatively paltry amount of L3 cache in the Skylake-X CPUs. The Core i7-6950X ten-core CPU has 25MB of L3 cache, for comparison's sake.

The information didn't come with any pricing details, so we'll have to wait and see what the bottom line looks like for these chips—assuming this information is even accurate. If so, that means we'll be seeing these parts next month save for the 12-core Core i9-7920X. That part is supposedly launching in August.

Comments closed
    • Lord.Blue
    • 2 years ago

    Lots to digest here, but shouldn’t the 7640k be a core i5?
    4 non-hyper-threaded cores = core i5, right?
    Just an observation.

    • southrncomfortjm
    • 2 years ago

    Sorry, I get lost in all the naming conventions and designators. I forget what an X v. an E processor is.

    Just looking at what’s up there, the 7740K looks like the right thread/core count for high end, single GPU gaming. Am I right? Are these even enthusiast parts, or are they above that?

    I guess I’m just confused how the 7740K fits in with the 7700K and whatever the upcoming Coffee Lake Equivalent to the 7700K is. Any clarity would be appreciated since I’m biding my time for a CPU/Mobo/Ram upgrade.

    • End User
    • 2 years ago

    The Core i9-7820X looks nice.

      • Krogoth
      • 2 years ago

      Hopefully pressure from Ryzen 7 1800X keeps the price around $399-$600

    • Mad_Dane
    • 2 years ago

    Makes no sense at when they already have 6 and 8 core i7’s!

    • BIF
    • 2 years ago

    I like this strategy. It’s about time to split the i7 group into two for better organization.

    I’ve also disliked how the i5/i7 nomenclature doesn’t align to the Xeon nomenclature. i7 equates to E5? Why would you do that, Intel?

      • ImSpartacus
      • 2 years ago

      I guess that’s why the “E#” nomenclature appears to have disappeared.

      • psuedonymous
      • 2 years ago

      E5 always was a bit of a weird bastard designation. You had the E5-1xxx series and the E5-2xxx series. E5-1xxx could be overclocked, were single socket (no QPI bus), and basically were i7-extremes that had a higher binning and in some markets ended up a bit cheaper. E5-2xxx were double-socket locked dies with no consumer-targeted equivalent.
      With the silly new Gold/Platinum/wtc designation, at least the single-socket dies have been split off into their own designation (and even own socket).

    • Anovoca
    • 2 years ago

    Is this where I get to say “CALLED IT”

    [url<]https://techreport.com/forums/viewtopic.php?f=2&t=119314&p=1343218#p1343218[/url<] [quote="Anovoca"<][quote="derFunkenstein"<]The only reason the Pentium brand got demoted to mean "budget" is because Intel ruined the name with Prescott and later Netburst chips. I would expect Core to stick around as a name as long as it's associated with good performance.[/quote<] Pentium has made a comeback. Once all the people that jumped ship to the i3 for web browsing pcs realized that the Pentium processors were running circles around them in single threaded applications. So i guess if anything maybe expect intel to drop the i3 and load up a new chip with the i9 moniker with even MORE CORZZZZ.[/quote<]

      • ImSpartacus
      • 2 years ago

      I don’t think it’s much of a stretch to say that Intel intentionally saved “i9” and of course it would bring an increase in cores as compared to i7.

      Furthermore, I don’t think the i3 branding will be dropped.

      • the
      • 2 years ago

      Well if you want to see called it, I offer [url=https://techreport.com/forums/viewtopic.php?f=2&t=93911&p=1206147&hilit=refresh#p1206147<]this post from [i<]three years ago[/i<][/url<]. Not 100% accurate but looking back I think I got most of it right so far. My big failure was expecting Sky Lake to support 4 way SMT, which I think is still coming down the road in a future core revision and the lack of eDRAM options on some parts. But to my credit, I did predict SkyLake-E would be called the Core i9, predicted Kaby Lake as Sky Lake refresh, got the core count correct with Broadwell-E and consumer Cannon Lake (which is now 14 nm Coffee Lake). I was also wrong about the increase in PCIe lane counts but kinda did happen in the chipsets for NVMe. The funny thing is that things are still hazzy after Cannon Lake.

        • derFunkenstein
        • 2 years ago

        Y’all don’t tear a rotator cuff patting yourselves on the back. More cores coming in a new architecture? A bump in marketing numbers? Alert the media! There’s a new prognosticator in the house.

          • the
          • 2 years ago

          I prefer the term procrastinating prophet. I will tell you why but that’ll have to wait for another time.

          • BIF
          • 2 years ago

          When I saw the first music CDs, I predicted storage on hard drives and eventually entire business basing their models on streaming content. Unfortunately, the internet wasn’t created yet, so nobody can prove I’m a liar. 😀

          Hey, does anybody have some Icy-Hot or Australian Dream? 😉

        • Ninjitsu
        • 2 years ago

        Heh my speculation is just above yours 😀

        Correctly called the Pentium and i7 Extreme spec bumps, wrong about the rest so far. I think i expected Broadwell-E to be skipped for Skylake-E.

    • techguy
    • 2 years ago

    I hope this naming scheme turns out to be nothing more than the resurrection of this ancient rumor:

    [url<]http://gizmodo.com/5411119/intel-core-i9-benched-six-cores-of-pure-joy[/url<]

    • albundy
    • 2 years ago

    the i9’s are extreme processors? how many mortgage payments is that gonna be?

      • Krogoth
      • 2 years ago

      They are probably going from $499-$1499. R7 Ryzen line-up should be putting price pressure on Intel.

        • Klimax
        • 2 years ago

        If frequency is correct, I don’t expect price cuts. (Unless Intel decides that AMD outlived its usefulness)

          • Krogoth
          • 2 years ago

          The lower-end i9 parts are going to be pair-up with R7 Ryzen line-up while the high-end stuff is going take up Intel’s $999+ top of the line spots. The only way Intel will do a price cut on the high-end stuff is if AMD came out with their high-end HEDT parts based on Naples-architecture and try to undercut the high-end i9 stuff.

          I doubt will we see a repeat of ~$1704 MSRP for the top of the line part though.

    • Unknown-Error
    • 2 years ago

    I guess we must thank AMD for poking a sleeping dragon.

      • the
      • 2 years ago

      And Intel is still holding back. Intel has some pretty crazy parts that could fit into socket 2066 if they really were threatened (say 32 cores?).

        • shank15217
        • 2 years ago

        So? Its a server socket, AMD could hit right back with Naples.

          • the
          • 2 years ago

          The Xeon socket for SkyLake-EP is actually LGA 3467. If Intel wasn’t holding back, they’d release the Core i9’s using it.

          AMD’s enthusiast platform according to rumor is using the same socket as Naples but with the half the memory channels and PCIe lanes. AMD could counter with Naples but it would be crippled compared to the full Opteron implementation. Still this opens up oppertunities for AMD down the road.

            • shank15217
            • 2 years ago

            So, basically another socket segmentation. This would mean all LGA 2011-3 platform that could probably be updated easily to Skylake is now EOL. Intel just gave the consumer another middle finger and nobody bats an eye.

            • the
            • 2 years ago

            That is because Intel has a two generation cadence for their sockets:
            LGA 1366 is Nehalem/Westmere
            LGA 2011 is Sandy Bridge-E/Ivy Bridge-E
            LGA 2011-3 is Haswell-E/Broadwell-E

            The thing about the pattern above is that Intel is actually changing the feature set with each generation by adding more memory channels and/or changing memory type. No one batted an eye as they were able to see improvements.

            LGA 2066 is the first platform where there the updates are minor but they’re there: more PCIe lanes and better configuration of them.

            • Chrispy_
            • 2 years ago

            Rumoured, yes.

            Personally I don’t think that multi-processor, quad-memory-channel processors can work in a single 1331-pin socket without being horribly restricted.

            I would guess that there will be a dedicated server socket for Naples and Snowy-Owl-based Opterons, simply because the additional pin-count required for the inter-CPU logic and two extra memory channels must be huge. Intel requires an extra 860+ pins for that specific functionality. If AMD are doing it in zero extra pins, they have some new-fangled voodoo magic I do not understand.

            • chuckula
            • 2 years ago

            There’s literally no way for the 1331 pin socket for AM4 to support a quad-channel memory configuration. There are no physical pins present to actually hook up the different memory channels. That’s why the rumored 16 core workstation parts are using the same 4094 connector (LGA) socket as Naples server parts, although many of those connectors are unused in the workstation-grade systems.

      • blastdoor
      • 2 years ago

      You can buy a 10 core i7 broadwell today, so this isn’t a big deal.

      What would be a big deal is a price cut. Say, $800 for the 8 core one.

        • shank15217
        • 2 years ago

        A server socket means that the platform is way more expensive, Ryzen really shouldn’t be compared to anything else but Intel’s i7 + GPU line. I am sure AMD will have a qualified “workstation” line with Naples to fight Intel’s Xeon based core i7s. AMD might position the Ryzen to fight against the Core Extremes but that platform is a server platform in workstation clothing.

          • the
          • 2 years ago

          Using a server socket for the high end desktop is nothing new for Intel (see LGA 1366, LGA 2011, LGA 2011-3 etc.). SkyLake-X will actually be the odd ball by using a different socket than the Sky Lake-EP Xeons (LGA 3467).

    • willmore
    • 2 years ago

    This is Intel’s response to Ryzen in the enthusiast end of the market? What a huge bundle of ‘meh’. Too few cores. Too slow. Too hot. And, we all know it’ll be too expensive.

    It’s been so long since Intel had to be ‘lean and mean’ that it looks like they forgot how to do it.

      • DancinJack
      • 2 years ago

      Too expensive, maybe, but too few cores, too slow, and too hot? You’re confused I think.

      I don’t know what the different turbos mean, but i’m guessing base, all four turbo, single core turbo. Since when is 8 Skylake cores @ 4.3 “too slow”?

      You must really hate Ryzen considering the 1800X is 3.6-4GHz on 8 cores with lower IPC than Skylake.

    • brucethemoose
    • 2 years ago

    When marketing departments go to war!

    This is exactly what AMD wanted, as now that have an excuse to use the “R9” branding in CPUs.

    Which, to the joy of meme enthusiasts everywhere, will force Intel to respond by taking it up to 11.

      • DancinJack
      • 2 years ago

      The Ryzen CPU’s aren’t “R3” or “R7,” but it would have worked out well if they were…

    • Chrispy_
    • 2 years ago

    I’m so glad Intel have finally stopped calling them i7 [i<]Extreme[/i<], and just admitted that they're nothing to do with the other i7's whatsoever. "EXTREEEEME DEWD" is so 80's.

      • CScottG
      • 2 years ago

      Having lived in that period, it’s totally not, like: fur shure.

      -I think you’d have to fast-forward about 25 years.

      (..now where is my 1040ST and a copy of The Bard’s Tale?)

        • JustAnEngineer
        • 2 years ago

        [i<]The Bard's Tale[/i<] runs pretty well on your Android phone these days.

      • shank15217
      • 2 years ago

      Its Intel Xeon with multi sockets disabled, fu consumer!

        • Chrispy_
        • 2 years ago

        It has always been this way, ever since Intel decided to split consumer and server platforms (Bloomfield/Lynfield).

        Consumers don’t want the cost of a high pin-count socket and quad memory controller that will often be wasted, so LGA1366 has evolved into LGA2011 and the consumer sockets have stayed steady at LGA115x

        If you’re hoping for a high-core count without needed to change platform altogether, you’ll need to go to AMD. Intel stoutly refuses to put more than four cores on a dual-channel, affordable LGA115x design.

        • chuckula
        • 2 years ago

        Post your mailing address so I can send you every single AM4 dual-socket motherboard that’s available on Newegg and Amazon.

        I’ve got the basket already to go and I’m pretty sure it won’t max out my CC.

          • shank15217
          • 2 years ago

          AM4 is not a workstation / server platform. Why would think this would be a multi-socket version and Intel product segments their line like crazy, mostly to maximize profit over considering their customer base. Defending Intel for artificially segmenting their Xeon line is fanboy-ish.

            • Chrispy_
            • 2 years ago

            AM4 is [i<]absolutely[/i<] a workstation platform. There aren't really any significant consumer tasks that need 16 threads yet. At the very least it's prosumer/workstation and all of AMD's product demos and press-releases leading up to Ryzen were squarely focused on the comparison with Socket-2011 workstation/prosumer Intel chips, very little focus on the mainstream Socket-1151 comparisons. I'm not sure where you think we're defending Intel. It's just a plain statement of how they do things; Whether I agree or disagree with that isn't something I've mentioned or inferred.

            • Krogoth
            • 2 years ago

            AM4 is desktop/entry-level workstation platform a.k.a single-socket w/ ECC support

            • shank15217
            • 2 years ago

            yet there isn’t a single board with ECC support available, at least as a far as i know.

            • Krogoth
            • 2 years ago

            There are a few AM4 boards out in the field with ECC support and I’m certain more are coming down the pipeline.

            • shank15217
            • 2 years ago

            It is absolutely not a workstation platform, see AMD future products, there is a very high chance AMD will disclose a naples based single socket workstation platform soon (along with their server platform) that actually competes directly with core extreme.

            • shank15217
            • 2 years ago

            Opps, you’re wrong

            [url<]http://wccftech.com/amd-ryzen-9-lineup-threadripper/[/url<]

            • chuckula
            • 2 years ago

            Yeah, and look: Single socket-only support.

            What was it you just said in this thread?

            Oh yeah: Its AMD Naples with multi sockets disabled., fu consumer!!

            • shank15217
            • 2 years ago

            Sure, lets put 180w tdp x 2 under your desktop

            • chuckula
            • 2 years ago

            Yeah, typical fanboy respones: When Intel does X: IT’S EVIL HOW COULD THEY!!

            When AMD does literally the exact same X: THANK YOU AMD FOR SAVING HUMANITY AGAIN!

            • shank15217
            • 2 years ago

            By producing an alternate viable cpu platform AMD actually did make a difference, if you cant see that then you’re the fanboy.

            • chuckula
            • 2 years ago

            Uh… I seem to recall Raj showing off 2 Rx 480s running in tandem as the end-all be-all solution for standard gaming PCs.

            That sure as hell sounds like 2×180 watt parts (more if you overclock!) running in what is supposed to be a “cheap” desktop PC to me.

            So you’re saying that Raj is an idiot and that professional grade workstations are incapable of using parts with the same power draw and cooling requirements as supposedly “cheap” desktop PCs that are advertised on stage by AMD Executives.

            You know, for an AMD fanboy you sure seem to have no problem calling every single AMD employee an idiot in the pursuit of your dreams.

            • Chrispy_
            • 2 years ago

            I’m not sure what the point of this post is.

            A) I’m not wrong, and a company in London I collaborate with has already purchased over 200 Ryzen 7 workstations for 3D modelling, because they are incredible value for money that quite literally blow Intel out of the competition when it comes to multithreaded performance/dollar. We only have 8 ourselves – mostly used for very high-resolution renders and environmental simulations.

            B) The existence of upcoming upcoming ‘socket 4094’ quad-channel platform does not automatically mean that AM4 cannot be a workstation platform. They’re not mutually exclusive; Both AMD and Intel have had workstation/consumer crossover between their consumer and server platforms for the best part of a decade. Again, this isn’t exactly news, just wikipedia search i7 and Xeon: You’ll see the lines are so blurred that this hard distinction you’re trying to make is not only wrong, it’s utterly pointless in the first place. Same exact thing [i<]used[/i<] to happen with AMD opterons, back in their heyday. It seems (unless I'm being completely stupid here) that you're countering your own first argument with this second argument. I mean, you started this rant with "it's Xeon with mutli-sockets disabled, fu consumer" and now you're defending it with "here's AMD's [i<]Xeon equivalent[/i<] with multi-sockets disabled" I don't know what you're getting at man. I really don't understand what angle you're even trying to argue anymore.

            • chuckula
            • 2 years ago

            1. Once upon a time (less than 3 months ago) Lisa Su compared the 1800X to a 6900K onstage to show off how awesome RyZen is. Sounds like she thought the 1800X is a competitor to Intel’s HEDT lineup to me!

            2. But yes, you are right and Lisa Su doesn’t know what she’s talking about. So Intel’s HEDT lineup is some evil artificial marketing segmentation strategy or something. Well, since you yourself just said that RyZen doesn’t compete in the workstation market, care to point out AMD’s dual-socket workstation motherboards to me? Because I’m not seeing any other than repurposing a dual-socket server motherboard and calling it a workstation motherboard, and you could do that with Intel boards right now.

            The fact that Intel has products in a category that AMD chooses to ignore is no reason to blindly bash Intel and defend AMD based on some executives at AMD deciding that some market segments don’t count.

            • shank15217
            • 2 years ago

            Yea, its because she compared on AMD’s strength of having more threads over IPC. The most effective comparison was with the 6900K. And you are wrong, AMD will release a workstation class platform along with their server platform because it makes sense.

      • EzioAs
      • 2 years ago

      Hey hey hey! Some people like the 80s.

    • Krogoth
    • 2 years ago

    If the stats hold true then Kaby-Lake-X = massive disappointment. It cannot even take advantage of Socket 2066 platform so what is point of it? The only reason Intel would bother is that ignorant buyers will end-up being beta-testers for their Socket 2066 platform.

    Skylake-E stuff makes sense for HEDT related-stuff but depending on price it may end-up fighting an uphill battle against the R7 Ryzen line-up.

      • mganai
      • 2 years ago

      It’s only for giving people a gateway upgrade path. Though at that point they may as well just go in with one of the pricier choices.

        • Krogoth
        • 2 years ago

        It is not much of an upgrade path.

        LGA2066 is the successor to the LGA2011. You might as well get a real HEDT chip instead of a gimp desktop-tier chip (it cannot utilize its IGP) in a workstation-tier platform.

      • Bauxite
      • 2 years ago

      Also known as a joke.

    • RAGEPRO
    • 2 years ago

    5 GHz and >4000 MT/s memory better both be more or less “gimmes” on that 7740 part.

      • RAGEPRO
      • 2 years ago

      I also wonder if it’s possible to have an LGA 2066 motherboard with four DIMM slots that supports dual-channel operation with two DIMMs per channel on Kaby Lake-X, and quad-channel with four DIMMs (one per channel) on Skylake-X. Seems improbable, but that’d be pretty nice for Kaby Lake-X. It’d drive me crazy to have to leave half my RAM slots empty.

      [sub<][i<]Edited to fix a typo that screwed my meaning, but I think DancinJack figured it out anyway.[/i<][/sub<]

        • DancinJack
        • 2 years ago

        Considering the MC is in the CPU, I’d actually bet that is the case. Having to make ANOTHER set of SKUs for each KL-X and SKL-X would probably not sit well with OEMs.

        • MOSFET
        • 2 years ago

        I think it’s going to drive me crazy when we’re forced to consider the 7740K “HEDT-material” (so, soon)

          • DancinJack
          • 2 years ago

          meh. it’s not like you (or anyone) has to buy one. I don’t see what the big deal is.

            • MOSFET
            • 2 years ago

            and I won’t!

            meh is the dumbest word on the internet.

            • DancinJack
            • 2 years ago

            meh.

    • jihadjoe
    • 2 years ago

    Now that they finally introduce i9 (I believe they first thought of doing this back with Gulftown) maybe Intel can adjust the model numbers to be consistent with each family, instead of HEDT being a generation behind.

    i7-8xxx/i9-8xxx = hopefully Coffee Lake / Coffee Lake X, instead of i7-7xxx/i9-7xxx being Kaby and Skylake-X.

      • DancinJack
      • 2 years ago

      I think they said a while back they’re going to mix generations within the same i7-7, i9-7, etc. They’re not gonna wait a year to release the HEDT stuff and name it after “last year’s” mainstream SKU’s.

      You’ll definitely have to pay attention if there are specific features you want from a generation.

      • DavidC1
      • 2 years ago

      Since the Skylake-X is based on the server core, there is no Kabylake or Coffelake. The next is either Cannonlake or Icelake.

        • the
        • 2 years ago

        Intel indicated that Sky Lake-EP is using the updated process found in Kaby Lake. Thus from a foundry perspective Sky Lake-EP = Kaby Lake-EP.

        Intel has also indicated that the 10 nm server chips will be of the Cannon Lake family but will use the revised 10 nm process after consumer Cannon Lake. Intel has intentionally been slowing down server releases to an 18 month to 24 month cadence.

    • DancinJack
    • 2 years ago

    I mean, Intel can’t let AMD have a 3, 5, and 7 just like they do! Gotta up it to nine bruh.

      • Redocbew
      • 2 years ago

      A 16 core chip would take it to 11.

        • DancinJack
        • 2 years ago

        Speaking of, I really, really want a 7820X. I have no idea how high they’ll price that thing, but I want one.

        • maxxcool
        • 2 years ago

        /facepalm/

        • jihadjoe
        • 2 years ago

        How about an 11-core chip?

      • Anton Kochubey
      • 2 years ago

      AMD’s graphics division has R3, R5, R7 and R9 GPUs. I won’t be surprised if AMD will release 16-core Ryzen 9 😛

    • kmm
    • 2 years ago

    If you’re doing the kind of real work / computation where you need two graphics cards with 16 PCIe 3.0 lanes each, or are an enthusiast benchmarking top-of-the-line setups, I don’t think the price of the i7-7900X or i9-7920X is that much to ask and it’s probably appropriate anyway.

    Are people really that desperately in need of over 28 lanes? Legitimately, what for? I mean, more is nice and better, don’t get me wrong. But the need is ____ ?

      • Redocbew
      • 2 years ago

      16 for a GPU, plus another 8 for 2 NVMe SSDs gets you to 24, so I can see 28 not being completely unreasonable. It’s certainly a niche, and something you’d have to work at to fully utilize. 44 lanes is indeed a bit silly though. I guess(maybe?) there could be a use if there are many concurrent users involved, but in that case you’re not really in workstation territory anymore anyway.

      • brucek2
      • 2 years ago

      My issue is more with 16 than with 28. Needing 29+ lanes is probably not mainstream; but one graphics card + one PCIe SSD either soon will be or already is and it’s frustrating that most parts do not provide enough lanes for this case.

        • brucethemoose
        • 2 years ago

        Chipsets have x4 connections now, which is enough for a PCIE SSD.

      • MOSFET
      • 2 years ago

      “a pair of graphics cards with a full 16 lanes of PCIe 3.0 for each”

      Also wondering exactly what the [i<]need[/i<] here is. Last I looked, x8 @ 3.0 was enough for all but the benchmark-obsessed. To be clear, I too want more PCIe lanes. I want them for video and audio and storage and networking, just like most of us. But there are other bottlenecks to worry about before being too concerned about limiting a GPU to x8 (such as DMI).

        • the
        • 2 years ago

        nVidia has limited SLI to two cards with Pascal so that side of the argument is satisfied with 8 lanes per card. Granted 16 lanes each ever so slightly faster but not worth the price of a platform upgrade.

        AMD still permits quad GPU crossfire but only needs four lanes per card. Since they got rid of the Crossfire connector increasing the amount of PCIe bandwidth does help. The real problem for AMD is they they haven’t had anything on the high end for awhile that would motivate people to spend money on several cards. It is simply wiser to get a beefier single GPU solution than get two or more smaller GPUs. That’ll likely change with Vega. Beyond Vega, AMD is a member of the openCAPI group so their future GPUs would benefit from that (think nvLink for AMD). This would require and AMD platform to fully realize. Long story short more PCIe may become advantageous on the AMD side.

    • blastdoor
    • 2 years ago

    The only reason this is potentially interesting is from a marketing/pricing perspective.

    Does Core i9 connote a lower price than monikers like “Xeon” and “Extreme”?

    If not, then *yawn*.

      • ImSpartacus
      • 2 years ago

      No, I think i9 just means “better than i7”.

      In prior HEDT parts, Intel kinda sucked. They weren’t consistently good at gaming compared to cheaper options.

      Now with those clocks (especially Turbo Boost 3.0) and Skylake still being the best architecture in the land, Skylake-X can actually be convincingly better than Skylake. Therefore, Intel can name it that way.

    • GatoRat
    • 2 years ago

    Ah, the joys of competition.

    • chuckula
    • 2 years ago

    OMG they used a combination of LETTERS AND NUMBERS for their models? WORST NAMING SCHEME EVAR!!

    Slightly more seriously, I have no problem with the i9 part of it.

    The i7 models are too confusingly similar to existing LGA-1151 models, however, and they ain’t socket compatible, so that could be an issue.

      • Redocbew
      • 2 years ago

      [quote<]I have no problem with the i9 part of it[/quote<] Le gasp! I'm not sure why anybody would have a problem with these being called i9's, but that may just be me being reasonable again. I do that a lot.

      • DPete27
      • 2 years ago

      [quote<]Slightly more seriously, I have no problem with the i9 part of it. The i7 models are too confusingly similar to existing LGA-1151 models, however, and they ain't socket compatible, so that could be an issue.[/quote<] Yep. I think i9 is a good notation. However, it looks like they're just tagging the i(x) with number of cores. Why not just make it i7-7740X and i7-7640X.

      • CuttinHobo
      • 2 years ago

      Intel is in full-on panic mode.

      Ryzen’s release caused CEO Krzanich to pull a Gary Oldman in [i<]The Professional[/i<] - "Bring me EVERYONE!!"

      • abiprithtr
      • 2 years ago

      Core i9? Bah !!

      I am holding my breath for the launch of the $40 Core i1 processors with a 1-core, 2GHz speed non-HT non-turbo part that can overclock to 4.8GHz and will prove to be an IPC monster giving awesome single core benchmarks to show on my 2008 netbook and can games when unburdened by any other process/anti-virus running in the background.

      (To be read very fast with no breaks for breathing, like they do in TV ads that say “Conditions Apply”)

Pin It on Pinterest

Share This