Rumor: Core i3-8350K and Core i3-8100 CPUs pop up

When we talked up the rumors about Coffee Lake CPUs last week, there was a remark about a claimed "Core i3-8300"—a would-be quad-core, 65W CPU part. At the time we didn't report on it, since the PCEva forum poster himself said the information was shaky. However, today a poster on the massive Chinese PTT.cc forum posted the image below describing purported Core i3-8350K and Core i3-8100 CPUs. Although the table looks fairly complete, we have no way to judge its legitimacy or origin, so take it with as many grains of salt as your blood sodium level will tolerate.

The rumored CPUs are supposedly quad-core units without Hyper-Threading—a combination that would be a first for Intel's Core i3 line. However, these chips are similar to existing Core i3 CPUs in that they also lack Turbo Boost support.

It's hard to imagine the lack of Turbo Boost being a major concern for these CPUs, though. Even the alleged Core i3-8100 is listed with a 3.6 GHz base frequency, and the overclockable Core i3-8350K would run at 4 GHz. If the info is solid, that CPU would be in rarefied territory so far occupied only by a handful of other K-suffix units. The Core i3-8350K also supposedly has a full 8MB of L3 cache, which would make it unique not only among Core i3s but also among quad-core CPUs without Hyper-Threading. Meanwhile, the Core i3-8100 is listed as having 6MB of cache like many past Core i5 CPUs.

Everything else in the table is more-or-less believable. The claimed 65W TDP of the Core i3-8100 is higher than the TDPs on past Core i3 CPU, but not out of line for a 3.6 GHz quad-core CPU. Likewise, the 91W TDP of the Core i3-8350K is pretty high, but again reasonable for a 4 GHz quad-core unit—especially an overclockable one. Notably, both of these CPUs are listed as having ECC support. Some existing Core i3s support ECC memory, but the overclockable Core i3-7350K does not.

Overall, these rumored CPUs would look a lot more like what we'd expect from Core i5s than Core i3s. If these rumors pan out, that leaves us wondering what the Coffee Lake Core i5 lineup will end up looking like. Given the rumors we've collected about upcoming hexa-core Core i7s, it's possible the Core i5s will end up being four-core, eight-thread CPUs like the Core i7s of today, or even six-core, six-thread units. At any rate, it's not going to be long until we know more, seeing as Intel's set to debut its 8th-generation Core CPUs on August 21.

Comments closed
    • freebird
    • 2 years ago

    What’s up with Intel naming one of their i3s after AMD’s FX-8350?!?!? All they did was remove the FX in front and add a K to the end…

    Nice “originality” :0 Probably gonna compare it against the FX-8350 in benchies too I bet… 😉

      • Klimax
      • 2 years ago

      What the hell are you talking about? Intel follows their naming convention for longer then Bulldozer exists. (For once they kept it!)

    • Chrispy_
    • 2 years ago

    Has Intel finally seen reason and decided that dual-core is only good for the lowest-budget parts at last? It always amuses me that a single CPU core makes up such a tiny portion of the die area in a modern Intel desktop CPU, and yet Intel relentlessly kept pricing quad cores high…

    [url=https://en.wikichip.org/w/images/thumb/2/27/kaby_lake_%28dual_core%29_%28annotated%29.png/650px-kaby_lake_%28dual_core%29_%28annotated%29.png<]Here's a Kaby dual-core die shot.[/url<] With fixed-area hardware like the memory controller, system agent, IGP, fixed-function media logic, Display I/O and System I/O logic making up about 80% of the chip, each pair of cores is only around 15% with L3 cache and the interconnect bringing the die-area per core to about 10%. In other words, it would cost Intel only 20% more die area and almost nothing extra in packaging, binning, testing, and other misc expenses to make only quad-cores and kill off dual-core desktop parts for good. Let's not forget that most Intel CPUs are best described as [i<]SoC-focused GPUs with CPUs bolted on as an afterthought.[/i<] Here's hoping that with Coffee Lake, Dual-core parts exist solely for ultra-mobile, low-power consumption reasons. Anything less than that should be the realm of quad-core Apollo Lake chips instead.

      • NoOne ButMe
      • 2 years ago

      How many dual cores do they sell a year from that die right now? 20-25% higher costs per die off the foundry for tens-of-millions of units?

      I doubt Intel is going to cut margins.

        • Chrispy_
        • 2 years ago

        They could drop the IGP from 24 execution units to 8 and 95% of their customers wouldn’t notice, saving them 25% or so of their die area. So no, this isn’t about margins on the dual-core die areas; It’s about the very high markup on the quad-cores.

        As long as they’ve artificially kept entry and mainstream models on dual-core due to a lack of viable competition, their quad-cores can be priced [b<]MUCH[/b<] higher relative to the dual cores because you're getting "twice" the CPU capability. AMD's incoming Raven Ridge APUs will really shake up the mainstream laptop market and entry level desktop market (or what's left of it at least). Competition is a wonderful thing, with quad core low-power AMD chips with superior IGP performance making dual-core Intel products with lousy IGP perfomance something that only appeals to those wanting power saving above all else.

          • NoOne ButMe
          • 2 years ago

          I think Intel doesn’t make anything under 24EU. They do units of 24EU I think..
          24/48/72 units.

          And all leaks to me seems that Intel rather cut margins than iGPU… odd

    • NTMBK
    • 2 years ago

    The power of competition.

    • ermo
    • 2 years ago

    Loved the picture accompanying the article on the front page.

    • Goty
    • 2 years ago

    Given the posted TDPs for these parts, does it seem to anyone else like these parts won’t be much of a leap forward in terms of efficiency? The i5-7600K is a four-core, four thread part at a 3.8GHz base frequency, and it shares the same 91W TDP. What sort of architectural changes are we looking at for Coffee Lake?

      • Beahmont
      • 2 years ago

      200 Mhz for the same TDP isn’t peanuts at this level of design given that frequency increases scale power in a worse than linear curve. Also I imagine there is still a fair amount to a lot of headroom to overclock the k chip.

      • maroon1
      • 2 years ago

      For comparison 65w AMD RYZEN 3 1200 has base clock of 3.1 (3.4GHz turbo)
      And 95w AMD RYZEN 3 1300X has base clock of 3.5GHz (3.7GHz turbo)

      i5 7500 beats 1300X in majority of benchmarks. So, both of those new i3 will beat the the top Ryzen 3 part

      • jihadjoe
      • 2 years ago

      Hmm a 91W i3… 4.8GHz base 5.2GHz boost?

        • auxy
        • 2 years ago

        Someone didn’t read the post… (*´艸`*)

          • jihadjoe
          • 2 years ago

          I did, but 4GHz base just doesn’t add up to a 91W TDP for 2 cores. 7700k has 4 cores at an even higher 4.2GHz base and stays within 91W.

            • Klimax
            • 2 years ago

            i3 has no Turbo boost.

            • Goty
            • 2 years ago

            The article you supposedly read literally refers to these parts as quad cores five separate times.

            • jihadjoe
            • 2 years ago

            Ok, skimmed. Downvote away.

    • Kurotetsu
    • 2 years ago

    What’s really interesting is that the Core-i3 8350K supports ECC. If I’m not mistaken, that would make it the first consumer-oriented (i.e. non-Xeon), quad-core, unlocked processor that supports ECC that Intel has ever produced.

      • ermo
      • 2 years ago

      Yeah, it’d likely make a rather decent NAS software-RAID CPU (FreeBSD + ZFS and the like).

      My home server is currently an AMD PhII 955 X4 BE w/4x4GB DDR3-1600 ECC with WD Red 4TB disks. It spends most of its time at 800MHz@1.0V.

      For use with encrypted disks, the i3-8100 CPU would definitely be a BIG upgrade. Then again, a similarly priced RyZen might perform just as well or better, so…

        • Beahmont
        • 2 years ago

        Ryzen does not support ECC. The parts are kinda there in the memory controller, but the mobo chipset can’t talk to that part of the controller and the traces are not there on the boards.

          • ermo
          • 2 years ago

          [quote=”AMD”<][b<]AMD_Robert:[/b<] ECC is not disabled. It works, but not validated for our consumer client platform. [b<]Q:[/b<] What does "validated" mean in this context? What sort of stumbling-block does that represent to those who want ECC? Will it still be possible to build ECC-enabled servers with consumer-grade (and consumer-price-range) hardware on the Ryzen platform? There are a significant portion of users who want ECC for their NAS/Homelab setups. [b<]AMD_james:[/b<] Validated means run it through server/workstation grade testing. For the first Ryzen processors, focused on the prosumer / gaming market, this feature is enabled and working but not validated by AMD. You should not have issues creating a whitebox homelab or NAS with ECC memory enabled. [b<]Q:[/b<] So the Ryzen has full ECC support, if I install a ECC memory, it would work in ECC mode, not non-ECC mode? [b<]AMD_james:[/b<] yes, if you enable ECC support in the BIOS so check with the MB feature list before you buy. [b<]Q:[/b<] Thank you for the answer! So, the AM4 platform / socket theoretically has everything to fully support ECC and it's only up to mainboard manufacturers. Is that correct? [b<]AMD_Robert:[/b<] Bingo.[/quote<] source: [url<]https://www.reddit.com/r/Amd/comments/5x4hxu/we_are_amd_creators_of_athlon_radeon_and_other/def6vs2/[/url<]

            • Klimax
            • 2 years ago

            Non-validated ECC is not exactly useful. How do you know it works correctly when there si no validation?

            • ermo
            • 2 years ago

            You trust your motherboard vendor to have validated their ECC implementation.

            If that’s not good enough for you, pay up for an AMD (or intel) validated platform. That’s the gist of it.

            • Markopolo
            • 2 years ago

            I don’t see anything in your quote that shows AMD is doing die level tests on the ECC hardware in the ryzen CPU… isn’t that also required?

            If not, how do you know the ecc hw is actually catching hw faults on you ram?

            • ermo
            • 2 years ago

            Under Linux, you can verify this by:

            1) Confirming that the Linux kernel EDAC module detects that you’re using ECC RAM, and
            2) Check that the system can correct errors by e.g. overclocking your RAM and checking that the EDAC module reports that the error has been caught and corrected (there’s a link to just such a test in the reddit thread).

            From what I understand, Gigabyte treats ECC-support as a product segmentation feature and only enables it on higher-end RyZen boards. ASRock I believe enables it on its Taichi boards. There are conflicting reports on the ASUS X370 Prime Pro boards. For MSI I haven’t a clue.

            Again, if you don’t trust the motherboard vendor’s validation, perhaps you’re better off purchasing officially validated ECC platforms from AMD or intel. Caveat emptor etc.

            • Markopolo
            • 2 years ago

            Thanks for the detailed response re how to check ecc works.

            To be clear: I don’t question the motherboard manufacturers validation.

            My question should have been worded: does AMD verify the ecc HW on each ryzen CPU despite claiming it is not validated or am I playing silicon lottery?

            our CI / build server is having trouble keeping up with our nighlies, and so is due for an upgrade… but our budget is tight. Ryzen 1700 would be such an improvement, and fits under that magic 1k usd where we can use discretionary spending instead of waiting for next calendar year (august 2018).

            • Klimax
            • 2 years ago

            You still need ECC validated on chip. Unvalidated ECC is useless. (Validation needs to be done on each produced chip!)

      • UberGerbil
      • 2 years ago

      Yeah, I’m guessing that’s an error or hopeful thinking. The combination of i3, “K” and ECC seems about as un-Intel as you can get.

        • Kurotetsu
        • 2 years ago

        You still don’t get Turbo Boost and you lose Hyper Threading. So there’s still some segmentation there to keep the marketers happy….

          • Klimax
          • 2 years ago

          2c/4t are less powerful then 4c/4t almost by definition. Why would you expect Intel to keep HT in i3 when they get extra 2 full cores?

    • SecretMaster
    • 2 years ago

    Good lord. Reading that press release by Intel reeks of marketing hubris.

    “Watch as two exciting moments align: the Great American Solar Eclipse and the unveiling of Intel’s most powerful family of processors for the next era of computing.”

    Unless Coffee Lake is a fundamental revolution in computing, there is no way I’ll ever remember that a near once in a lifetime event coincided with a processor unveiling. Even associating the two just seems absurd to me.

      • chuckula
      • 2 years ago

      Sounds about par for the course as far as these things go. Meaning dumb but about expected.

      Intel would have been better served to say: [quote<]Ever since the beginning of time, man has yearned to destroy the sun. We shall do the next best thing: block it out![/quote<] As for how big a deal it is... Intel is livestreaming the event, which AMD went out of its way to avoid for Vega's "launch" last month.

      • morphine
      • 2 years ago

      [quote<]Reading that press release by Intel reeks of marketing hubris.[/quote<] It's reasonably light as these things go. I take it you don't read a lot of PRs? (Please don't unless you absolutely have to).

      • ImSpartacus
      • 2 years ago

      This is pretty tame for marketing.

      They are always going to push the truth to it’s breaking point. That’s Intel’s and every other company’s marketing team.

        • Klimax
        • 2 years ago

        And what happens when either PR or products fail, one can just look at AMD. (Bulldozer, Vega for recent history)

      • freebird
      • 2 years ago

      I think they mean Coffee lake will be “eclipsed” by “Ryzen”

      • Kretschmer
      • 2 years ago

      Marketers gotta market; this is better than most.

    • chuckula
    • 2 years ago

    Take this story with a whole mountain of purported halides.

Pin It on Pinterest

Share This