Core i9-9900K takes eight cores, 16 threads, and solder to 300-series mobos

At its Performance Unleashed event this morning, Intel announced three new product families. For most PC builders and enthusiasts, the trio of ninth-generation CPUs for Intel's 300-series LGA 1151 platforms will likely be of the greatest interest.

The long-rumored and long-leaked Core i9-9900K is Intel's first 5-GHz CPU to ship in volume, and it'll bring eight Skylake cores and 16 threads to Z370 and Z390 motherboards. The ninth generation of Core desktop CPUs also marks the long-requested return of solder as the thermal interface material (or TIM) beneath those chips' integrated heat spreaders (IHS), enhancing heat transfer from the chip to the heatsink above. That solder should let builders keep overclocked ninth-gen systems both cool and quiet without resorting to delidding and repasting.

Base

clock

speed

(GHz)

Peak

single-core

boost

speed

(GHz)

Cores/

threads

TDP L3

cache

Memory

support

Suggested

price

i9-9900K 3.6 5.0 8/16 95 W 16 MB Dual-channel

DDR4-2666

$488
i7-9700K 3.6 4.9 8/8 12 MB $374
i5-9600K 3.7 4.6 6/6 9 MB $262

To get that combination of peak single-core performance, multi-threaded grunt, and overclocking prowess, however, builders will need to pony up. At $488 for the i9-9900K, Intel is giving builders what they want, but it's clearly not attempting to engage AMD on core count for the dollar. Instead, thanks to Skylake's continued superiority in performance and clock speed, the company is positioning the i9-9900K at the top of the heap for eight-core, 16-thread chips.

Within Intel's product line, the i9-9900K comes in at $100 less than the Skylake-X i7-7820X commanded at launch. Even with half the memory channels of that part and no AVX-512 support, it's hard to complain about the company making eight of its best client cores available on mainstream motherboards for what may be the lowest price ever for such a complement. Intel also let slip that it expects an all-core Turbo Boost speed of 4.7 GHz from the i9-9900K. No matter how you slice it, this chip will be formidable. We'll be testing the i9-9900K and will be able to say just how formidable it is soon.

Moving down the stack, the i7-9700K offers eight cores and eight threads running at a 4.9-GHz peak single-core speed. It has the same soldered IHS as its i9 sibling. This is the first Core i7 desktop CPU we've seen without Hyper-Threading in years, and it'll be interesting to see just how that omission affects performance. Intel doesn't seem to expect the drop in overall thread count to harm the 9700K versus the 8700K too much, though, as it's pricing the ninth-gen part at only a small premium versus the i7-8700K's $359-$370 range.

At the base of the ninth-gen stack, the i5-9600K offers six cores and six threads with a 4.6-GHz single-core peak speed. A 300-MHz peak single-core boost and a soldered IHS will be the primary benefits buyers will enjoy versus the i5-8600K, and they only add $5 to the 9600K's suggested price versus its predecessor. We expect buyers who care more about peak speed and overclocking headroom than absolute multithreaded grunt will find a lot to like in the i5-9600K, even versus AMD's Ryzen 5 chips.

All ninth-gen processors will work in existing Intel 300-series motherboards, as well as just-introduced Z390 boards. Stay tuned for our review of Intel's latest and greatest.

Comments closed
    • End User
    • 1 year ago

    Gamers Nexus followed up directly with Principled Technologies regarding the paid for by Intel report. Definitely worth a read/watch:

    [url<]https://www.gamersnexus.net/industry/3374-principled-technologies-interview-intel-testing-concerns[/url<]

    • End User
    • 1 year ago

    Principled Technologies (the name itself is beyond ridiculous) used 1920×1080 for all their benchmarks.

    It is laughable to isolate testing to 1920×1080. 1920×1080 clearly favours Intel.

    Principled did not even bother to use a proper gaming display. They used a Dell Ultra HD 4K Monitor P2715Q. So ya, they could have run the benchmarks at 4K but the could not be bothered to do so (and yes, the GPU becomes the bottleneck at 4K, but that is my point).

    Hey Intel. Frack off!

      • Voldenuit
      • 1 year ago

      A 4K test would only show that CPU choice is (mostly) irrelevant if you’re GPU bound.

      But we all knew that already, anyway.

      Now a more realistic ‘real world’ test, and something TR does, is gaming performance while CPU encoding and streaming on the same box.

        • End User
        • 1 year ago

        I agree, as long as tests are performed @ 2560×1440 and above

    • Convert
    • 1 year ago

    I’ve been waiting for this!

    My lab systems are always spec’ed for higher core count, within reason.

    I typically go between releases like this, where Intel offers higher core count chips on consumer boards, and the HEDT offerings.

    This release lets me switch my default x299 platform with one of these. I save some money on the chip, plus I don’t need a GPU, I also come down from 140W which I greatly appreciate. The systems are on par for core count and I save a few hundred dollars.

    Needless to say I’m very excited for this chip! Now if only the OEM we use for system builds would stock things sooner than 3+ months after release.

    • anotherengineer
    • 1 year ago

    I predict a lack of supply, leading to pricing higher than the “suggested” above.

    • Mr Bill
    • 1 year ago

    Pssst! The thermal solder is Ice-nine.

    • evilpaul
    • 1 year ago

    When are the Ice Lake/10nm chips supposed to be coming out?

      • chuckula
      • 1 year ago

      And the aptly named, “Sir not launching commercially in this decade.”

      • NoOne ButMe
      • 1 year ago

      2H2019 probably. 2Q2019 if we get really lucky

        • BurntMyBacon
        • 1 year ago

        2Q2020 if we get really unlucky.

        Thankfully, that is really unlikely.

    • firewired
    • 1 year ago

    No thanks Intel.

    I picked up my R7 1700X for CAD $260 a couple months ago and a B450 motherboard for less than any of your Z370/Z390 new-motherboard-every-CPU-release garbage would cost me.

    I really do not care if your CPU’s have higher clocks or higher IPC. Ryzen offers more cores per-dollar, more threads per-dollar, on a significantly less-expensive platform which saved me the money to buy what really mattered: a high-end video card instead of mid-range.

      • Kretschmer
      • 1 year ago

      CPU-limited gamer says “what?”

        • ptsant
        • 1 year ago

        I think that in real life most gamers are GPU limited, ie that the configuration of monitor resolution and settings that they use is more likely to stress the GPU. You really need a 1080 or faster and/or game at 1080p60 to be at real risk of being CPU-limited.

          • synthtel2
          • 1 year ago

          I end up CPU-bound in many games with an R7 1700 and RX 480 at 1440p because I turn settings down and don’t find most AAA stuff very compelling.

          • Kretschmer
          • 1 year ago

          I mostly agree, but once you get into strobing tech (which currently doesn’t work with adaptive sync), maintaining a steady 100/120/etc. FPS really requires a CPU with minimal compromises. I also think that most gaming CPU benchmarks are less than indicative by virtue of being repeatable (jogging down a forest path with no other characters is easily repeatable but less strenuous than a firefighter). In addition, a CPU that is merely suboptimal today becomes a bottleneck tomorrow.

          Everyone should choose a CPU based on their unique blend of needs. A 1700X isn’t what I’d recommend for a gamer, but it’s not likely to be a maddening bottleneck.

          • derFunkenstein
          • 1 year ago

          This checks out. I do tend to play games in real life, and I do tend to crank the details as much as the graphics card can handle.

    • Kretschmer
    • 1 year ago

    I’m really curious to see if the extra cores will help much in mainstream games, or if this processor will be a repeat of the 7700K/8700K benchmarks when fast RAM was used.

      • psuedonymous
      • 1 year ago

      The dual-core 5GHz turbo mode is more likely to be a factor than throwing extra cores at lightly threaded workloads.

        • Kretschmer
        • 1 year ago

        Exactly. I’m tempted to play around with lightly OCing my 7700K, but there aren’t a lot of noticeable slowdowns with a 100Hz GSync monitor. I’d be overclocking like mad if I had a 200Hz display…

      • Krogoth
      • 1 year ago

      You’ll probably see almost little or no return with gaming applications unless you plan on doing some multi-tasking on the background and/or streaming.

      The larger L3 cache pool will bump 9900K and 9700K ~1-2% over 8700K and 8600K respectfully in gaming applications.

    • wierdo
    • 1 year ago

    Any comment on this news?

    [url<]https://www.hardocp.com/news/2018/10/09/hardware_unboxed_analyzes_intels_commissioned_core_i99900k_benchmarks/[/url<] "Hardware Unboxed did a short analysis of a few of the benchmarks as their team felt that the i7-8700K benchmarks and the AMD Ryzen 2700X numbers were incorrect. They found that Principled Technologies had allegedly gimped the AMD CPUs by using different coolers, incorrect ram timings, and possibly even disabled some of the cores on the AMD Ryzen 2700X" [url<]https://www.patreon.com/posts/21950120[/url<] "The benchmarks carried out by Principled Technologies are even more bogus than we first thought. A few viewers pointed out that the Ryzen 7 2700X was listed as tested in the “Game Mode” within the Ryzen Master software and I foolishly thought they might have just made a simple copy and paste error in their document as they would have used this mode for the 2950X. This does explain why the Threadripper CPUs were faster than the 2700X in every test. What this means is a CCX module in the 2700X was completely disabled, essentially turning it into a quad-core. I’ve gone ahead and re-run the XMP 2933 test with Game Mode enabled and now I’m getting results that are within the margin of error to those published by Principled Technologies."

      • chuckula
      • 1 year ago

      [quote<] "The benchmarks carried out by Principled Technologies are even more bogus than we first thought. A few viewers pointed out that the Ryzen 7 2700X was listed as tested in the “Game Mode” within the Ryzen Master software and I foolishly thought they might have just made a simple copy and paste error in their document as they would have used this mode for the 2950X. This does explain why the Threadripper CPUs were faster than the 2700X in every test.[/quote<] AMD went out of its way to make "optimization" software that includes an advertised "game mode" that's actually bad for game performance? Looks like Intel was smart enough to call AMD on its bluff and turn on the "optimizations" that AMD advertises in its own product. As for canned benchmarks, I don't trust them from AMD/Intel/Nvidia/anybody but I'm not going to countenance the usual suspects acting all butthurt when Intel does a better job of playing the same game that AMD does whenever it launches a new product.

        • thx1138r
        • 1 year ago

        Nope: “AMD’s user guide for Ryzen 7 specifically states that Game Mode is reserved principally for Threadripper and that Ryzen customers shouldn’t use it”
        see:
        [url<]https://www.extremetech.com/computing/278526-the-core-i9-gaming-benchmarks-intel-commissioned-against-amd-are-a-flat-lie[/url<]

          • chuckula
          • 1 year ago

          Great Idea: If AMD is really full of these geniuses who are that much smarter than anybody else… how about they write their own “optimization” software so that it won’t even let you engage this “game mode” on 8 core chips!

          What a concept!

            • thx1138r
            • 1 year ago

            Or they could just trust their users and give them a big red warning when they try to do something that may not be beneficial. Oh wait, that’s what they did.

            • chuckula
            • 1 year ago

            Where’s the big red warning on the screenshots from that article?

            Oh, there’s a footnote somewhere buried deep in AMD’s documentation that nobody reads?

            Once again: [b<]THIS IS AMD'S OWN SOFTWARE THAT THEY WROTE AND CONTROL 100%[/b<]. I haven't slung code in 10 years but I'd bet you AMD's entire 2018 production run of Threadrippers that I could throw in a trivial CPU-id check and change "game mode" to enable all 8 cores on standard RyZen parts in less than a day. As for "oh look we give users the option to turn off cores because FREEDOM! Thanks AMD!" line, I've had the ability to turn off cores on practically any chip in software since multicore CPUs became a thing. AMD doesn't deserve to be blindly worshipped because their "optimization" software is designed by third-rate amateurs and somebody had the gall to actually click the big buttons AMD provided for them.

            • thx1138r
            • 1 year ago

            I do apologize, the warning was actually in yellow:
            [url<]https://www.extremetech.com/wp-content/uploads/2018/10/Ryzen-Master-Guide.jpg[/url<] You're entitled to you opinion of course, but I still the "personal freedom" approach of giving people the ability but warning them of the possible consequences rather than prohibiting an action entirely

            • chuckula
            • 1 year ago

            [quote<]the warning was actually in yellow:[/quote<] Wrong, the photoshopped picture from the article has added yellow highlighting to a text bubble that's not part of the main interface. Once again, the onus is on AMD to make "game mode" in its own freakin' optimization software actually help or at least not harm games. That's the real story here since the results of the tests are completely irrelevant.

            • thx1138r
            • 1 year ago

            That’s not some random photoshopped picture, that’s from the official Ryzen Master Reference Guide:
            [url<]https://download.amd.com/documents/ryzen-master-quick-reference-guide.pdf[/url<] While regular users need not consider such documents, I would expect professional reviewers to consider them while attempting to create a non-biased review.

            • chuckula
            • 1 year ago

            1. As I said, a minor note buried half way down the documentation is only proof of a failure on AMD’s part.

            2. Did you bother to actually read the linked document? There’s NO HIGHLIGHTING AT ALL in the original. Your click-bait conspiracy losers who think that Intel suddenly invented marketing benchmarks out of thin air last week added the “OMG IT’S HIGHLIGHTED IN YELLOW” using photoshop in their own article.

            3. Does AMD control the software? Yes. Did AMD screw up? Yes. Did this third-party firm take advantage of AMD’s incompetence? Yes. Is this even in the same universe of the worst marketing stuff I’ve ever seen? Not even remotely close.

            Call me back when Intel’s CEO sits 3 feet away and smiles approvingly from an Intel employee who flat out lies about how an upcoming product is an “overclocker’s dream”. [b<]THAT[/b<] is over the top dishonesty, and Lisa Su is personally guilty of it.

            • thx1138r
            • 1 year ago

            I admit that the highlighting color was mis-represented on that link, but is the color really that important. Because what you really saying that it’s ok for a professional reviewing company to ignore the Reference Guide for the software they are using. You might be fine with that, I’m not and none of your bluster is going to change my mind about that.

            And why are you bringing up some minor slip by Lisa Su, is that really equivalent to some of the outrageous illegalities that intel has actually been convicted of in the past? We could play that game all day.

            I honestly don’t see why you are so outraged, all companies are guilty of indiscretions of some scale or other. All of the indiscretions should be called out, as this one has been. Bringing up past indiscretions by the (currently) wronged side just looks like sour grapes, particularly when you try to imply that they are equivalent to some other indiscretion.

            So I assume you don’t want to take that bet then…

      • K-L-Waster
      • 1 year ago

      As with any product from anyone, ignore any performance figures that are supplied by the vendor.

      • jihadjoe
      • 1 year ago

      “Principled” Technologies lol

        • derFunkenstein
        • 1 year ago

        more like “Principaled” Technologies, amirite?

        [url<]https://www.merriam-webster.com/dictionary/principal[/url<] [spoiler<]definition 2 of course[/spoiler<]

      • derFunkenstein
      • 1 year ago

      The EIC is totally subtweeting you today. [url<]https://twitter.com/jkampman_tr/status/1049725963012136960[/url<]

      • Jeff Kampman
      • 1 year ago

      I don’t know why we would comment on something we didn’t deem worthy of reporting on to begin with. Wait for independent reviews.

        • ptsant
        • 1 year ago

        The point is to reinforce what you are saying, ie that only truly independent reviews are worthwhile.

          • ronch
          • 1 year ago

          Unfortunately data like this may be taken seriously by anyone who doesn’t know a thing about computers or proper benchmarking.

            • K-L-Waster
            • 1 year ago

            I assume you’re using the word “data” in the loosest possible sense?

            (And yes, you are right about the risk to less-educated consumers.)

            • ronch
            • 1 year ago

            Yes, ‘data’.

      • thx1138r
      • 1 year ago

      Another humorous point from PT’s “Report” is that they used the stock cooler on the 2700X and every other CPU used the rather nice Noctua NH-U14S.
      “Principled Technologies”, another company named in the Orwellian tradition.

        • chuckula
        • 1 year ago

        Given the fact that AMD advertises its stock Wraith cooler as being so amazing I don’t see a problem with that either. I seem to recall numerous AMD fanboys complimenting the Wraith as being amazing. Oh, and none of the chips were overclocked, so it’s not an issue of running the chips outside of their normal range. Are you suggesting that AMD’s Wraith coolers are inadequate for running RyZen at stock clocks?

        As for the Noctua, that’s not even a water cooler, so I really don’t see much of an issue using two different air coolers.

          • freebird
          • 1 year ago

          So has Gamer Nexus stated in their video, since the Intel CPU didn’t include a cooler it shouldn’t have been tested with one, using that infallible logic.

            • chuckula
            • 1 year ago

            No, I’m saying that AMD emphatically recommends that its RyZen parts be used with Wraith coolers… so that’s how they were tested.

            And this is suddenly considered to be “unfair” to AMD.

            Incidentally, those Noctua coolers were used to test Threadripper parts from AMD too, but for some reason nobody complained about that.

          • thx1138r
          • 1 year ago

          [quote<]Given the fact that AMD advertises its stock Wraith cooler as being so amazing[/quote<] I don't know where you saw the Wraith being advertised as "Amazing", I've combed through the marketing materials and can't find it. The nearest thing I could find is on the page: [url<]https://www.amd.com/en/technologies/cpu-cooler-solution[/url<] where they say their Wraith coolers have [quote<]capable thermal performance[/quote<] Also, I'm pretty sure you're misquoting that random AMD fanboi that you are so fond of. Anyway, I'm pretty sure the fanbois quote was not [quote<]complimenting the Wraith as being amazing[/quote<] They were in fact: [quote<]complimenting the Wraith as being amazing for a stock cooler[/quote<] See, if you can make up stuff that a fanboi might say, I can too. Anyway, [quote<]Are you suggesting that AMD's Wraith coolers are inadequate for running RyZen at stock clocks?[/quote<] Inadequate, no, but I do have a suspicion that the Noctua NH-U14S might be able to keep the 1700X at boosted clocks for longer. To be honest I don't know for sure, I can't find a thorough enough review of the Wraith Prism to be sure.

            • chuckula
            • 1 year ago

            The funny thing is when AMD disingenuously said that a Skylake X “needs” to have liquid cooling at its canned-demo press event for Threadripper 2 nobody batted an eyelash at the fact that they were flat out lying.

            Furthermore they parroted the line.

            But now that a third-party has committed the ULTIMATE CRIME of using air coolers on two chips including an air cooler that AMD flat out says can cool its own chip at stock speeds… OMG END OF THE WORLD.

            Tell ya what, let’s make this interesting: Who wants to bet that the RyZen 2 miracle chip in 2019 mysteriously fails to beat the 9900K in gaming benchmarks by the same margin or greater than the 9900K beats the previous miracle RyZen chips this year.

            And no, I’m not talking about third-party marketing fluff that nobody cares about. I’m talking about TR’s official review.

            • thx1138r
            • 1 year ago

            Didn’t see that skylake-x-needs-liquid-cooling thing, but thanks for bringing it to my attention, I agree that all such nonsense should be highlighted and called out, do you have any links?

            Nevertheless, I would still argue that the two things are not equivalent. An offhand comment about users needing to use a particular cooler is not the same thing as deliberately crippling a competitor in an attempt to skew the only publicly available benchmarks of a new product.

            [quote<]Who wants to bet that the RyZen 2 miracle chip in 2019 mysteriously fails to beat the 9900K in gaming benchmarks by the same margin or greater than the 9900K beats the previous miracle RyZen chips this year.[/quote<] That's a very telling way of putting it, so the Ryzen 2 will have to outpace the 9900k by presumably decent margin. You could (and much more easily) have just made the bet that Ryzen 2 would/wouldn't beat the 9900K in gaming. So that means you must think the Ryzen 2 will take the gaming crown next year, interesting... Care to make a bet on the absolute gaming champion next year?

            • freebird
            • 1 year ago

            Chuckla the troll does that all the time to make a point…makes all sorts of claims by supposed “fanboys” so that he can denigrate the claims only he is making… He’s put up so many “straw-man” arguments on here that the price of scarecrows has gone up like the price of DRAM that last year or two…

            • thx1138r
            • 1 year ago

            Not even on wccftech have I seen the type of comments that Chuckula attributes to AMD fanbois…….

            ……checks wccftech, tries again………

            Only on wccftech have I occasionally seen the type of comments that Chuckula attributes to AMD fanbois.

            That site has a lot to answer for.

            • BurntMyBacon
            • 1 year ago

            Look at what they’ve created. It isn’t hard to see how chuckula’s comments could become increasingly unreasonable as a direct result of and in answer to the wccftech brainwashed fanboi comments. It would be appreciated, however, if chuckula would direct these comments towards them exclusively and post here when he’s done [s<]venting[/s<] [i<]correcting the imbalance[/i<]. His posts can be quite informative, insightful, and even comical when he wants .

            • thx1138r
            • 1 year ago

            I agree, if more and more people start quoting wccftech style posts, then it’s only a matter of time before this sites starts to go the way of wccftech. I think there should be a ban on making this sort of quote on this site, or at the very least it should be called out and frowned upon. If you want to quote and complain about a wccftech post then the best place to do that is wccftech itself.

      • freebird
      • 1 year ago

      Looks like Gamers Nexus commented alot on the subject:

      [url<]https://gaming.youtube.com/watch?v=D1mJMI_uaa8[/url<]

        • jihadjoe
        • 1 year ago

        I agree with Steve these guys are grossly incompetent, rather than outright malicious.

        Why incompetent? Because they also did stuff that was in favor of the slower CPU, like Civ VI’s completely useless and broken graphics benchmark.

          • BurntMyBacon
          • 1 year ago

          I rather think they are both incompetent and malicious. Perhaps less malicious than attributed and more incompetent, but I can’t really pass off some of their choices as unintentional.

      • Chrispy_
      • 1 year ago

      Yeah. Intel have screwed up big time by not immediately distancing themselves from this ridiculous and stupidly-biased attempt at a comparison.

      With that memory, cooling, and cores handicap, Principled Technologies have effectively shown that the 9900K is only 30-50% faster than an 18-month old, [url=https://www.newegg.com/Product/Product.aspx?Item=9SIA2W08547755&cm_re=1500X-_-19-113-436-_-Product<]$145[/url<] Ryzen5 1500X. Any gamer that realises this is going to realise that the extra $340 not spent on the GPU is going to let them buy a 2080Ti instead of a vanilla 2080 for a huge gaming advantage.

      • wierdo
      • 1 year ago

      Some interesting analysis relating to this situation:
      [url<]https://www.youtube.com/watch?v=3V8pEsjNa4Q[/url<] Apparently Intel did this before with Bapco, some interesting history mentioned that was entertaining to hear about relating to Toms, Vans and HardOcp etc.

        • Redocbew
        • 1 year ago

        Someone is wrong on the Internet, and someone else is doing something they probably shouldn’t be doing. Good luck trying to get that under control.

          • wierdo
          • 1 year ago

          The history is entertaining, the details of how Bapco served as cover for Sysmark while the code is fully developed in-house at Intel, was the interesting info. More amusing is their offices using the same street address as Intel. Coincidence?

          And this new prop, Principled Technologies, doing the same song and dance. It’s just interesting how history repeats itself, except this time Intel is obligated by court order to post a disclaimer on their work due to past infractions… hence the disclaimer showing up in the report from this so called independent organization, extra amusing.

    • 1sh
    • 1 year ago

    I paid nearly $400 for 8700k less than a year ago… I’ll take 2 more real cores on the 9700K over my extra 8 threads.

      • K-L-Waster
      • 1 year ago

      Same here, and… same here 🙂

      The usual Q & D estimate of the value of an HT thread is roughly 30% of a physical core (heavily dependent on the task of course). 6 cores * 1.3 comes out to 7.8, so just slightly less than 8 physical cores.

    • shank15217
    • 1 year ago

    These cpu boxes are getting increasingly ridiculous, maybe I can cut a hole in the side and make a house for my gerbil wtf.

      • Chrispy_
      • 1 year ago

      I threw out a cubic meter of Threadripper packaging today.

      It’s all going to end up in the ocean, I suspect 🙁

      Okay, I’m exaggerating, [url=https://i.imgur.com/VvkAOVk.jpg<]the trash bag of Threadripper and Noctua cooler boxes[/url<] is only 0.95m x 0.75m diameter which is only 0.42 cubic meters ('just' 15 cubic feet of plastic and polystyrene trash in freedom units for you).

    • auxy
    • 1 year ago

    [quote=”the people who will absolutely never make use of more than 6 threads”<]Yaaay! We're finally getting moar coars! Thanks AMD!!! \(*'▽')ノ[/quote<] ;つД`)

    • dhmhtries22
    • 1 year ago

    (((((it’ll bring eight Skylake cores))))))

    Are these indeed skylake cores or just miswriting???

      • Star Brood
      • 1 year ago

      They stopped innovating on core architecture for their + series, they are just throwing in more cores (which they already had from their higher-end kits before).

      They don’t want to spend the money re-designing cores for a dead-end node size. The stuff they have actually innovated on is waiting for their 10nm.

      I’m hoping that – when that 10nm finally comes out – they can actually meet the market demand and get prices back to normal. That and hopefully bring some really nice IPC improvements to settle down those games which still can’t figure out how to run optimally on more than 1 or 2 cores.

    • ronch
    • 1 year ago

    No thanks I’m going Ryzen. You’re only giving us all this because of AMD and if it weren’t for them you’d still be shoving 4C/8T down our throats for $350 until who knows when. So out of principle alone, I’m going AMD all the way until the Apocalypse.

    • ronch
    • 1 year ago

    The real story here is the 9700K and the 9600K. At those prices they are competing more directly against the Ryzen 2600X and 2700X. AMD still gives you more for your money though, but people who game first and foremost will have more reasons to get Intel. Will AMD drop prices in response?

      • Krogoth
      • 1 year ago

      Not likely, the 9700K and 9900K are going to be faster then 2700K at everything but sell at higher price points.

      9600K is pretty much a 8600K with higher turbo boost at the same price point. 8700K and 8086K will remain at its current spots.

      Intel is really banking on the clockspeed advantage of Coffee Lake to retain its dominance in the mainstream market.

      Things are much more interesting in the HEDT world though.

        • ronch
        • 1 year ago

        The 2700X is $295 at the Egg right now, from $320-$330 in recent weeks and months, and the 2600X is $200, from $210. I thin AMD is quietly adjusting prices to make more folks buy from them now. The more folks buy AMD right now the less folks will be buying these upcoming Intel chips when they come out. People buy computers to use for a long time. That’s why in this market these companies know the importance of making people buy NOW.

    • Pville_Piper
    • 1 year ago

    So why has Intel, who always pushed Hyper threading, suddenly not adding it on the i7?

      • ronch
      • 1 year ago

      Intel is the master of product segmentation.

      • freebird
      • 1 year ago

      Maybe thermals, maybe more HT side channel attacks looming. I think AMD’s threading maybe a little better than Intel’s between that and thermals maybe it doesn’t compare as good in multi-threaded apps. Most likely they are just trying to maximize the “feature upgrade” syndrome of people wanting the “Best Gaming CPU ever” to opt for the i9900K with HT.

    • DPete27
    • 1 year ago

    Seeing all the extra CPU power headers on the Z390 boards, I’d be very cautious about plunking an i9-9900K into a Z370 board.

      • DeadOfKnight
      • 1 year ago

      Those extra headers are usually overkill and put there so they can brag about some professional overclock done on them.

      • Waco
      • 1 year ago

      Even my Threadripper board only has the 8 pin and 4 pin connectors…and the 4 pin is optional.

    • crystall
    • 1 year ago

    It’s wonderful what a competitive AMD does to the PC market.

      • drfish
      • 1 year ago

      Now about those GPUs guys…

      • Klimax
      • 1 year ago

      No AMD is just yet another incentive. Main incentive is and always was to get poepl to upgrade. No upgrades, no large sales. (You get only replacements and new devices and that’s not that much)

      Note: GPUs are very different case and pretty much opposite…

        • crystall
        • 1 year ago

        The first Intel quad-cores were released in Q1’07. Intel only released quad-cores for their mainstream desktop offerings until Q4’17 when they launched Coffee Lake, only a few months after Ryzen launched. It took Intel 10 years to go from 4 cores to 6 and now they’ve bumped it to 8 already. Coincidence? I don’t think so.

    • ozzuneoj
    • 1 year ago

    Maybe after 7 years a $370 CPU will finally give me both a large increase in per-thread performance and twice as many cores over the 2500K I spent $200 on in 2011.

    If I’d waited this long to upgrade last time I’d have gone from an Athlon XP or single core Athlon 64 to Sandy Bridge (four times the cores and multiple times the per thread performance).

    I’m hoping with the soldered IHS the temps at stock speed will be at least close to what I’m used to with my current CPU when overclocked… though that’s probably a bit optimistic.

    • Thresher
    • 1 year ago

    Is it just me or did intel just raise the price of their CPUs by about $100 at every segment?

      • colinstu12
      • 1 year ago

      Well they did double the cores at every segment so…

        • Thresher
        • 1 year ago

        The price points were pretty much established no matter how many cores there were.

      • Krogoth
      • 1 year ago

      I suspect it is because they are trying to wrestle with the current 14nm production shortages and they also want a bigger piece of the pie that etailers enjoyed during the short-term Coffee Lake shortages at its launch.

    • leor
    • 1 year ago

    Why is Intel segmenting their products with hyper-threading???

      • K-L-Waster
      • 1 year ago

      They always have — the difference here is the cut off for HT / non-HT has moved from the i5 to the i7.

      • DPete27
      • 1 year ago

      Welcome to Earth.
      More threads = more $

      • drfish
      • 1 year ago

      It makes perfect sense add HT to i3s, remove it from i9s. Don’t you get it?

      • NoOne ButMe
      • 1 year ago

      Because they need to squeeze every last bit of performance out of their SKUs.

      8C/8T will typically be slightly better than 6C/12T. Not by much, but 5-10% in MT.

      And the i5 can’t use 6C/12T because that gets to close to the i7 now.

      And 6C/6T > 4C/8T.

      It is all to push for maximum performance while being able to segment.

      • f0d
      • 1 year ago

      And does the same
      All Ryzen 3 cpus have hyperthreading removed for “segmentation”

    • Chrispy_
    • 1 year ago

    The 9700K looks like the one to get, I don’t think SMT is worth another $114 – especially since the target audience is probably gamers and SMT is hit/miss when it comes to adding or subtracting performance.

    Those who need cores yet aren’t gaming would have picked up an 8 core chip from the other team a year ago and at a lower price.

      • Sahrin
      • 1 year ago

      The extra 4 Meg of cache on the other hand could be quite valuable to gamers.

      • Voldenuit
      • 1 year ago

      [quote<]Those who need cores yet aren't gaming would have picked up an 8 core chip from the other team a year ago and at a lower price.[/quote<] Intel pretty much only makes sense to me if you /need/ 120+ fps gaming, and that also means the only viable chips in their lineup are the 8600K, 8700K, 9600K, 9700K and 9900K. And for the 8xxx series, that's only if you're going to delid. But with the recent price hikes/supply shortages, the value proposition for practically any intel chip is dead right now. If you're streaming, a 2600X or 2700X is a much better choice, as you'll be better at encoding while you stream, and nobody's streaming 120 fps video anyway. And for practically any other productivity scenario, AMD gives better perfomance/price.

      • JustAnEngineer
      • 1 year ago

      Specter/Meltdown/etc. mitigation patches have been knocking down the performance of SMT, anyway, haven’t they?

        • BIF
        • 1 year ago

        Yes they have. Don’t know if this one would be affected, but I’ve noticed it even in a CPU made as recently as within the last 2.5 years.

        I’m feeling “meh” for CPUs unless I know they’ve got protection built in. Or not affected by those vulnerabilities.

        I also want to see some real prison time for those who would create the vulnerabilities.

          • Chrispy_
          • 1 year ago

          [quote<]I also want to see some real prison time for those who would [b<]EXPLOIT[/b<] the vulnerabilities.[/quote<] FTFY, I think.

      • srg86
      • 1 year ago

      I’m in that bracket, though I prefer the reliability of Intel chipsets and memory controllers (AMD’s have always caused me problems long term, hence they are on my avoid list, I just don’t trust them).

      If I was in the market for a CPU right now, I’d definitely go for the i9-9900K, running at stock. 8 cores + hyper-threading + extra cache + integrated GPU….just about perfect. That price is steep though.

      That said, I’m waiting for IceLake. Mainly because it fits my timeline, but also should have Spectire/Meltdown fixes in hardware.

        • Chrispy_
        • 1 year ago

        As someone who has managed thousands of machines, including several hundred Haswell models, I can assure you that Intel’s chipsets and memory controllers are far from flawless.

        Haswell’s FIVR simply goes bad over time. It’s the only CPU in my lengthy employ that has reliably been a “failing CPU”.

        Historically AMD has had some flaky third-party chipsets but no more, proportionally than Intel.

        I’m not saying AMD make the better chips. If money is no object, Intel’s best will beat AMD’s best still; Unlike Intel, AMD are still making decent generational improvements and their product line is affected by way fewer of the Spectacle/Melt-face shenanigans, all ahead of Intel to market and for more overall performance/efficiency/value than Intel has provided in half a decade or more. Outside of high-fps gaming where every bit of IPC and clockspeed counts, Intel really is cashing in on its brand image to sell their products. They don’t really stand up to Ryzen at the current prices under their own merit.

          • Convert
          • 1 year ago

          What about the FIVR on Haswell?

          We also have thousands of Haswell machines deployed and I am unaware of the issues you describe. I’ve also never heard of any reports about it, other than people complaining that the chip ran hotter or it restricted their OC.

          • synthtel2
          • 1 year ago

          How has that been going? I’ve seen one more (i7-4700MQ) fail since we last talked about it, but that one looked more like Toshiba’s fault (their heatsink design failed to properly cool a corner of the die).

          • srg86
          • 1 year ago

          I’m not saying they are flawless, but I’m staying with what works better for me, and I’ve had better experiences with Intel, plus I like fast cores with integrated graphics.

          I’m willing to pay a reasonable amount more for the relative piece of mind.

          I’ll also add that at least for what I do, the Spectre/Meltdown mitigations have hardly hit me at all, I’ve not really noticed any difference, though even I’m willing to wait for hardware fixes.

      • Convert
      • 1 year ago

      Oops, wrong post!

    • Thresher
    • 1 year ago

    Will there be non-K processors?

    It seems like there would be room for a 8core/16thread 9900.

    If I were intel, I would have also kept the 9700K as a 6core/12thread, but it’s possible that performance was just too close to what the 9700K could mete out.

      • Freon
      • 1 year ago

      I would expect the non-K variants to be announced soon. Give it a few weeks.

      Lack of a 6/12 part is interesting, but perhaps they didn’t want to deal with further market confusion between 6/12 vs. 8/8.

        • Krogoth
        • 1 year ago

        That’s the 8700K and 8086K spots.

    • jensend
    • 1 year ago

    Skylake Refresh Refresh Refresh: Now with a couple bucks’ worth of solder FOR FREE!

      • defaultluser
      • 1 year ago

      The didn’t even have the decency to include a price cut this time around.

      The Core i5 spins in-place, while the high-end chips get a tiny performance bump (10-20% for the Core i7 9700k over the 8700k, assuming you can exercise all 12 threads) And the bump the price into the stratosphere if you actually want a performance improvement.

        • JustAnEngineer
        • 1 year ago

        Micro(processor)economics. If Intel has neglected to develop enough productuon capacity to meet demand, price increases are a natural result of the intersection of the supply and demand curves. Volume goes away from consumers unable/unwilling to spend so much or those who switch to competing (AMD) products.

          • Klimax
          • 1 year ago

          That’s why Intel recently announced they are investing extra billion of dollars in manufacturing to alleviate supply constraints.

    • xeridea
    • 1 year ago

    They created a fake new selling point, now with solder!!! No longer runs 90C stock!

      • Krogoth
      • 1 year ago

      Intel chips haven’t been running 90C at stock unless you have poor/subpar cooling. This was even true back with the infamous P4 Prescott and P4 Smithfield chips.

        • DPete27
        • 1 year ago

        They’ll run at 90C on Intel’s stock cooler.

          • Krogoth
          • 1 year ago

          Sorry, they ran around ~80C back in the day at stock settings and cooler.

          90C non-sense is either from overclock chips or poor/improper cooling (HSF isn’t fitted right or the person forgot to remove the plastic cover on the bottom of the stock HSF)

            • DeadOfKnight
            • 1 year ago

            They weren’t properly hydrated.

          • wingless
          • 1 year ago

          They sure will!

            • Krogoth
            • 1 year ago

            Only if you are an idiot who leaves the MCE on and other auto-overclocking garbage and/or doesn’t properly [b<]install[/b<] the HSF. The whole 90C at stock settings is meme-tier non-sense from foolish builders. Not even Bulldozer chips got that high at load even the infamous FX-9590.

          • K-L-Waster
          • 1 year ago

          … which is irrelevant because these SKUs don’t include Intel’s stock cooler anyway…

        • Voldenuit
        • 1 year ago

        There’s definitely been variability in Coffee Lake thermals from reviewers. TR didn’t have any problems (78C), but [url=https://arstechnica.com/gadgets/2017/10/intel-coffee-lake-8700k-review/<]ars got to 90C... with a 280mm AIO water cooler[/url<]. [quote<]At stock speeds on auto settings, the 8700K reaches a toasty 90°C under full load when paired with a substantial 280mm liquid cooler (in this case a CoolerMaster MasterLiquid Pro[/quote<] Did ars just get a bad sample? Most other review sites had no issues, but the reddit thread on Coffee Lake thermals has many users with similarly high humbers.

          • K-L-Waster
          • 1 year ago

          Was that with Multi-Core-Enhancement enabled or disabled? (I’m betting enabled.)

            • Voldenuit
            • 1 year ago

            They don’t say, but it’s probably enabled, as they used an ROG motherboard.

            From what I’ve seen, Asus boards default to enabled, while Gigabyte boards default to disabled. The target market for K processors will be the type to tweak settings, and usually enable MCE anyway, but it would be nice for reviewers to be more specific.

            IIRC, TR tested their 8700K on a Gigabyte board, and noted that MCE was disabled by default.

            • K-L-Waster
            • 1 year ago

            My ASUS ROG STRIX had MCE enabled by default. If Ars didn’t turn it off for their tests, that would easily explain the “stock” temps being that high.

            • Voldenuit
            • 1 year ago

            I mean, does anyone with a K processor actually turn MCE ‘off’ unless they are trying for very high single core runs anyway?

            • K-L-Waster
            • 1 year ago

            I did — in my case, the system was unstable with MCE enabled, and it wouldn’t boot period with both MCE and the memory running on XMP rather than JDEC.

    • Krogoth
    • 1 year ago

    9900K will certainly beat 2700X and “2800X” assuming that AMD cherry-picks “2700Xs” with higher boost and base speeds akin to the 8086K.

    I suspect that dual-channel DDR4 bus is going to be an issue for eight-core Coffee Lake chips under some workloads and ring-bus topology’s weaknesses will start to expose itself. The 9900K and 9700K are probably going to need some decent to high-end cooling if you want to avoid throttling at maxmium loads. They will also benefit much more from factory overclocked DDR4 DIMMs due to being stuck with dual-channel DDR4.

      • DPete27
      • 1 year ago

      The Ryzen 2700X has dual channel memory controller that’s less efficient than Intel’s.

        • Krogoth
        • 1 year ago

        The margins are much closer than you think though. There’s a reason why Intel originally gave their six and eight-core HEDT designs quad-channel DDR4.

        Eight-core Coffee Lake chips are merely a stopgap designed to usurp 2700X’s crown at multi-threaded workloads in the normal desktop market. The biggest victim of this are some of the Skylake LCC SKUs.

          • freebird
          • 1 year ago

          “Margins” is an interesting word… the margins should be ok on the 9000 series chips. But now that Intel is basically saying that 6-8 cores is now the new mainstream, do they have enough capacity at 14nm to make the larger 6-8 core cpu dies for desktops massively larger Server CPUs and also the tens of millions of 5G modems they will be providing to Apple?

          Probably why all other Intel CPUs seem to be going up in price over the last month or so…

          Where does that leave the inventory of 2-4 core CPUs they used to be able to charge $200-$300?
          Maybe they have a big enough inventory of that to last awhile and can stop making them, I would. Laptop 4-core+ low power CPUs would be much more profitable.

            • Voldenuit
            • 1 year ago

            I’ve read that the fab capacity for intel is so bad right now, they’re reopening their 22nm process lines.

          • stefem
          • 1 year ago

          [quote<]There's a reason why Intel originally gave their six and eight-core HEDT designs quad-channel[/quote<] Yes, the reason is that they were targeted at a different sector of the market which have different needs

            • Krogoth
            • 1 year ago

            Those six and eight-Coffee Lake SKUs are starting overlap into that aforementioned market. Intel’s artificial segmentation non-sense for the past decade is just rearing its ugly head.

        • biffzinker
        • 1 year ago

        My Ryzen 2600x dual channel memory controller seems plenty efficient too me. Out of a theoretical 51.2 GB/s I’m getting 49.7 GB/s on read, 48.2 GB/s for writes, and 45.4 GB/s copying with the XMP of 3200 MHz.

          • synthtel2
          • 1 year ago

          AMD’s uncore is better at bandwidth and scaling, Intel’s (ring bus) is better at latency, power consumption, and die area. The scaling concern with the ring bus can’t really be drawn out that easily, though.

            • Krogoth
            • 1 year ago

            Ring-bus is only better at die space until you start going beyond six cores or more. It starts becoming more and more expensive on die-space as you scale-up on core count. The latency advantage also starts to wane away as well. You also run more into resource contention issues assuming that all cores are sharing the same cache pool.

            Alternative topology schemes begin to make more scene with high-core count designs but they have their own set of trade-offs.

            • synthtel2
            • 1 year ago

            8C worth of Intel’s ring bus is still pretty tiny compared to whatever’s going on in AMD’s uncore. I did mention scaling.

            • psuedonymous
            • 1 year ago

            Ring bus has proven scaling up to 12 cores per ring (or 22 per double-ring). And that was with ‘only’ 4 channels (2 per ring) on the Broadwell HCC die.
            I don’t think a mere 8 cores is going to stress the ring bus.

            • Krogoth
            • 1 year ago

            It does scale but there are diminishing returns beyond six cores though that progressively get worse as you move up with core count.

            That’s the main reason why Intel went with mashed topology with Skylake-X. Skylake-X HCC and XCC completely outclass their Broadwell-EP and HCC predecessors at performance, scalability and power efficiency.

      • drfish
      • 1 year ago

      This is going to be important. If I was hypothetically looking at either the 9900K or 2920X for an upgrade, the cost savings of not needing to buy two extra sticks of fast memory would actually be a factor for me.

      Completely unrelated, anyone looking for a good deal on a 7700K and Z270 mobo? 😛

        • K-L-Waster
        • 1 year ago

        Didn’t anyone tell you should should have waited, fish?

          • drfish
          • 1 year ago

          *hangs head in shame*

            • RAGEPRO
            • 1 year ago

            [url<]https://www.youtube.com/watch?v=noJKWn9XkvM[/url<]

      • ptsant
      • 1 year ago

      AMD can beat them on price and $488 is close to TR territory. And, at least where I live, the actual selling prices for Intel are definitely over MSRP.

        • Krogoth
        • 1 year ago

        Threadripper lacks the clockspeed for workloads that care more about that then having more threads though.

          • ptsant
          • 1 year ago

          Place for a 2800X at +5% frequency and $350?

        • NoOne ButMe
        • 1 year ago

        $488 is over Threadripper territory.
        1920x is 500 currently
        1900x is $320-350

        Motherboard+RAM sways things towards Intel.

        But that comes down to how much heavy CPU stuff you do outside of gaming.

        If you do a lot, Threadripper’s easy “drop in” upgrade to a much higher core count may be useful.

        Otherwise the 9900k is probably better.

      • lycium
      • 1 year ago

      > ring-bus topology’s weaknesses will start to expose itself

      Oh my, that sounds even more exciting than all this “teasing” one hears about these days.

        • Krogoth
        • 1 year ago

        We already have known about it from Broadwell-E and Broadwell-EP chips which do show some scaling issues when going beyond six-cores. That’s why Intel went with meshed topology with Skylake-X and future high-core count designs.

          • jihadjoe
          • 1 year ago

          The mesh topology didn’t exactly give a good accounting of itself vs good old ring-bus for low-core count chips. IMO it’s good for 12 cores and up, but 8-10 cores is still well served by ring-bus.

          The first Skylake-X chips were such a small IPC improvement over Broadwell-E, despite desktop Skylake being a much bigger jump.

            • Krogoth
            • 1 year ago

            Meshed topology yielded noticeable gains under heavily threaded loads but that was only for HCC and XCC Skylake-X chips. There are returns for LCC SKUs but they were marginal at best. They were slower than their desktop counterparts at lightly threaded loads because they had dedicated smaller L3 pool(s) instead of a single large L3 cache (where one or two cores can have of it all). The same single large L3 cache pool does become a problem if you have a ton of cores contesting over it.

            Skylake’s jump over Broadwell/Haswell was almost owed due being first mainstream Intel platform to offer DDR4 which alleviated the bandwidth issues with Haswell/Broadwell.

            The performance problems under heavily-threaded loads for 8-core Coffee Lake SKUs will be further compounded by being stuck on platform that only has dual-channel DDR4.

      • the
      • 1 year ago

      The most hops Intel’s ring bus ever had was 13 nodes I thinks. It could go to 16 but Intel has never released such a chip.

      Of note is that the GPU, PCIe and memory controller count as nodes on the 9900K. That would equate to 11 nodes.

      Midrange/high end Xeons would also need to count QPI links and on-die ring bridges toward the number of nodes per rings. However past 10 cores, Intel splits the ring bus into two partitions. Also there are also two memory controllers.

      • Mr Bill
      • 1 year ago

      [quote<]under some workloads and ring-bus topology's weaknesses will start to expose itself[/quote<] "and in the darkness, bind them"

        • BurntMyBacon
        • 1 year ago

        Knew there was a LotR reference in there somewhere. Though it seems the One Ring(-bus) got thrown into the fires of Mt. Doom by Skylake-X.

    • Takeshi7
    • 1 year ago

    [quote<] The ninth generation of Core desktop CPUs also marks the long-requested return of solder as the thermal interface material (or TIM) beneath those chips' heat spreaders, enhancing heat transfer from the chip to the heatsink above.[/quote<] So Intel is pulling a Trump and taking credit for fixing a problem they created?

      • drfish
      • 1 year ago

      I applaud Intel’s foresight. They planted a rabbit in their hat years ago to pull out for just this occasion. Courage! Or something…

      • Voldenuit
      • 1 year ago

      At least they fixed the problem instead of just claiming they’d fixed it while doing nothing of substance.

        • Krogoth
        • 1 year ago

        It was only an issue with Ivy Bridge and first batch of Haswell chips. It has been mostly fixed since then.

        Soldering is going to make fitting more consistent at the factory. Just don’t expect miracles from it.

          • Chrispy_
          • 1 year ago

          Agreed, though the improved TIM for the Haswell refresh just moved the line in the sand from “inadequate” to “barely adequate”.

          TIM was of no consequence to the vast majority of Intel’s customers but was incredibly insulting to overclocking buyers who were paying at least $100 of Z-series and K-series tax only to be given a chip that was hindered by a thermal interface limitation that saved Intel single-digit US-cent savings per CPU.

            • Voldenuit
            • 1 year ago

            Yeah, I side-graded from Haswell to Devil’s Canyon (don’t look at me like that, I wanted more threads for Vegas Pro, okay?), and noticed no difference in TIM performance (pre-delid).

            • Krogoth
            • 1 year ago

            It was because Intel was more concerned about dealing with a potential “Bumpgate” on their hands due to potential long-term thermal cycling/material fatigue issues with soldering/IHS.

            K series chips share the same packaging as the regular non-K and other SKUs meant for other markets. It makes no fiscal sense to make a separate line for chips with soldering until now. IMO, the whole move with going back to soldering is just a marketing stunt to help Intel retain marketshare/mindshare. It isn’t going to create miracles.

            The irony is that hardcore overclockers actually prefer TIM/IHS setup since it makes de-lidding a lot less risky.

Pin It on Pinterest

Share This