Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Intel Power Hardware Technology

Intel Launches Core i9-9900KS 8-Core CPU At 5GHz Across All Cores (hothardware.com) 89

Posted by BeauHD from the setting-the-bar dept.
MojoKid writes: As the "S" in its name implies, the new Intel Core i9-9900KS that launched today is something akin to a Special Edition version of the company's existing Core i9-9900K 8-core CPU. The processors are built from the same slab of silicon -- an 8-core, Coffee Lake-refresh based die and packaged up for Intel's LGA1151 socket. What makes the Core i9-9900KS different from its predecessor are its base and turbo boost clocks, which are rated for 4GHz and 5GHz across all-cores, respectively, with enhanced binning of the chips to meet its performance criteria. The Core i9-9900KS is arguably the fastest processor available right now for single and lightly-threaded workloads, and offers the highest performance in gaming and graphics tests. In more heavily-threaded workloads that can leverage all of the additional processing resources available in a 12-core CPU like the Ryzen 9-3900X, however, the 8-core Intel Core i9-9900KS doesn't fare as well. It did catch AMD's 12-core Threadripper 2920X, which is based on the previous-gen Zen+ architecture, on a couple of occasions, however. Intel's new Core i9-9900KS desktop processor is available starting today at $513 MSRP.
This discussion has been archived. No new comments can be posted.

Intel Launches Core i9-9900KS 8-Core CPU At 5GHz Across All Cores More

Intel Launches Core i9-9900KS 8-Core CPU At 5GHz Across All Cores

Comments Filter:
  • Wow...
  • With a brand new CPU coming out from Intel, I wonder if all the known vulnerabilities that affect their existing chips will be fixed, eg. Spectre, Meltdown, Rowhammer, etc., etc.

    • Define fixed:

      The microcode fixes are baked in the microcode that comes from the factory from da 1.

      Nonetheless, it is a Coffee Lake refresh, which in turn is a Cannon Lake refresh, which in turn is a.... you get the idea...

      Of course, since the mitigations are already baked in, the performance you measure is the performance you'll get.

      But, being a coffee Lake refresh, the underlying architecture has not changed, which means, the vulnerability remain, which means that there may be more vulnerabilities lurking

    • Re:I wonder if the security issues will be fixed. (Score:5, Interesting)

      by gweihir ( 88907 ) on Wednesday October 30, 2019 @05:53PM (#59363626)

      Not anytime soon. Intel needs to completely redesign speculative execution for that. Any real fix at this time will come at a massive performance penalty and they are not going to accept that because then the lose the last (pseudo-) argument against AMD. AMD, on the other hand, has redesigned speculative execution and, in particular, branch-prediction that is probably not vulnerable in practice quite unlike this old Intel design.

      Expect it to take at least 3 years before the have something fixed and then expect it to not have good performance, as basically all current Intel CPUs get their performance from excessive detail-optimization (otherwise their old manufacturing process could not compete at all). Detail optimization is a slow, manual process. It will probably take several additional years after they have a new design out until that design is competitive again.

      • You have nailed it.

        Intel is fucked, and they have known that they are fucked, and they have known what their big mistake was, and they can't undo it. They admitted it in April of 2016 with massive layoffs and the announcement of their new "cloud strategy" which is code for "we will remain competitive in the server space longer than in the desktop space."

        Intels big mistake was obsessively trying to get 3D trigates to be practical at 10nm. They wasted half a decade on it, instead of half a decade solving

        • Re: (Score:2)

          by gweihir ( 88907 )

          You have nailed it.

          Thanks.

          "we will remain competitive in the server space longer than in the desktop space."

          With the 64 core Epyc CPUs and the continues Spectre issues, even that looks highly doubtful now.

          Intels big mistake was obsessively trying to get 3D trigates to be practical at 10nm. They wasted half a decade on it, instead of half a decade solving the yield problem, which is also a (lesser) problem at 14+++nm.

          I think the problem here is that Intel never had really good CPUs, but they had superior manufacturing for a long time. Then they suddenly had not and tried to fix that by concentrating their efforts on something nobody else was doing, probably in a mistaken belief in their own superiority. In essence, Intel learned absolutely nothing from the first time AMD overtook them. They would probably have had to

          • From a business perspective, Intel could definitely just have TSMC build their chips, and instantly crush AMD. Or, of TSMC want Intel to suffer, they could try e.g. Samsung, and offer them some profit advantage over their competitor TSMC.

            Then, because Intel are and have always been sneaky sleazebags, they could of course build their own new fab tech from scratch without keeping around all the pricy baggage of an old process node.

            Unless they did something like that already, and nobody wants to be Intels fab,

            • Re: (Score:2)

              by gweihir ( 88907 )

              From a business perspective, Intel could definitely just have TSMC build their chips, and instantly crush AMD.

              Business, maybe, but tech, no way. First, they do not have the tooling to adapt to a manufacturing process different from their own. Second, the results would be nowhere as good as what they derive from their current manufacturing process. The only way they can compete at all is by careful, manual detail optimization of their outdated design on their outdated manufacturing process. They cannot do that when they have the design manufactured by somebody else. So, no, they would not "crush" AMD, they would act

            • From a business perspective, Intel could definitely just have TSMC build their chips

              Your statement would be correct if you said "from a technical standpoint" ... but on the business end of things, TSMC is PurePlay and absolutely would not ever in a million years make chips branded by a competing non-PurePlay foundry, certainly not the epitome of vertical integration that is Intel. Intel is literally as far as it is possible to get from being able to rent time on a PurePlay fab.

              As far as Samsung, they would only do it if there was no conflict of interest, and Intel would only do it if yi

              • Re: (Score:2)

                by gweihir ( 88907 )

                So from a businesses perspective not either. They have truly painted themselves into a corner. The old combination of greed, arrogance and stupidity at work.

    • Which brand new CPU? Not this, it's existing silicon...

  • Fanboi-only prpduct (Score:3)

    by gweihir ( 88907 ) on Wednesday October 30, 2019 @05:47PM (#59363602)

    Nobody sane is going to buy this. Too expensive, not really better than the competition and still has the Spectre issue unfixed.

    • Too expensive

      not really better than the competition

      For all general purposes I agree there. But then this is hardly targeting a general purpose PC, throw a decent AMD in there.

      and still has the Spectre issue unfixed.

      And no one outside of a cloud services provider gives a crap.

      • Re: (Score:2)

        by gweihir ( 88907 )

        and still has the Spectre issue unfixed.

        And no one outside of a cloud services provider gives a crap.

        People are stupid about security, I agree. Especially as there are JavaScript based demo-exploits, so this is definitely not only a cloud thing. Just a matter of time until ordinary users will be hit.

        • You can bet your grandma, that Five Eyes, the FSB, China, the Mossad, NSO group and every similar enemy number one of humanity has and is actively using this already. Likely even before it got public. Remember: They have their own fabs, and in their league, you casually play around with dopant-level hardware backdoors!
          This js child's play to them.

          • Re: (Score:2)

            by gweihir ( 88907 )

            Don't think so. Exploits are probably still in development or very careful, very restricted use. But they will become generally available if they turn out to work reasonably well. May still take a while though.

        • Re: (Score:3, Interesting)

          by thegarbz ( 1787294 )

          People are stupid about security, I agree.

          Yeah people are. They get overly excited about something that is impossible to execute in practice and fall back on a javascript demo which doesn't work outside a lab and produces no meaningful result without detailed information about the target machine.

          Just a matter of time until ordinary users will be hit.

          I'm waiting patiently. I mean with proof of concept code, and a completely unpatchable fault everyone should have their dickpics shared online right? Wrong! The attack is insanely complex to execute compared to every other attack possible out there and requ

          • Re: (Score:1)

            by gweihir ( 88907 )

            I am an actual security expert. You are just a cretin with a big ego and no clue.

            • I am an actual security expert.

              Oh that makes sense. I guess you need to drum up a completely false sense of fear in order to get some business. To security "experts" everything is a high risk, your business depends on it. I'll keep that in mind when I read your future posts, honestly I thought you were a professional troll but it all makes far more sense now.

              But since I have no clue, educate me. Point out to me where the potentially most serious, prevalent, and still unpatched issue is being actively exploited. I'll wait. I mean every se

      • No matter how you look at it on both sides.. Once you hit the $500 range for CPU's it's no longer just general consumer market, and falls more so under "enthusiast" market if we're going to be brutally honest.
    • It does a good bit better than the 3900X in gaming and I'd argue that most gamers don't care about 12 cores or Spectre. This is targeted at PC gamers and I think it hits the mark fairly well. For anything else, I'd agree that AMD is probably the way to go.

      Gamers Nexus has a review. Gaming benchmarks start around 17 minutes in: https://www.youtube.com/watch?... [youtube.com]
      • ....and Gamers Nexus concludes that above 1080p gaming on the fastest video card you can possibly get, the extra mhz is meaningless as it doesnt translate into higher fps when you are already gpu bound.
      • Comment removed based on user account deletion

        • Yeah about that....if you game you can just get the Ryzen 3600 which is $300 cheaper which means you can get most of your system done for less than the cost of the Intel CPU alone.

          I play quite a few games without an issue on my Intel i7 6800K (Broadwell) CPU (overclocked at 4.25 GHz)... and the Ryzen 3600 is 30% faster on average. So yes, a Ryzen 3600 would suffice in all cases except niche ones (such as "progamerers" playing CS:GO at 500 FPS).

          The reason I can't upgrade my PC at the moment is the (un)availability of enough PCI Express lanes. My current CPU has 28 lanes, barely enough for the hardware I have thrown in my PC. All Ryzen and Intel CPUs except prosumer ones offer no more

          • Re: (Score:2)

            by Kokuyo ( 549451 )

            I'm confused, I thought Ryzens have 24 PCI-E lanes?

          • The new Ryzen 3000 chips all offer 24 PCIe lanes. The Prosumer and Server chips offer even more.

            • They offer 20 PCI Express lanes for devices (16X for graphics and 4X for nVME/SATA), because the rest of 4 lanes are the interconnect with the X570 chipset. Nevertheless, that's less than the 28 PCI Express lanes the 6800K offers.

          • Re: (Score:2)

            by AmiMoJo ( 196126 )

            Is AMD trolling Intel with names? Intel release the X299 chipset and then AMD announces their X399 platform. Intel go for X390 and AMD releases Z490.

            A used Threadripper and new mobo isn't too expensive these days.

          • Comment removed based on user account deletion
        • Yeah, I wouldn't recommend Intel to anyone who asked - that money would be better spent on a GPU. I bought a 9700K and it's great, but if I had to buy today, I'd go AMD. I'm glad they finally got their act together and I can only hope that they start to push Nvidia, because GPUs are completely out of control. I'm also hoping that Ryzens start getting into laptops. Of course, my 2011 Lenovo x220i refuses to die, so who knows when I'll be ready for an upgrade :P

        • Re: (Score:2)

          by Holi ( 250190 )
          "Ya see this is why Intel is fucked"

          Like they were when AMD launched the 1st 1 ghz processor, or was it when AMD beat them with 64bit? History proves time and time again while AMD can make great products and be great competition to Intel, they can't hold the lead.
          • I think that now since Intel had to abandon it's 10nm process and move toward 7nm like AMD did they'll have to play catch up. The bigger difference versus 20+ years ago is AMD isn't slowing down nor is it rehashing its current process. With that being said I believe AMD can hold the lead longer now until Intel releases something far better if they can release it before AMD's next big jump. Only Intel and AMD can dictate who maintains the lead now.

      • Re: (Score:2)

        by gweihir ( 88907 )

        It does a good bit better than the 3900X in gaming

        Nope. It does a small, insignificant bit better in some benchmarks. In actual gaming, the difference is irrelevant.

        • Agreed. as far as gaming goes.. AMD is on par with Intel at best so there is almost no noticeable difference unless you're counting small FPS percentage gains for bragging rights/eSports. Buying the 3900x for better productivity on the other hand is a more sensible reason versus Intel's flagship 9900K/9900KS at the $500 price point.

        • Re: (Score:3)

          by jwhyche ( 6192 )

          Nope. It does a small, insignificant bit better in some benchmarks. In actual gaming, the difference is irrelevant.

          This is the truth. There are plenty of videos on youtube and articles that show comparisons between AMD and Intel. In similar class chips the difference is usually with in 2% to 5% in fps in favor of Intel. In gaming the difference is irrelevant but the price isn't.

      • Agreed. When I chose the 3900x I did so for the multi-core productivity. Being able to game almost entirely on par with the 9900K is a bonus.
    • This is what is referred to as an enthusiast product. If someone wants the fastest process at 5Ghz all core they'll buy it. There is pretty much a market for everyone.
  • Binning of the chips as in tossing into rubbish? As in, where's our free-and-open CPU alternative to Intel's and AMD's spy microcode?

    • Feel free to buy the design software and create your open-source CPU which violates nobody's patents. Then use your crate full of millions of dollars to make masks and a fab run, and hope that there will be no errors requiring more millions for new masks and fabrication (fat chance).

      Whatever their flaws, the designers at AMD and Intel are not idiots. Bleeding edge general purpose CPU design is difficult and expensive.

      • RISK-V is the keyword, Alibaba has even made a 16 core 2.5GHz CPU on it though of course nobody is building computers aimed at the end user out of it. ARM laptops are a thing though, closed IP but not x86 at least.
    • Binning as it's all the same silicon coming off the assembly line, some going to top shelf bin, others the bottom shelf bin. AMD has taken it to the limit, their entire processor lineup is made out of the same damn Zen 2 chiplet branded and packaged into different processors based on how fast it will run and how many working cores it has.
  • I do rendering, raytracing, and scientific computing, so the lightly threaded workload isn't my thing.
    • 4 More of AMD's Cores
    • I dont think that its a thing for anybody that has thought about it.

      For those that doubt... ask yourself... what do you wait for? What processing is annoyingly slow?

      It isnt single-threaded stuff. All of it is multi-threaded stuff. Encoding videos, compiling code, rendering, ... thats the stuff you sit there waiting for. If you want to spend less time waiting, you buy the most cores you can for the money.

      • "His name was James Damore"

        His big old nose you'll see no more,
        He blamed his bullshit on autism,
        More likely it was backed up jism,
        Good fucking riddance James Damore

      • I think its because people want to argue fastest versus most productive even if it means 4% faster on FPS for gaming on certain games. AMD's 3000 series chips are now the best overall chips on both gaming and productivity at a much better cost to performance ratio.

  • Sounds like just the chip for the long cold winter months, or if you need to fry an egg. That 14nm+++++ tech you know.

  • Intel launches AMD FX-9590, but way too late (Score:3)

    by Jody Bruchon ( 3404363 ) on Wednesday October 30, 2019 @11:20PM (#59364552)
    AMD already beat you to this punch a long time ago with the FX-9590. Sure, the Intel chip is a newer chip so it's going to be a lot faster, but that 8-core 5 GHz nerd wank crown has long since been claimed. I don't see why anyone would spend $500 on a CPU like this when the price-to-performance ratio is massively better on other chips such as, say...AMD's high-end chips.
    • To be fair, Intel is winning in the sub-$100 market now [cpubenchmark.net] with the Xeon E5-2620 price slash from its former over $400 to its current under $70.

      Intel has never dominated the sub-$100 market before! This is groundbreaking!

      I wonder how many warehouses are full of those for the price to drop so much. Selling them at a loss for sure.
      • I wouldn't say groundbreaking as older Xeon chips are being re-sold well below MSRP ad they age. The real issue is the cost for Xeon boards and how the Xeons age versus the chips released 2 years ago. They're good for a low-end VM server though.

    • but that 8-core 5 GHz nerd wank crown

      ...is owned by Intel since IPC is actually a thing and ultimately peak GHz only matters within a specific architecture.

      • LOL modded troll for correcting an absurd fanboi statement. Yep they give modpoints to just about anyone these days.

  • Almost the same clocks on all cores can be achieved on the 9900K. This is what the 9900K was supposed to be and is just a product rehash with a factory overlclock on all cores. The only way this would be a more attractive buy is if the IPC was improved at least 15% at a minimum. Overall this is more of the same on the Intel product line in the CPU space.

  • The Core i9-9900KS is arguably the fastest processor available right now for single and lightly-threaded workloads

    Argue all you want you are still wrong. 9900KS offers exactly no single thread performance gain over 9900K, your busiest thread runs on the fastest core which for both is exactly the same thing running at exactly the same 5GHz. The only difference for KS is that all the other cores are also the same 5GHz instead of being limited. That improves multi thread performance a bit, but not enough to catch up to AMD CPU-s that have an advantage of a node jump. And Zen 2 still beats it all in single thread performan

Slashdot Top Deals

Beware of all enterprises that require new clothes, and not rather a new wearer of clothes. -- Henry David Thoreau

Close