Intel’s Core X-series CPUs and X299 platform revealed

AMD’s Ryzen CPUs and their oodles of cores and threads have taken the market by storm of late, and the company seems poised to deliver even more computing power this summer with the Ryzen Threadripper high-end desktop chip. Of course, you can’t make noise like that in the CPU market without awakening a sleeping giant.

Intel is reasserting its x86 dominance this morning with a broadside of Core X-series high-end desktop CPUs, and it’s also introducing a new host platform with the X299 chipset and LGA 2066 socket. This platform can host chips ranging from four cores and four threads all the way up to a monster 18-core, 36-thread Extreme Edition chip: the Core i9-7980XE.

Model Base

clock

(GHz)

Turbo

clock

(GHz)

Turbo

Boost

Max 3.0

clock

(GHz)

Cores/

threads

L3

cache

PCIe

3.0

lanes

Memory

support

TDP Socket

Price

(1K

units)

i9-7980XE 18/36 LGA 2066 $1999
i9-7960X 16/32 $1699
i9-7940X 14/28 $1399
i9-7920X 12/24 $1199
i9-7900X 3.3 4.3 4.5 10/20 13.75MB 44 Quad-channel

DDR4-2666

140W $999
i7-7820X 3.6 4.3 4.5 8/16 11MB 28 $599
i7-7800X 3.5 4.0 NA 6/12 8.25MB Quad-channel

DDR4-2400

$389
i7-7740X 4.3 4.5 NA 4/8 8MB 16 Dual-channel

DDR4-2666

112W $339
i5-7640X 4.0 4.2 NA 4/4 6MB $242

Before we talk about the summit of the Core X-series lineup, let’s look at base camp. In what may be a first for Intel’s high-end desktop chips, the Core X-series counts two different generations of microarchitectures among its members. At the base of the pyramid, the four-core, four-thread Core i5-7640X and the four-core, eight-thread Core i7-7740X transplant Intel’s highest-performance consumer Kaby Lake chips onto the X299 platform with higher TDPs, higher base clocks, and dual-channel DDR4-2666 memory support.

Intel seems to think that enthusiasts will choose these Kaby Lake-X chips as a gateway to future upgrades on their X299 mobos over their LGA 1151 counterparts. That might be an interesting prospect for some enthusiasts, but aside from their higher base clocks and TDPs versus their LGA 1151 doppelgangers, Kaby Lake-X CPUs aren’t the most intriguing parts Intel is introducing today.

The Core i7-7800X on up is where the action really starts to happen for this launch.  Intel has tapped its Skylake Xeon production line to repurpose server silicon for X299. The Skylake-X dies (likely 18-core and 12-core base dies) that Intel is using for these parts gives the company a wealth of knobs and dials to tweak as it creates products, and tweak it has. The Skylake-X family boasts a whopping seven chips this time around, compared to the usual four Intel has historically launched with each high-end desktop refresh.

Unfortunately, Intel played its cards rather close to the chest about the highest-end Skylake-X CPUs ahead of today’s Computex keynote. We do know that starting with the 10-core, 20-thread Core i9-7900X, Intel will be using the Core i9 brand to set apart its most extreme desktop processors. The 12-core, 24-thread Core i9-7920X, the 14-core, 28-thread Core i9-7940X, the 16-core, 32-thread Core i9-7960X, and the beastly 18-core, 36-thread Core i9-7980XE (for Extreme Edition) all herald a new era of computing horsepower for the high-end desktop PC— all for a pretty penny, of course. More on that in a second.

The 18-core Skylake-X die. Source: Intel

Aside from those core counts, however, we know next to nothing about Core i9 CPUs as of this writing. Base clocks, Turbo clocks, and Turbo Boost Max 3.0 speeds all remain under wraps as of Intel’s Computex keynote. We do know that those many-core Core i9s will not have modest thermal envelopes. Expect a 165W TDP from the Core i9-7920X on up, and expect clock speeds to decrease as core counts climb.

Intel strongly recommends liquid cooling for these chips, and it’ll be offering an Asetek-produced 120-mm all-in-one cooler for LGA 2066 CPUs called the TS13X.  Folks looking to overclock Core i9 CPUs will likely need even more substantial cooling hardware to extract the full potential from their chips.

Even with this dearth of basic information, the architectural improvements in Skylake-X might give AMD’s Ryzen Threadripper CPUs reason to look over their shoulders as the red team prepares those many-core parts for launch later this year. Let’s delve into why now.

 

Skylake-X gets wider and cachier

While we don’t have full details of how the server version of Skylake stacks up against its mainstream desktop counterpart yet, Intel has let slip a few details. First off, Intel has implemented AVX-512 support with Skylake-X. That means the width of each AVX data register has doubled compared to the desktop version of Skylake, which uses 256-bit-wide registers.

Source: GNU GCC

AVX-512 support also nets 32 data registers for Skylake-X, compared to 16 for typical desktop Skylake. In theory, AVX-512 will yield a doubling of the FLOPS per second per core compared to Intel’s older architectures.  You’ll hear Intel call the Core i9-7980XE its “first teraflop CPU,” and the move to AVX-512 is why. Assuming developers recompile their software to use AVX-512 instructions, SIMD-heavy programs could get a big speedup from these chips.

In perhaps the biggest change to its microarchitecture, Skylake-X also introduces a new cache hierarchy. Each core now has a 1MB private L2 cache, up from 256KB in mainstream Skylake cores. The shared last-level cache, or L3, for Skylake-X has shrunken a bit in turn, to as much as 1.375MB per core. Broadwell-E boasted as much as 2.5MB of L3 cache per core. This new L3 structure is no longer inclusive of the L2 cache, either, as it was with Broadwell-E. As a rule of thumb, quadrupling the cache size as Intel has done here will cut the miss rate in half, assuming the same associativity. We’ll be digging for more details of these changes when we meet with Intel about its X-series CPUs soon, but expect a major performance boost from this rejiggering of the caches on Skylake-X regardless.

It’s not strictly a microarchitectural feature, but Skylake-X also debuts a new take on Intel’s Turbo Boost Max 3.0 technology, first seen with Broadwell-E chips. In the first iteration of Turbo Boost Max 3.0, Intel identified the highest-boosting core on a Broadwell-E die. In tandem with a Windows driver, TBM 3.0 could let those high-potential cores boost well above their normal Turbo speeds for high performance in lightly-threaded workloads. This time around, Intel is snooping out the best two such cores on a Skylake-X die for even more of a performance improvement, and no more Windows driver is needed to let TBM 3.0 work its magic. Intel also tells us to expect more consistent support and implementation of the TBM 3.0 feature from motherboard manufacturers this time around, a welcome improvement compared to the state of the feature on Broadwell-E motherboards.

The X299 platform takes a dip in Kaby Lake

To host Skylake-X and Kaby Lake-X CPUs, Intel is introducing a new socket, called LGA 2066, and a new chipset, called X299. We’ll be covering many, many X299 motherboards this week from Computex, but here are the basics. LGA 2066 uses more pins than LGA 2011v3 before it, but its outer dimensions are identical to the older socket’s. Builders will obviously need new motherboards to build with Core X-series chips, but they’ll at least be able to reuse their existing cooling hardware.

The X299 chipset might also evoke some deja vu among observant gerbils. That’s because Intel based the X299 PCH on its consumer Kaby Lake chipsets, rather than following in the footsteps of the server-sourced X99 PCH. The infusion of Z270 in the X299 chipset offers several benefits to Intel’s high-end desktop platform, although it breaks little new ground. That said, X299 upgrades the link between the CPU and the chipset to DMI 3.0, which offers roughly the same bandwidth as four lanes of PCIe 3.0. Broadwell-E still relied on the X99 platform’s DMI 2.0 link.

X299 also gives motherboard makers 30 flex I/O lanes to play with. Those flex lanes can be configured as up to 24 PCIe 3.0 lanes, as many as eight SATA 6Gbps ports, and up to 10 USB 3.0 ports. Thanks to its Z270 roots, however, X299 still doesn’t boast onboard USB 3.1 Gen2 support, so expect that connectivity to come by way of external controllers on X299 mobos.

X299 will also support Intel’s Optane Memory and Optane SSD technologies. We doubt many X299 builders will still be considering using hard drives as boot devices, but Intel plans to offer a full-bore 3D Xpoint SSD later this year that seems like a perfect complement for a Skylake-X system. When that storage device arrives, X299 PCs will be ready for it.

 

Pricing and positioning

Many hoped (myself among them) that Ryzen’s arrival would force Intel high-end desktop CPU prices down compared to their historical stickers, and for two chips in the Skylake-X lineup, that’s sort of what’s happened. The Core i7-7820X brings eight cores and 16 threads to the $600 slot for the first time in Intel’s history, while the Core i9-7900X shifts the 10 cores and 20 threads that were previously the domain of the $1650 Core i7-6950X down to a grand. I was disappointed that Intel didn’t choose to add more cores for the same prices as Haswell-E chips when it launched Broadwell-E, so it’s heartening to see the company returning to that practice with Skylake-X. 

Meanwhile, the six-core Core i7-7800X rings in at the same $389 price tag as the Core i7-5820K of Haswell-E fame. Compare that to the $434 one had to pay to get into the six-core Core i7-6800K, and it seems Ryzen CPUs are keeping Intel honest at the low end of the high end, if nothing else.

Of course, Intel couldn’t resist taking away as it gave us more cores for less money. To get 44 lanes of PCIe connectivity from the CPU with Skylake-X, one now has to pay $1000 for the Core i7-7900X. Both the Core i7-7800X and Core i7-7820X have the dreaded 28 lanes of connectivity that used to be the sole domain of the most entry-level Intel HEDT CPUs, a move that will doubtless leave some builders a little sour as they plop down $600 for an i7-7820X.

Even though Intel’s decision to make the full 44 lanes of PCIe 3.0 more exclusive stings here, it might not sour builders as much as it might have in the past. Interest in multi-GPU as a route to higher graphics performance seems to be dwindling by the day, and multi-GPU was the primary reason we dinged Intel’s 28-lane i7-5820K when it first appeared. Times have changed since. A single GTX 1080 Ti is enough graphics card to power all but the most excessive gaming systems, even with a 4K screen hooked up. Assuming one decides to chance SLI’s hit-or-miss game support with twin GTX 1080 Tis, we’re guessing they can afford a $1000 CPU or better to go with them.

The biggest demand for PCIe these days beyond graphics is for storage devices, and the lower-end Skylake-X parts still do well enough here. Motherboard makers could conceivably wire up three full-speed M.2 slots or U.2 ports from the remaining 12 lanes on the i7-7800X and i7-7820X, and that’s before we consider what’s possible with 44 lanes from the higher-end chips. Mobo engineers should also have fun with the freedom of X299’s flex I/O lanes, although devices connected to those lanes will be bottlenecked by the roughly PCIe 3.0 x4 bandwidth of the chipset’s DMI 3.0 link.

What about Core i9 CPUs? Well, more cores mean more money, but the Core i9 lineup actually seems pretty reasonably priced for what it offers, assuming you need more cores. The Core i9-7920X’s 12 cores are a $200 upcharge over the $999 Core i9 7900X’s 10-core, 20-thread complement, and another $200 jump gets you 14 cores and 28 threads on board the Core i9-7940X. From there, the delta changes to $300: first for the $1699 Core i9-7960X, and then to $1999 for the Core i9-7980XE big daddy with its 18 cores and 36 threads.

Even at $1999, the Core i9-7980XE seems like a fair bargain compared to many-core Broadwell Xeons. As just one example (and presuming your X99 motherboard offered support for it), one could get 18 Broadwell cores and 36 threads in the Xeon E5-2695 v4 for $2400 at retail. X299 likely won’t offer server-grade niceties like ECC RAM support, but there’s no denying that it’s making a certain class of compute capability more accessible—and with higher potential performance—than ever before on Intel’s high-end desktop platform.

Intel says Core X-series CPUs will be available “in the coming weeks,” and we’ll be learning more about them soon. For now, however, you can join us in pinching ourselves about the prospect of 18 Skylake-X cores in a high-end desktop PC. Hopefully we can stop dreaming and start testing something like the Core i9-7980XE soon.

Comments closed
    • Chrispy_
    • 2 years ago

    I think this is the first time in a long time that Intel’s naming scheme has gone to ruin.

    6000-series = Skylake
    7000-series = Kaby Lake

    Unless you don’t actually have a product to match AMD, and then you hastily repurpose an old chip and slap a new sticker on it. Ugh, I used to respect Intel’s naming system but it seems they’ve sunk to the same low standards as everyone else and will just be using it to upsell old product now….

      • ImSpartacus
      • 2 years ago

      What’s new?
      [list<] [*<]Sandy Bridge-S used 2000 while Sandy Bridge-E used 3000. [/*<][*<]Ivy Bridge-S used 3000 while Ivy Bridge-E used 4000. [/*<][*<]Haswell-S used 4000 while Haswell-E used 5000. [/*<][*<]Broadwell-S used 5000 while Broadwell-E used 6000. [/*<][*<]Skylake-S used 6000 while Skylake-X used 7000. [/*<][*<]Kaby Lake-S used 7000 and Kaby Lake-X still used 7000 because it is materially identical to Kaby Lake-S. [/*<] [/list<] Aside from the minor weirdness with Kaby Lake-X (which is forgivable given its unusual positioning), Intel has been remarkably consistent. Nothing is ruined as far as I can see.

        • the
        • 2 years ago

        The naming convention in that regards has been consistent but not the product releases. Broadwell-C for the desktop (i7-5775C) was a rather low profile release to be eclipsed by SkyLake-S a months later. Haswell-E was the 5000 series parts that most consumers would encounter.

        It also doesn’t help that Kaby Lake is virtually a Sky Lake refresh. The only real change is 4K video support on the GPU side.

        There is also not going to be a Kaby Lake-E part to match up with Cannon Lake’s 8000 series naming scheme.

      • chuckula
      • 2 years ago

      [quote<]I used to respect Intel's naming system[/quote<] Really? It's been pretty arbitrary (at best) for a long time. I actually like the Core i9 part of it because it clearly distinguishes the socket right up front. The worst part is Kaby Lake X where it's still "i7" with model numbers that are almost indistinguishable from LGA-1151. I'm sure Intel would have caught a lot of rebranding flak for calling Kaby Lake X an "i9" and using more easily distinguishable model numbers, but frankly I'd prefer that to model numbers that are easily confused with incompatible chips.

    • ronch
    • 2 years ago

    AMD Threadripper vs. Intel Walletripper.

    • NovusBogus
    • 2 years ago

    Intel silicon is expensive to research, expensive to make and expensive to sell. Moar coarz is great news but anyone expecting a massive price drop hasn’t been paying attention.

    With those kids of core counts, I bet the i9 series runs pretty slow. Not a problem for those who truly need all the cores, but I anticipate them being even less popular enthusiast choices than the previous iterations.

      • chuckula
      • 2 years ago

      It depends on how you look at it. The 7820X (8 core) and 7900X (10 core) will certainly overclock better than their Broadwell equivalents, which weren’t particularly fast. Of course, if you really want moar hertz they’ll still lose to a quad-core Kaby Lake, but they should be respectable.

      As the core count jumps up, and especially when you go to the higher core-count silicon for the biggest parts, the clock headroom for running a large number of cores at high speeds will certainly fall off quite a bit.

      • ImSpartacus
      • 2 years ago

      With Turbo Boost 3.0, it won’t matter.

      They can pick the best two cores in a given chip and selectively clock those up for single or dual threaded loads.

      I believe that tech scales out to any number of cores. There’s no fundamental reason why it couldn’t.

        • JustAnEngineer
        • 2 years ago

        Consider, too, that Intel is advocating a water-cooling setup for these new chips. They might be able to really ramp up the voltage to those couple of cores with that much cooling available.

      • JustAnEngineer
      • 2 years ago

      Intel’s silicon is cheaper to make than the competition’s silicon. Sure, they’ve sunk billions into research and development and then more into marketing, but their variable costs are quite low.

    • beck2448
    • 2 years ago

    I9s will be great. Love to the 18 core beast in the new Mac Pro!

    • bjm
    • 2 years ago

    Hey Jeff,

    Honest question: Do you find chuckula’s incessant fanboy baiting acceptable?

    I ask because he has only stepped up the aggressiveness in the past few months and is clearly baiting everyone into some fanboy flame war. I don’t take issue with his technical points, in fact I agree with many of them. I feel Intel has the superior product, but his tone has grown tiresome. Its clear to me that he has no interest in discussing technical merits, but rather has an intent focus on drawing AMD fanboys into a flame war. Yes, we all have our biases, but he goes the extra mile with personal attacks and repeated flame baiting. I think he’s getting out of hand.

    Do you?

      • DancinJack
      • 2 years ago

      There is literally no reason to do this in the comments except so others can see how you’re “standing up” to chuckula.

        • bjm
        • 2 years ago

        Just like how chuckula is standing up to AMD fanboys, right? Its the tone that has been set here as of late and its been set by chuckula. He constantly calls out and baits fanboys out of the woodwork, so why can’t I call him out on that?

        • willmore
        • 2 years ago

        Not everyone hangs out in the forums. Many of us prefer the comment threads and rarely if ever go into the forums.

          • DancinJack
          • 2 years ago

          And?

          Yes, god forbid he try and deal with it personally through a PM to Chuckula or to Jeff. Y’know, that capability that every single account has access to for free?

            • bjm
            • 2 years ago

            Why do it that way when chuckula publicly points out his “usual suspects” in nearly every thread when he also has that access to personally PM these “usual suspects”? Does he have some greater reason for doing so other than to “stand up” to these fanboys and bringing them out of the woodwork?

            This has gone on and on for months. Is it acceptable to do that here? Or is it not?

        • cegras
        • 2 years ago

        Of course it does – it gives a handy little vote counter for a collective voicing of agreement.

    • Fonbu
    • 2 years ago

    I wonder how many transistors in that 18core? And recommended water cooling unit hmm.

    • ronch
    • 2 years ago

    I think it needs mentioning that when Bulldozer came out and fell on its face someone at Intel said something like they just don’t consider AMD to be a threat anymore. Like AMD is finished and it’s time to move on. At that time many were also anticipating ARM to be the Next Big Thing from servers to bread toasters.

    So yeah, remember that, Intel guy? 🙂

      • DancinJack
      • 2 years ago

      -Intel reads ronch’s comment.
      -Intel checks their market cap of 169.67B
      -Intel checks AMD’s market cap of 10.68B
      -Intel relaxes

      • bjm
      • 2 years ago

      Sure, they remember, because its still true to today. AMD isn’t a threat. More competitive today than before, yes. A threat? No.

        • ronch
        • 2 years ago

        Sure, AMD won’t drive Intel bankrupt but board members now think of AMD in the shower.

    • wingless
    • 2 years ago

    I can’t wait to buy a $2000 Intel CPU!

    /s

    • abele2017
    • 2 years ago
    • meerkt
    • 2 years ago

    So Intel calls it USB 3.0. Tsk tsk. 3.1 Gen 1 no longer in vogue? 🙂

      • DancinJack
      • 2 years ago

      Man, the industry really f’d that naming scheme up. Who the heck wants to write USB 3.1 Gen 1 when you can just write USB 3? Such a stupid mess that could have been avoided pretty easily.

        • JustAnEngineer
        • 2 years ago

        Engineers name things to make them identifiable. Marketers name things to obscure identification and wring more dollars out of the pockets of gullible consumers.

        • Ninjitsu
        • 2 years ago

        or maybe even, you know, 3.2. 😛

      • ImSpartacus
      • 2 years ago

      Shoulda called it USB 3. Sig figs!

      • tipoo
      • 2 years ago

      I wouldn’t even be mad if they retconned gen1/gen2 and just did USB 3/3.1 again. How moronic the naming had become.

    • sophisticles
    • 2 years ago

    I’m getting old! I find myself unable to get even a little bit excited by these processors. There was a time years ago that the news of a new high end processor would send me to every tech site, looking for any info that the other sites didn’t have, waiting for the benchmarks and wishing I could go build a top of the line system.

    Today, I find myself excited by the relatively cheap lower end technology; I was in a Microcenter 3 days ago, they had the 1600 for $200, with $50 off any compatible motherboard, so a decent $65 Gigabyte MB would cost just $15 and they had 16Gb of GSkill ram for about $60, so for under $300 you could pick up a 6C12T cpu/mb and 16Gb of ram and I still didn’t buy it. Why?

    Because at this moment I find myself waiting to see what Coffee Lake will bring, specifically in the realm of hardware encoders, Kaby Lake already has hardware encoding for VP9, VP8, MJPEG, MPEG-2, H264, HEVC and AAC and they have continuously improved Quick Sync with each generation, in quality, speed and capabilities, so I’m hoping that this trend continues with Coffee Lake.

    Other than that, I’m waiting to for Optane Dimms to become widely available and see what they bring to the desktop, I have a feeling that will be the next revolutionary step.

    These high core count processors, unless you have “penis envy” and just need to brag about how many cores your cpu has, unless you do a lot and I mean a lot of code compilation, or you run a multi-user environment, like some kind of server, you won’t likely get the benefits some are probably hoping for.

    Ironically enough, these processors would be ideal for a type of gaming that those that would build a gaming rig around said processors will probably never do…chess. Chess tends to be one of the most cpu intensive games you can play, if you turn up the analysis algorithm and the search depth. It’s funny, but the people that play chess games would never spend the money on this type of processor and the people that would spend the money tp build a gaming system would never play computer chess.

      • brucek2
      • 2 years ago

      Even then, intensive compute computer chess feels like it should be a cloud service vs. owned hardware. The data to and from the server is super tiny, the structure of the game can accommodate a little (or a lot) of latency beautifully, and if you’re taking it to the max I bet access to A LOT of cores on a tiny fractional basis is much superior to fewer cores on a 100% owned basis.

    • southrncomfortjm
    • 2 years ago

    So, where does Coffee Lake fit into all this? Isn’t that supposed to come out later this year too?

      • VincentHanna
      • 2 years ago

      Coffee lake may come out this year.

      Coffee lake X299/ Extreme edition processors will be refined and held back until 2018, like always. Probably Q4

        • southrncomfortjm
        • 2 years ago

        I just wonder what the Coffee Lake 4 core/8 thread part will look like compared to the 7740K. There may not be a lot of room left for improvement without a process shrink, right?

      • ronch
      • 2 years ago

      Covfefe Lake.

      FTFY.

        • southrncomfortjm
        • 2 years ago

        Yes, yes you did.

        • ImSpartacus
        • 2 years ago

        That’s a fresh meme, never even been frozen.

      • HERETIC
      • 2 years ago

      There was talk of it being brought forward to this year.
      Latest news-back to Feb 2018.
      [url<]http://wccftech.com/intel-coffee-lake-delayed-2018-8th-gen-kaby-lake-refresh/[/url<]

    • Ninjitsu
    • 2 years ago

    I was thinking about the last two chips of the line-up, and why they exist here. Hooking the customer on to X299 makes sense, and well it’s a clearer update path – same socket for two generations, with multiple core counts available.

    But what’s the point for the consumer? Well, I’d guess that the IHS is soldered and that there’s no iGPU, so better overclocking. Apart from that, nothing I can think off.

    [quote<]Assuming one decides to chance SLI's hit-or-miss game support with twin GTX 1080 Tis, we're guessing they can afford a $1000 CPU or better to go with them.[/quote<] I've long thought that this is a non-issue for dual GPU builds, and haven't seen evidence to the contrary. And even according to nvidia itself, more than 2 GPUs is a rarest of rare case. So overall it's a non-issue for GPUs. I'd really like to see proof that 8x+8x PCIe 3.0 is limiting for the GTX1080 Ti, because implying a $1000 CPU is required for dual GPU (even 1080 Tis) sounds a bit much. Wouldn't it be possible to run 16x+8x, anyway? Finally, if you really need more than 16 lanes for PCIe storage, 28 lanes still seems plenty given that there aren't any alternatives from AMD (yet) in that price range.

      • Krogoth
      • 2 years ago

      Actually, there are reports that the new Skylake-X chips have gone with thermal paste + IHS setup from people who managed to get engineering samples and delidded them.

      It is possible that it is only applicable to the “Core” tier stuff while the Xeon-branded stuff might still use solder.

    • WaltC
    • 2 years ago

    Awwww….isn’t that cute! ‘de lil’ baby Intel wants to keep up with the Big Dogs @ AMD…;) cootchy-cootchy-coo…;)

    To run with the big dogs, Intel, you will have to do better than produce massively overheating and overvolted wanna’be’s that require liquid cooling…;)

    Especially troubling to Intel, I’m sure, is the fact that AMD has announced that Zen2 will tape out this year…! IMO, AMD is going to keep Intel scrambling for the foreseeable future…;)

    Intel: “The CPU especially made for the Catch-Up crowd–for those who truly savor nostalgia!”…;) (Sorry–can’t resist!)

      • chuckula
      • 2 years ago

      To run with the Big Dogs at AMD Intel will clearly have to find a way to liberate itself of that quarterly profit it makes on an interminable basis.

      Intel will clearly fail to steal AMD’s innovation in that regard.

        • Mr Bill
        • 2 years ago

        Even though they take a profit for every CPU sold, they will make it up on volume.

      • bjm
      • 2 years ago

      It’s hard to play catchup when you’re leading in all the benchmarks.

    • bjm
    • 2 years ago

    Bring on the core count battles! It’s nice to see competition again and the only battleground really left is thread count. Unfortunately for AMD, Ryzen was an overrated second place winner before this announcement and they may not be even that now. This really isn’t new ground for Intel either, they’ve been producing high core count Xeons since Haswell and they just needed a reason to move out of the server market. AMD has been telegraphing their moves for a long time, so Intel will always be more than ready to quickly counter with products like these.

    • srg86
    • 2 years ago

    Good, competition is bringing more computing power for the price.

    This also kills any interest I had in Ryzen though.

    That said, I’d be more interested in the 6 core Coffee Lake than these HEDT chips.

      • chuckula
      • 2 years ago

      Just wait until AMD threadrips its way back to the top.
      It’s gonna be Epyc.

    • Anovoca
    • 2 years ago

    As a hardware enthusiast this news has got me pretty stoked. As a home owner with a mortgage payment, I am secretly praying for a Pentium refresh.

      • Ifalna
      • 2 years ago

      Yup, I’m looking forward to seeing the benchmarks while still clinging to my trusty old 3570K because I’m a stingy bastard. 😀

    • Bensam123
    • 2 years ago

    LOL still trying to sell 6/12 for $390 when you can buy a 8/16 for $315 from AMD with roughly ~10% less performance in some of todays games. Meanwhile you can find a 6/12 from AMD at $220.

    Not even close.

      • chuckula
      • 2 years ago

      Man, you’ve really gone downhill from [url=https://techreport.com/discussion/24879/intel-core-i7-4770k-and-4950hq-haswell-processors-reviewed?post=735330<]your glory days.[/url<] Oh, and change your underwear.

      • Krogoth
      • 2 years ago

      Skylake-X chips are going to faster, but it appears that Intel is placing massive blind faith in their Core brand hoping that it can justify rather steep premiums on their HEDT line-up.

      i7-7820 and i7-7800 are only SKUs that are remotely sane while i9-7900 and i8-7920 are tough sales. i7-7720-7740 are just fool’s errands and i9-7960 and i9-7980 make the previous the i7-6900-6950 look sane by comparison.

        • Bensam123
        • 2 years ago

        Sure, but the gaming performance that was tested was against something like a 7700k. If you look at the reviews for Ryzen, it was already tested against that. That’s the recommendation the Intel fanbois are making today. Buy a 7700k so you can get 10% better performance in some games (and of course half the cores) for the same price.

        Older architecture is just going to be that much worse and close the gap between the two. It basically comes down to price unless you’re talking data center for some very specific tasks (which this platform wont be used for anyway). In this case there is no question which is a better value.

          • Krogoth
          • 2 years ago

          7700K is a mediocre buy at best right now. 7600 is a far better buy for a gaming build. It almost as fast at gaming and same overclocking ceiling for almost $100 less. That difference could land you a beefier GPU which will yield far greater returns.

          The HT on the 7700K is only useful for workloads and situations where the Ryzen R5 and R7 would normally stomp it.

            • Bensam123
            • 2 years ago

            I agree, but the kids all buy i7s and it’s basically the ‘best case scenario’ to hold up to for gaming performance for Intel. You knock it down a rung to i5 and then you’re going to be comparing 6/12 core processor. However, we aren’t really talking specifically about those buys, rather the launch of Intels new 6/12 and 8/16 core chips, which are insanely over priced. Their performance per core should be the same as a 7700k though.

            • Dr_Gigolo
            • 2 years ago

            [url<]https://www.youtube.com/watch?v=XylVCItVhS4[/url<] Looking at many benchmarks now a days, hyper threading actually boosts quite a few games. Of course, caches might make a difference, but it looks like hyper threading is actually being utilised in games now. Enough so that I finally jumped the gun and am upgrading from a 4670k to 7700k. I would like 6 or 8 cores, but really, I all I use my desktop for these days, is gaming and emulation (gaming), where high frequency is king.

            • Krogoth
            • 2 years ago

            Ryzen R5 and R7 outperform the i5 -7600 and i7-7700 in those situations though.

      • bjm
      • 2 years ago

      Its close if you favor higher IPC. Simple because AMD has more cores doesn’t necessarily mean its the right choice for everyone. For many, paying the extra for Intel to gain 10% more performance is entirely acceptable.

        • Bensam123
        • 2 years ago

        Yet it’s going to get worse, much worse in the future. You’re talking about squeaking out a little extra performance in games made for four core processors. Good for Intel, however newer games and even updates to older engines will make more use of more cores, meaning this advantage may not only dwindle, but disappear completely and go in the opposite direction.

          • K-L-Waster
          • 2 years ago

          In my experience, buying now in anticipation that it will be the right part eventually is at best a crap shoot. Frequently by the time the software you were buying for shows up, the part you bought is most of the way to obsolete.

            • Bensam123
            • 2 years ago

            I guess in this case if the software companies not only endorsing, but talking about supporting it into the future equates to a crap shoot. Higher then quad core processors have been around longer then Zen launch.

          • bjm
          • 2 years ago

          If and when newer games make use of more than four cores, it will benefit Intel CPUs as well. I don’t understand your “go in the opposite direction” comment. Intel has been making 18 and up core CPUs since Haswell. High IPC is not exclusive to a high core count.

          And to add to K-L-Waster’s point, we have had quad core CPUs for more than 10 years now. Games have only now just started to make use of them. Provided that your CPU has enough threads, high IPC will always mean more performance. As it stands today, for gaming, a quad core with a high IPC is the ideal configuration.

            • Bensam123
            • 2 years ago

            Except in order to get the higher core counts you have to pay an insane amount of money, which we’re circling back to the original point. You can’t purchase a 8 core Intel for $315. That’s the argument here. Basing a $315 processor against a $1000 is a fools errand.

            Sure and there haven’t really been accessible higher then eight core processors till now. Bulldozer was essentially a quad core processor and a unique one at that, which made it a niche case scenario, this is just more cores and very similiar per core performance.

            Yes more IPC means more performance, 10% more with half the core count is kind a big deal bro. And as far as games today, you can see a pretty meaty performance boost going from a quad to a hex core in Overwatch. I did it, I went from a 4690k to a 5820k and saw a 30% performance increase (if you aren’t being limit by your GPU). There are already games out that take advantage of this, your information is old.

            • bjm
            • 2 years ago

            10% more with half the core count?

            The article that you’re posting under has an Intel 8-core/16-thread CPU for $599 and 6-core/12-thread for $389. Your information is old. The 5820k that you just mentioned is $409 on NewEgg right now. Again, high IPC is not mutually exclusive to a high core count, nor is a high core count exclusive to AMD. Top of the line/King of the hill gaming performance, however, is exclusive to Intel.

            • Bensam123
            • 2 years ago

            Except you can get a 8c/16t AMD for $315 ($285 difference) and a 6/12 for $220 ($150 difference).

            We can use simple math. Divide $600 by $315. Is there a 90% improvement in performance between those two processors? No?

            Lets do $390 by $220. Is there a 77% improvement between those two processors? No?

            You can STILL get a 8c/16t AMD for less then the cost of Intels 6/12. Once again a $75 difference. That just exasperates the comparison even more because you’re adding MORE cores on top of the price difference, which is what we’re basing everything on seeing as the per core performance difference is pretty close to nill and will get better over time as software moves forward.

            Yup, a 5820k is a old processor and I bought it off eBay for $300 about a year ago. I wouldn’t buy it again today. Options on the market were different at that time and really not even sure what you’re using it for in your paragraph as I mentioned it showcasing the fact that core count DOES indeed matter even right now in todays titles. More so because it was a direct apples to apples comparison 4690k vs 5820k are both Haswell processors.

            • bjm
            • 2 years ago

            We’re just going to have to agree to disagree then, I suppose. I’m not arguing against Ryzen being a great value, because it is. But as it stands today, an Intel 7700k and 6700k is beating Ryzen in just about all gaming benchmarks, despite having 4 cores. In terms of absolute top end gaming performance, Intel is at the top. If you disagree with that, then well.. Let’s just agree to disagree.

            • Bensam123
            • 2 years ago

            This isn’t a argument of opinion. Not only is it a great value, it also will end up hands down superior as multithreaded performance of games improves (in some games this is already the case IE Overwatch).

            I guess you can completely shit the bed and just act like prices don’t matter. And then, yes, you are indeed correct. a 8c/16t Intel for $500 is superior to a $315 8c/16t AMD. 10% more performance for an additional 90% price hike – literally almost 2x the cost.

            I do like how you stipulate ‘even though it has four cores’ as if the extra cores on Ryzen were being fully utilized. You should also stipulate that as games become better at multithreaded workloads, which big game companies have already made clear they’re working on, that 4 core processor is basically going to become the equivalent of buying a two core Pentium today.

            • bjm
            • 2 years ago

            [quote<]And then, yes, you are indeed correct. a 8c/16t Intel for $500 is superior to a $315 8c/16t AMD. 10% more performance for an additional 90% price hike - literally almost 2x the cost.[/quote<] No, I'm saying that a 4c/8t Intel 7700k for $339 is faster than any AMD CPU at any price for [b<]gaming[/b<]. Benchmarks prove it. Do you really dispute that? You can add any and all stipulations you want, but the fact is, Intel has the fastest CPU for gaming today. You don't need to pay a 90% price hike to get it either. Ryzen 7 1800X being outperformed in Overwatch by even a 4790K: [url<]http://www.techspot.com/review/1345-amd-ryzen-7-1800x-1700x/page4.html[/url<]

            • Redocbew
            • 2 years ago

            Oh thank Bob. I’m glad someone posted actual test results. I always want to correct stuff like this but never want to spend the time doing it.

            • the
            • 2 years ago

            Math isn’t your strong point. $500 over $315 is a ~59% increase in price. Certainly more than raw performance gain but far from ‘literally almost 2x the cost’. You also ignore the idea that here could be other benefits for the $500 price tag like higher memory capacity or more PCIe lanes.

    • ltcommander.data
    • 2 years ago

    Any word on whether Skylake-X is treated like the rest of Skylake by Microsoft/Intel and has official Windows 7/8.1 support? Or will Broadwell-E remain the fastest/last CPUs that have official multi-boot support for Windows 7/8.1/10?x

      • synthtel2
      • 2 years ago

      Skylake Win7/8.1 support only goes through July 2018.

        • ltcommander.data
        • 2 years ago

        [quote<]Today Windows 7 is in extended support and support will end for Windows 7 on January 14, 2020, and support for Windows 8.1 will end January 10, 2023. 6th Gen Intel Core devices on Windows 7 and Windows 8.1 will be supported with all applicable security updates until the end of support for Windows 7 and Windows 8.1.[/quote<] [url<]https://blogs.windows.com/business/2016/08/11/updates-to-silicon-support-policy-for-windows/[/url<] I believe Microsoft acquiesced and is now supporting Skylake with security updates through EOL for both Windows 7 and 8.1 rather than terminating support prematurely. It seems like Skylake-X doesn't count as part of the rest of the Skylake family though.

          • synthtel2
          • 2 years ago

          They’ve done clarification on that more recently, and it turns out that OEM SKL systems are fine, but systems we build using them aren’t. See [url=https://techreport.com/news/31741/updates-for-windows-7-and-8-1-on-kaby-and-ryzen-are-now-blocked<]this[/url<] story.

      • Krogoth
      • 2 years ago

      The CPU and platform aren’t going to be officially supported by MS for anything older then Windows 10.

      You can force Windows 7/8/8.1 to run Windows 10 drivers at your own risk (They all share the same codebase). If you are so inclined.

        • Waco
        • 2 years ago

        The problem is that updates aren’t supported, they are actively blocked.

        Has anyone come up with a way to bypass that yet?

        • ltcommander.data
        • 2 years ago

        That’s disappointing. For my work I need a workstation that can natively boot Windows 7, 8.1, and 10 so I guess I will need to look for a Broadwell-E one for my next upgrade.

    • fellix
    • 2 years ago

    The full scatter/gather support is good news for compiling vectorized code. Now developers will have more incentive to jump aggressively on AVX.

      • UberGerbil
      • 2 years ago

      While in theory I agree (and 512bit AVX to boot), in practice these aren’t going to sell in sufficient quantity to move the needle for most software developers. You’re still talking about a completely different codepath and all the testing effort that goes with it; unless and until these ISA extensions move fully into the mainstream CPUs that the fat majority of the market is buying, they’re going to remain the province of specialized applications that can dictate the hardware they run on.

        • smilingcrow
        • 2 years ago

        If you are using Intel’s Libraries the cost and time to add 512bit support should be minimal and Intel have done the heavy lifting for you testing wise.
        I used them years ago when some of them were still free and you could add SSE and SSE2 support with little effort.

    • DragonDaddyBear
    • 2 years ago

    What’s the cost of the platform/motherboard look like? When you factor that in it may cost more than just $100 USD like-for-like platforms.

    • Shobai
    • 2 years ago

    Does it strike anyone else as kind of… schizophrenic, perhaps, for Intel to highlight overclocking for these CPUs when they’ve literally just told i7 7700k owners they shouldn’t overclock their k-series CPUs?

    [edit – I wasn’t clear; thanks for highlighting this, Chuckula]

    Intel told people not to overclock their 7700k CPUs when people complained that these CPUs were spiking into the 90 degree C range in ‘trivial’ tasks, even when not overclocked.

    Given that this [bug?? unintended feature??] occurs with the Kaby Lake CPU, will we see it again with Kaby Lake-X? Will it appear with Skylake-X? Will adding more cores cause thermal throttling in CPUs affected by this feature?

    If it does appear, what’s the point of spruiking that these chips can be overclocked if they behave weirdly even when not overclocked?

      • Shobai
      • 2 years ago

      Perhaps “double minded” is the more appropriate term? [url=https://communities.intel.com/message/471425#471425<]For reference[/url<] If Skylake's issue with spiking core temps is bad with 4 cores, even on CPUs that aren't overclocked, how will adding more cores improve the situation? Has there been any word on whether these will have a soldered IHS? If so, will that be enough of a silver bullet?

      • chuckula
      • 2 years ago

      Where was your righteous anger at AMD in [url=https://techreport.com/news/31974/ryzen-agesa-1-0-0-6-exposes-more-memory-overclocking-options<]this thread that's still on the front page[/url<] where AMD expressly says that your warranty is voided if you install their AGESA software and use it? Oh wait, two different standards, I forgot. Tell ya what sunshine, let's make a little bet: It will be more likely that a 10 core Intel 7900X -- you know that chip from that evil anti-overclocking company Intel -- can get 10 cores up to 4.2GHz than the likelihood of getting 8 cores on your Ryzen 1800X to 4.2GHz.

        • Shobai
        • 2 years ago

        Oh, Mr Cheese-ula, calm down.

        I can see that I didn’t spell it out clearly in my original post; perhaps I should go back and edit it. In the meantime, please peruse the link I posted.

        The only double standards here are the ones in your mind.

        What’s not at issue here is that, for both Intel and AMD, overclocking voids your warranty.

        Similarly, what’s also not at issue here is whether an unreleased CPU can clock its cores in excess of the frequency that another released CPU can clock its cores.

        Finally, what’s further not at issue here is whether or not I personally have a Ryzen 1800X.

        In summary: you’re being childish.

        The situation, as I see it, is that owners of 7700k CPUs are complaining that their CPUs are exhibiting inordinate temperature spikes; the OP of the thread I linked is claiming 25-35 degrees C, others that I have seen referenced claim that their CPUs are spiking into the 90s C. This accounts are from people who have their CPUs at stock, have overclocked, have de-lidded, and various combinations. Intel’s response has been that this is normal behaviour and to tell people who haven’t been overclocking that they should not be overclocking.

        Now we are awaiting the release of Skylake-X and Kaby Lake-X, with increased core counts – can we expect to see the same issues with these CPUs? Will having more cores, which may be liable to temperature spikes like the 7700k, end up soaking the CPU and causing downclocking?

        Does that explain my original post more clearly?

          • Waco
          • 2 years ago

          More cores / larger die area should limit the effect of any particular hotspot. Dark silicon is good for many things, heat dissipation being one of them.

          • chuckula
          • 2 years ago

          You don’t have a point, you didn’t even believe the bravo-sierra copy-n-paste that you put in your original post you disingenous shill.

          Name one worhtwhile AMD product that overclocks to 5 GHz.

          Now answer this: Who is selling more chips that overclock to 5GHz this year? Yeah, that’s answered too.

          Now answer this: Who is selling any chips whatsoever with more than 4 cores that stand a real chance of overclocking past 4 GHz. Yeah, that’s answered too.

          Now that that’s out of the way: STFU when you make disingenous posts for the sockpuppet crowd that we both know you don’t believe.

            • Shobai
            • 2 years ago

            Calm down, chuckula, you’re ranting like an idiot. You’re so quick to cry “wolf!”/shill, it seems like you’re the pot calling the kettle black.

            • chuckula
            • 2 years ago

            You literally don’t believe what you are posting, and I frankly don’t care how many accounts you made to thumb yourself up.

            • Shobai
            • 2 years ago

            The very fact that you would accuse me of such behaviour ‘literally’ shows how little you know about any aspect of me, let alone what I do or do not believe.

            Your behaviour is tiresome. You’re welcome to the last word, your ego seems to require it.

            • chuckula
            • 2 years ago

            Your shilling and disingenuous horespucky is vastly more tiresome.

            Time to put up or shutup.

            If TR, whose founder is an AMD employee, can’t overclock an 8 core RyZen 1800X, which ought to be a cinch, more than it can overclock a ten-core 7900X, which ought to be damn near impossible, then you stop plagiarizing random forum posts and finally admit that maybe, just maybe, you’ve been a two-bit shill the whole time.

            • Waco
            • 2 years ago

            Dude, calm down.

      • nerdrage
      • 2 years ago

      [quote<]Does it strike anyone else as kind of... schizophrenic, perhaps, for Intel to highlight overclocking for these CPUs when they've literally just told i7 7700k owners they shouldn't overclock their k-series CPUs?[/quote<] Yes, perhaps, but I think that's more likely due to the former being from the Sales department, while the latter is from the Support department.

        • Shobai
        • 2 years ago

        I agree that that’s the likely cause, I guess I’m just hoping for some sort of utopic vision of a tech company, any tech company, that can marry the two groups for internal consistency.

        Interestingly enough, [url=http://www.gamersnexus.net/news-pc/2936-intel-i9-7900x-delidding-cpu-package-thermal-paste<]Der8auer's at it again[/url<] with delidding and has done a live i9-7900X delid using his new tool prototype - some interesting finds there, if anyone cares to take a peek.

    • tipoo
    • 2 years ago

    15 watt quad cores at last is pretty exciting. Could move down to a 13″ laptop for that.

    A 13″ quad core rMBP with Thunderbolt 3 and an external GPU could be pretty appealing. For the Windows boot camp side of course.

    • chuckula
    • 2 years ago

    Looks like The Threadripper won’t go over $1000 then. We’ll see how a 60% core advantage for Ripper ’98* works out over a 7900X. I’d expect it to be OK in the right kinds of parallel processes that don’t need inter-core communication, but I wouldn’t underestimate what the real Skylake architecture is capable of doing.

    *The R9 1998X, guaranteed to be faster, and more efficient with better access to the internet!

    [OK AMDbots, I take it by the downthumbs you think a 16 coar Threadripper is coming in substantially north of $1000? In that case: Let me be the first person to thank Intel for coming up with actually useful many-core products since AMD’s offerings are — according to you — suspiciously priced.]

      • kuraegomon
      • 2 years ago

      Don’t forget that doubling the memory bandwidth by channel (and likely also having more multipliers out of the box than Ryzen had) will definitely help ThreadRipper’s performance-per-clock. We just don’t know how big that delta will be. If it’s 3-5% that’s probably not enough. 5-8% might be. My money’s on the former – but my hopes are firmly for the latter 😉

        • ImSpartacus
        • 2 years ago

        Do we know how that affects the CCX interconnect (or the die interconnect for that matter)?

        I thought the CCX interconnect ram at a frequency proportional to that of your memory controller (i.e. RAM).

        All else being equal, it’s harder to run four DIMMs at a given speed than it is to run two DIMMs at that same speed.

        Therefore, in general, I’d expect Threadripper’s CCX interconnect to run slower than that of Summit Ridge.

        Obviously you can overcome that with other changes and I bet AMD did. But by itself, I’m not so sure quad channel is the best thing ever.

          • kuraegomon
          • 2 years ago

          AMD’s already made multiple updates to improve Ryzen’s ability to hit higher multipliers with a broader range of DIMMs. I think it’s a reasonable bet to assume that Threadripper’s interconnect will run as fast as Ryzen’s does now. After all, Threadripper is being sold as an [i<]HEDT[/i<] part. Given the higher platform price point that allows them to hit, I like their chances of being able to deliver along the lines I've predicted. I'll manfully take my medicine here if they fail to do so 🙂

            • ImSpartacus
            • 2 years ago

            For a 4-channel config like Threadripper to consistently hit the same memory frequencies as a 2-channel config like Summit Ridge, it has to have some special sauce that Summit Ridge doesn’t have.

            So far, I haven’t heard of any Threadripper-specific features that will help in this area.

            That’s not to say Ryzen hasn’t improved in this area, across the board. It has, as you pointed out. However, improvements across the board still leave Summit Ridge in an inherently better position than Threadripper.

      • synthtel2
      • 2 years ago

      It’s funny how often you have a neutrally-voted post I find upvote-worthy, then you edit it to add a rant in response to a downvote or two that I never saw, then I find it downvote-worthy and apparently others do too.

    • chuckula
    • 2 years ago

    Lots to unpack but it looks like most of the pre-launch rumors about the microarchitecture panned out as expected, especially including the revised cache being about 15% smaller on a core-for-core basis compared to Broadwell-E.

    There are clearly three pieces of silicon in this launch, first the Kaby Lake-X parts we’ve already seen, then the small core count Skylake-X which probably goes to 10 cores (maybe 12, we’ll see). Then there are the heavy-duty parts that get the medium core-count silicon all the way up to 18 cores.

    [Edit: Based on Anand’s article it looks like up to 12 cores is the low-core-count silicon and 14 – 18 cores is the “high” core count silicon that actually has 20 cores in the silicon. It’s “high” for LGA-2066 anyway, the highest core counts are of course on LGA-3647.]

    • blastdoor
    • 2 years ago

    Amazing prices for the 6 and 8 core parts. Thanks AMD!

    I had imagined Intel would sacrifice some market share to defend margins. I guess not!

      • chuckula
      • 2 years ago

      I don’t see what you are complaining about. Intel just announced the chips for sale in 2017 that will power the Mac Pro in 2018.

        • ImSpartacus
        • 2 years ago

        Oh shit, he’s calling it now.

        No Zen in the new Mac Pro.

          • chuckula
          • 2 years ago

          Nothing personal against The Threadripper, but there is no way Apple is ditching a single platform that can scale from 6 to 18 cores with a truly new microarchitecture for their high-end solution.

            • ImSpartacus
            • 2 years ago

            I know – it was a long shot for poor AMD.

            Intel is a known quantity for Apple and they have a pretty stellar track record in this space.

            • tipoo
            • 2 years ago

            The Mac Pro is probably also where half the AVX width will matter more, rather than general consumer parts.

            My prediction: No Zen in Mac Pro, but a Zen iMac instead. Possibly a semi custom APU. Pipe dream: the same semicustom APU update to the Mini.

            • derFunkenstein
            • 2 years ago

            Apple isn’t going to change anything it’s done for the last 11+ years. It’ll be Intel all the way.

            • w76
            • 2 years ago

            Tell that to Imagination Technologies. It’s business, not marriage, and even half of marriages fail. That said, I agree they’re likely to keep Intel until they roll their own CPU, but that’s just terrible business logic. If that’s really Apple’s logic, then they’re become totally calcified as a business with Jobs’ departure.

            • tipoo
            • 2 years ago

            Intel is actually the exception and not the rule with Apple, they like to flip flop on partners, likely to push deals (maybe secondarily for market health, creating a monopoly against themselves wouldn’t be wise). They stayed on Intel so long because there was no viable competition that wouldn’t be a regression in perf/watt since the Core Duo era, which is when they hopped on.

            The iMac is where I can see AMD making a play, not battery constrained, not massively behind in single core, and more core-iness, possibly an APU to not have a separate GPU taking space.

            • blastdoor
            • 2 years ago

            I wonder what the thermal constraints are for the iMac. Right now the max configuration is a 4 GHz 4 core skylake and a Radeon R9 M395x. How much headroom does Apple have? I suspect not too much — certainly not enough to put a real HEDT CPU in there. I’ll guess that maybe they can squeeze a 6 core coffee lake in there, but nothing more.

            • tipoo
            • 2 years ago

            I gather the 2014 5K iMac hit thermal limits while the 2015 did not, Apple isn’t a stranger to running right up against (un)comfortable thermal capacity. I guess they have at least 20% headroom in the current model? That’s currently a 125W GPU and a 91W CPU, so they could play a bit.

            Also wouldn’t mind them redesigning the thermal plumbing with what else they’ve learned, seems about time.

            • iMacmatician
            • 2 years ago

            Apple [url=https://techcrunch.com/2017/04/04/apple-pushes-the-reset-button-on-the-mac-pro/<]stated last month[/url<] that the iMac will get updates that benefit certain pro users. Shortly afterwards, a [url=https://pikeralpha.wordpress.com/2017/04/05/imac181-with-xeon-e3-1280-v6-processor/<]rumor from Pike's Universum[/url<] claimed that the iMac will get an update this October with the Xeon E3-1285 v6 as the highest end CPU. I think that's unusually specific for a rumor six months out, especially when the E3-1285 v6 isn't even released (yet). (The URL says "1280," which is released, but that's still a specific CPU.) The E3-1280 v6 [url=http://ark.intel.com/products/97475/Intel-Xeon-Processor-E3-1280-v6-8M-Cache-3_90-GHz<]has 4 cores (and a 72 W TDP)[/url<], and I assume that any E3-1285 v6 will also have 4 cores. From these factors, I have a suspicion that the above rumor might be a controlled leak from Apple as a way to pour cold water on any speculation that an "iMac Pro" will have a large(r) number of cores, at least this year.

            • JustAnEngineer
            • 2 years ago

            Has Apple done any innovating since 2011?

            • Shobai
            • 2 years ago

            Does being ‘courageous’ count?

            • Ninjitsu
            • 2 years ago

            Sounds unlike Apple two have to CPU vendors at the same time…

            • chuckula
            • 2 years ago

            Apple has used two different fabs to make (almost) identical versions of the same chip, but two radically different CPU types in what are very similar products (i.e. they are both Macs, we aren’t talking about the iPhone/Mac divide here) would be more unusual.

            • JustAnEngineer
            • 2 years ago

            Are you sure about that, Chuck? Your anti-AMD crusade seems to be very personal to you. Did Jerry Sanders run over your dog?

            • chuckula
            • 2 years ago

            I have never once made any comment that could be considered to be an anti-AMD “crusade”.

            I’ve made numerous completely correct comments on here against the AMD fan squad who literally can’t say anything nice about Intel — ironic since Intel clearly did more than 95% of the original designs that ended up being put into RyZen — while giving AMD a free pass on literally exactly the same issues that they attack Intel for.

            Just look at this thread: I was 100% factually correct in calling that Apple was going to be an enthusiastic adopter of Skylake X and in fact, for being such an “AMD hater” it turns out my predictions were [b<]pessimistic[/b<] about Apple's adoption since we now know that 18 core Skylake X parts are going into freakin' iMacs. So what is the response for being right about this thing called "reality"? That I'm on a "crusade" against AMD. You know who else is apparently on this crusade: reality.

        • blastdoor
        • 2 years ago

        I didn’t know that I was complaining, so I guess I don’t see what I’m complaining about either.

      • ImSpartacus
      • 2 years ago

      To be fair, Haswell-E launched the 6C 5820K at the same $389 price in 2014. There was no Zen encouragement then.

      It just looks better because Broadwell-E actually increased prices across the board for some strange reason.

      • chuckula
      • 2 years ago

      OK! I ADMIT WHEN I’M WRONG!!

      I said that Intel had announced the chips in 2017 that were going to be in Macs in 2018.

      I should have said: Intel had announced the chips in 2017 that were going to be in Macs in 2017. Including the iMac, not just the Mac Pro.

    • Krogoth
    • 2 years ago

    i7-7820 and 7800 are the only SKUs that make sense. The i9-7900 and 7920 are tough sells even with 10/12 cores at their disposal. The other SKUs are either a waste of time (“Ivy-Bridge X” abortions) or you might as get a real Xeon (I doubt high-core count Skylake-X units will overclock well).

    IMO, I would wait until Threadrippers come out and hopefully bring some competition and price pressure on Intel’s line-up.

    • Major-Failure
    • 2 years ago

    What an exciting time to build a new system, regardless of whether it’s Intel or AMD.

    • USAFTW
    • 2 years ago

    Hmm, I can’t quite put my finger on what caused Intel to drop the price of the 10/20 and 8/16 variants from ~1700$/1000$ to 1000$/600$. I guess some mysteries are meant to remain unknown.
    Edit: Ahh, apparently they’ve done so by crippling the PCI-E lanes down to 28.

      • ImSpartacus
      • 2 years ago

      Remember that Broadwell-E was painfully overpriced anyway. Prices actually INCREASED compared to Haswell-E (which was a hell of a deal in hindsight).

        • Krogoth
        • 2 years ago

        Prices did increased again for higher-end Skylake-X SKUs (beyond i9-7920).

          • ImSpartacus
          • 2 years ago

          There’s no meaningful precedent for an HEDT beyond the 7920X. You can’t say there’s an increase when there’s no “before” to compare to.

            • Krogoth
            • 2 years ago

            When compared to its Broadwell-E predecessors. The top of the line SKU debuted at a princely $1704 while the upcoming i9-7980XE is jumping towards $1999 line.

            • ImSpartacus
            • 2 years ago

            Oh cmon, Skylake-X’s “top of the line” isn’t comparable to that of Broadwell-E. You’re getting silly.

            We’re talking about a lineup of 9 parts topping out at 18C versus a lineup of 4 parts topping out at 10C. The scope of those lineups isn’t even close to being comparable.

            • Krogoth
            • 2 years ago

            Yes, it is.

            It is the fastest non-Xeon tier SKU that you can get. Intel has been pushing the price points in HEDT market ever since Sandy Bridge-E took over. The line-up use go from ~$999-$299 but Intel kept pushing the line-up higher and higher with every new generation due to the lack of competition in HEDT market. It is now hovering back at early to mid 1990s levels.

            Hopefully, Threadripper will bring in some much needed competition and price pressure into this segment.

            • Waco
            • 2 years ago

            I don’t understand your downvotes. What you’ve said here is 100% true – the top end Intel SKU has been rising in price steadily for the past decade…

            • Krogoth
            • 2 years ago

            It is because Intel fanboys can’t face the cold truth. Intel has been milking the HEDT line-up ever since they took over with Sandy Bridge-E.

            They don’t remember or don’t know that Intel high-end desktop SKUs used to go well over $1,000 to $2,000 back when Pentium Pro and Pentium II were kings.

            • chuckula
            • 2 years ago

            So what you are saying is that adjusted for inflation Intel has massively slashed its prices for high-end parts?

            Incidentally, once upon a time TR reviewed an AMD FX-62 with 2 cores that goes for over $1200 in 2017 money: [url<]https://techreport.com/review/10073/amd-socket-am2-processors/2[/url<] You know what's interesting about that part, other than the fact that Intel will sell you a 12-core part for less money? Literally the only difference between the FX-62 and parts that cost less than half as much was a few hundred megahertz of clockspeed. They didn't even throw an extra core at you for the effort.

            • Krogoth
            • 2 years ago

            Moving goalposts mate?

            The point is that heated competition is good for the customer, otherwise dominant company will just milk the market if there’s no competition.

            • K-L-Waster
            • 2 years ago

            Supply and demand. Charge what the market will bear.

            And of course, competition tends to change the second part of that equation.

            • ImSpartacus
            • 2 years ago

            Does anyone honestly not believe that Intel’s HEDT lineup is a terrible value?

            I mean, I’ve never seen anyone say that the HEDT lineup is a good value. It’s overwhelmingly a horrific deal for the money. This is effectively universally agreed upon.

            • synthtel2
            • 2 years ago

            It seems like the confusion here is whether we should be using the top of the HEDT line or the number of cores as the reference point. I’m more on the number of cores side – these “new” parts have had equivalents for a long time, Intel just hasn’t seen fit to sell them with core-i branding. That makes the top of the HEDT line pretty arbitrary, where core count is at least a reference point with a technical basis.

            I have no idea what prices equivalent Xeons sell for, but it seems relevant.

            • Waco
            • 2 years ago

            The top-end 1-2 socket Xeons list far higher at the top end than any of these chips.

            • synthtel2
            • 2 years ago

            Not those, but 8/10C Xeons to equalize core count. It looks like E5v4 prices bracket those of HEDT at the same number of cores, but the cheaper parts are pretty low-clocked, leaving HEDT looking like a pretty good perf/$ buy. IOW, extending HEDT to higher core counts does actually mean something for improved perf/$.

            • Waco
            • 2 years ago

            Keep in mind the list prices are rarely realized for Xeons. I routinely pay about half the list price for modern buys. They do absolutely get more worthwhile at the top end because of it.

            • synthtel2
            • 2 years ago

            If there’s no big price discontinuity between HEDT and Xeons, then I’m back to not seeing why it should matter where the HEDT lineup ends.

            • Waco
            • 2 years ago

            Oh, I see what you’re saying. Forgive my lack of coffee.

            • synthtel2
            • 2 years ago

            No worries.

            • smilingcrow
            • 2 years ago

            Are you referring to general retail pricing, eBay pricing or Engineering Samples?

            • Waco
            • 2 years ago

            MSRP versus delivered price from major vendors.

            • smilingcrow
            • 2 years ago

            Is that on a consistent basis or more ad hoc?

            • Waco
            • 2 years ago

            It’s been true for the last 3-4 years when I’ve been ordering servers. The E7 line is the exception, but most of the lower E3/E5 builds I’ve bought have all been fairly heavily discounted CPU-wise.

            • smilingcrow
            • 2 years ago

            Are you talking complete systems or lone CPUs? If the latter where do you recommned?

            • Waco
            • 2 years ago

            Complete systems.

            • smilingcrow
            • 2 years ago

            Thanks for the clarification.
            I have seen stunning deals for 2P Dell systems from the Outlet.
            You could just harvest the CPUs, memory and storage and throw the rest away and still save a fortune.
            But if you use them as a system they are ridiculously cheap especially when you consider that they usually have a three year next business on site warranty included.

            • smilingcrow
            • 2 years ago

            Skylake E actually beings the pricing more back in line with what it has been since Ivy Bridge where you get an extra 2 cores at $1,000.
            This after the price gouging of Broadwell E, so it’s a good thing but those extra 2 cores are less significant now as a percentage increase as clearly the jump from 6C to 8C is larger than jump from 10C to 12C.
            On top of that they’ve extended the range into new territory by taking the core count much higher than before.
            How you view all this is down to your bias as much as anything.
            If you look at Cores per Dollar then Skylake X is clearly back on track as it’s better value than all previous platforms at the $1,000 mark which is the traditional price point.
            But if you look at the highest price SKU the value isn’t so good which is what you expect at the top end but it’s still cheaper per Core than all previous top end HEDT CPUs.

            I7 4960X Ivy Bridge 6C $1000
            i7 5960X Haswell 8C $1000
            i7 6950X Broadwell 10C $1700
            i7 7900X Skylake 10C $1000
            i7 7920X Skylake 12C $1200

            So to get the extra 2 cores over top end Broadwell requires 12C which is $200 over the $1K price point.
            So it’s still an increase but compared to the obscenity that was Broadwell it’s passable.

      • ImSpartacus
      • 2 years ago

      Intel was just feeling generous.

      Intel is widely known to be a merciful and generous company.

        • chuckula
        • 2 years ago

        Just like Emperor Ming is known as the merciful.

    • raddude9
    • 2 years ago

    I find it interesting when competing companies do exactly the opposite thing in a push for market share.
    In this case I mean, Intel are putting desktop chips on a HEDT platform (the 4C8T i7-7740X and the 4C4T i5-7640X). Whereas AMD put HEDT-style 8C16T chips in the desktop AM4 platform. Both approaches have their merits, and it’s a real win for consumers to have the choice.

    • Aquilino
    • 2 years ago

    Intel doesn’t get tired of effin’ up people with the PCI lanes.
    No, thanks. Not this time.

      • thx1138r
      • 2 years ago

      Making-do with less PCIe lanes seemed reasonable ever since dual-GPU setups went the way of the Dodo, but it seems like a retrograde step these days with NVMe drives getting more popular.
      A basic setup with one GPU and two NVMe drives and you’re almost out of lanes, certainly, further expansion would have to be thought through very carefully.

        • Krogoth
        • 2 years ago

        Thunderbolt 3 and other upcoming ultra-high bandwidth peripherals are putting a challenge to that.

          • Bauxite
          • 2 years ago

          Thunderbolt needs lanes, they don’t come out of thin air lol. Even when they bake TB into the ‘soc’ or whatever it is going to come from the cpu root complex. Typical thunderbolt designs are lazy and use PCH lanes which are yet another bottleneck.

          The current misery rate of growth is a big problem: 16 lanes forevvvvvvver and +4 lanes after 5 years on >$1000 cpus only. They even regressed from 40 to 28 after haswell>broadwell on the same damn socket “just because”.

          Intel can pound sand.

        • the
        • 2 years ago

        Dual GPU setups are still around from both AMD and nVidia. What has gotten nerfed is that nVidia dropped triple and quad SLI support from their consumer cards (ironcically, the high end Telsa and Quadros can scale to 8-way now using nvLink).

        AMD still supports triple and quad Crossfire but since the R9 290X, there hasn’t been much reason to go in this direction.

      • chuckula
      • 2 years ago

      For $389 I’ll take 28 PCIe lanes on a 7800X over $500 for an 1800X with only 16 lanes any day of the week.

        • kuraegomon
        • 2 years ago

        That’s a fair point. There’s a decent chance that the 1800X is going to have to drop its price a bit once the 7800X ships.

        Of course, the counter-argument is that the [i<]platform[/i<] cost of 7800X vs 1800X - i.e. CPU + MB - is going to be virtually identical. And there are still parallel workloads where the 1800X might win. Overall though, 1800X is probably going to have to drop in price.

          • Beahmont
          • 2 years ago

          The problem is AMD loses on a Draw because they F’d around with Bulldozer for most of a decade and lost all credibility with most all the people that buy these things. There’s also the problem that none of the IT people want to deal with NUMA or Platform complaints from ID10T’s End Users who don’t know why their special toy doesn’t ‘work-right-but-they-know-it’s-your-fault’ on a Draw between platforms. It’s just not worth the headaches. Total Cost to Own is still higher for AMD than Intel right now when price and performance are a draw.

          Edit: Fixed spelling errors due to PLBSC errors caused by a migraine.

        • gerryg
        • 2 years ago

        7800X – 1800X = 6000. Intel is 6000 better, that’s a lot! If you want more “X”, there’s clearly only one choice.

    • NTMBK
    • 2 years ago

    I’m a bit worried about that new cache hierarchy. Exclusive caches, L2 roughly the same size as L3? Sounds an awful lot like Bulldozer…

      • the
      • 2 years ago

      But it has the Sky Lake front end. Bulldozer’s weakness was sharing decoders across two cores.

      Word is still out on cache latencies with Sky Lake-X but I’d bet that they’re better than Bulldozer too.

        • NTMBK
        • 2 years ago

        That was one of Bulldozer’s weaknesses, sure 😛 But poor cache design was another.

          • Ninjitsu
          • 2 years ago

          Devil’s in the details though.

      • maroon1
      • 2 years ago

      The low performance of bulldozer has noting to do with the size of L2 cache

        • NTMBK
        • 2 years ago

        Bigger cache generally means higher latency, and cache latency was one of the issues with BD.

          • Waco
          • 2 years ago

          Bigger cache *can* mean higher latency, but it isn’t strictly correlated.

    • ronch
    • 2 years ago

    Sorry, still Ryzen for me.

      • kuraegomon
      • 2 years ago

      I’m waiting for ThreadRipper personally, before deciding which vendor to choose for my next VM workstation build. I think that Quad-channel DDR4 (presumably with at least DDR4-3200 support) will really let the Ryzen architecture breathe. After that it’s all down to price-point, likely vs the i9-7900.

      That said, I’ll definitely admit that I like the cache architecture change that Intel has made with Skylake-X. I look forward to the benchmarks vs. ThreadRipper with _great_ interest 🙂

        • ronch
        • 2 years ago

        AMD will almost certainly still give Intel a run for its money, especially given the MCM nature of their Threadripper. That could help them be more cost effective. And given the much higher margins up there they will have lots more wiggle room.

      • USAFTW
      • 2 years ago

      Good man/woman.

      • DancinJack
      • 2 years ago

      We’re shocked, ronch…

        • ronch
        • 2 years ago

        I wonder how you’d feel if I buy Intel. ^_^

          • DancinJack
          • 2 years ago

          I’d be shocked.

            • ronch
            • 2 years ago

            Guess you’re in for a shock one way or the other. 🙂

    • DancinJack
    • 2 years ago

    They REALLY need a price point between the 7800 and 7820. 390 to 600 is quite a gap. Missing quite a bit of opportunity there IMO, especially considering Ryzen 8/16 is sitting at 500.

    Some cool stuff here though. 512b AVX + the new cache scheme should be fun.

      • jihadjoe
      • 2 years ago

      i7-7820X is gonna eat up R7-1800X. 8 higher IPC cores and turbo to 4.3/4.5GHz for $100 more seems like a no-brainer to me.

      It’s the i7-7800 and below that will be facing tougher opposition from the R7-1700 and R7-1700X, IMO.

      Edit: R9->R7. Thanks thx!

        • thx1138r
        • 2 years ago

        R9-1800X? you meant R7-1800x yes? It was rumored that the ThreadRipper series would have the R9 moniker, but that remains to be seen.

        Also, in fairness to the 1800x, the motherboard will also likely be $100 cheaper as well (for a total saving of $200 or so), and home users will be saving even more money overclocking their 1700’s.

        • ImSpartacus
        • 2 years ago

        I think the “true” HEDT starts at the 7820X.

        Everything under it is just for bragging rights. It’s a poor value.

        Coffee Lake-S will be out in a few months and it’s sure to bring 6C and 7700K-like clocks to the <$400 space.

        • Ninjitsu
        • 2 years ago

        not just that, quad channel and 28 PCIe lanes too.

      • ImSpartacus
      • 2 years ago

      Intel always does that.

      It’s supposed to tempt you to buy up to the $600 part.

      This isn’t Intel’s first rodeo. They know how to extract dollars from their customers.

        • JustAnEngineer
        • 2 years ago

        It’s too bad for Intel and good for consumers that AMD brought renewed competition in the CPU space, otherwise Intel could have continued to artificially segment the market and limit the pace of advancement to maximize profits.

        • DancinJack
        • 2 years ago

        I know what they are doing and did. What I am saying is that I believe a 500 dollar price point would have made them more money in the long run. There are surely people that just won’t jump to the 600 price point.

          • ImSpartacus
          • 2 years ago

          I don’t agree.

          MAYBE you cut prices by about $100 across the board after Threadripper releases (or when the 7800X loses relevance due to the release of coffee lake and the 7820X must come down a little to help out).

          But initially, you soak up that early adopter price discrimination if you’re Intel.

          This is textbook market segmentation.

            • DancinJack
            • 2 years ago

            Intel hasn’t had a 500 dollar competitor in the past decade either. This isn’t 2013 where Intel is the only game in town anymore.

      • Topinio
      • 2 years ago

      32 DP FLOPS per core per clock is very nice, but I just want Intel to publish the AVX clocks on ark.intel.com … is that unreasonable?

      • maroon1
      • 2 years ago

      But i7 6900K is already nearly on par with 1800X in productivity but better in gaming, and overclock easier to over 4GHz

      7820X has 400Mhz higher base clock than 6900K, higher IPC (512b AVX + the new cache might even make it higher than regular skylake) and that should make it easily better than 1800X in most productivity tasks

      You also still get more PCI-E (28 vs 16)

        • DancinJack
        • 2 years ago

        I’m not sure what you’re trying to show me?

        • kuraegomon
        • 2 years ago

        That’s great. But without mentioning the prices of the CPUs you’re comparing, you’re completely avoiding performance [i<]per-dollar[/i<] comparisons. And those matter - to a lot of (likely most) people.

        • gerryg
        • 2 years ago

        Encoding for alphabetic letter values, where A=1, B=2, … we see that the i in i7 is a 9, and the r in R7 is an 18. And the k in 6900K is an 11, whereas the x in 1800X is a 24. 18 > 9 and 24 > 11, so AMD is clearly the winner. ‘Nuff said.

        Edit: no sense of humor around here… see my “alternate” post above about Intel “winning”, also said in jest. sheesh

      • jihadjoe
      • 2 years ago

      I think this is one example where Intel’s process tech is too good for its own good.

      One of the hallmarks of Intel’s HEDT platform is everything is overclockable, so chips are distinguished with cores and PCIe lanes and not by clocks. If they were to make another 8/16 chip and price it below the 7820, it’d just end up killing 7820 sales.

Pin It on Pinterest

Share This