Rumor: Turbo Boost could come to Cannon Lake Core i3s

Among all of the market segmentation methods Intel has used to separate Core i3, i5, and i7 mobile processors over the years, the one constant was that Core i3 chips lacked Turbo Boost support. That's held true even as Core i5 and Core i7 CPUs have grown into quad-core, eight-thread parts in 15-W power envelopes.

That one major point of distinction could be about to change if new rumors prove correct. Online publication Laptop Media claims to have new information on a future eighth-generation Core i3 mobile part, the Core i3-8130U, that CPU investigator InstLatX64 claims will be among the first Cannon Lake CPUs produced on Intel's oft-delayed 10-nm process. InstLatX64 also points out a new entry in the Geekbench database featuring a Core i3-8121U processor that's purportedly from the same product family. Both parts could mark the first time Turbo Boost will make its debut on Core i3s—a move in keeping with Intel's reimagining of generational improvements as a measure of delivered performance instead of any particular silicon advance.

The Geekbench entry that InstLatX64 found purports to describe an as-yet-unreleased Core i3-8121U. This chip has a low base clock, an unusual CPUID string, and an extra megabyte of L3 cache compared to a seventh-generation Core i3 part. InstLatX64 takes to mean that the chip has a new architecture. InstLatX64 believes the extra megabyte of L3 is especially indicative of fabrication on Intel's 10-nm process. Gerbils that have been following Intel's moves from home will recall that the company's previous schedule of two architectures on each process node has broken down at 14 nm, and that Broadwell, Skylake, Kaby Lake, and Coffee Lake parts have all been fabricated on essentially the same process node.

If Laptop Media's information is to be believed, on the other hand, the eighth-gen Core i3-8130U mobile part it uncovered will slot in below the well-received four-core, eight-thread Core i5-8250U. This chip will reportedly sport Intel's familiar UHD Graphics 620 IGP, which is a simple rebrand of the HD Graphics 620 graphics processor found in the first round of Kaby Lake mobile parts. The site reports a 2.2 GHz base clock for the new Core i3 chip, down from the 2.7 GHz clock of the Core i3-7130U from the previous generation. The new chip will reportedly boost up to 3.4 GHz, however, a value that could go a long way toward explaining the increase in TDP from 7.5-15 W to 10-15 W. The amount of L2 cache increases from 3 MB to 4 MB, in keeping with the changes that InstLatX64 notes.

The evidence for either of these claims seems thin, but we do know that Intel has been talking up progress with its 10-nm process of late. Laptop Media has not published any details about the source of its Core i3-8130U specifications, and InstLatX64's claims are similarly unsubstantiated. The rumors look reasonable enough, however. We hope to hear more about Intel's possible new lineup of Core i3s—if they exist—soon.

Comments closed
    • johnsmith26
    • 2 years ago
    • Growler
    • 2 years ago

    Sweet! I always wanted to jump a computer over a train!

    [url<]https://media1.tenor.com/images/992ace05caa1ba90852787824f17d1da/tenor.gif?itemid=5597096[/url<]

    • mczak
    • 2 years ago

    FWIW the turbo rumor for mobile i3 cannonlake isn’t really much of news…
    Since even the kaby lake r mobile i3 (i3-8130u) are expected to feature turbo (and also 4MB of L3).
    And of course it’s pretty much the logical thing to do: previous to kaby lake r, the mobile (U) cpus (regardless of i3,i5,i7) featured 2 cores / 4 threads. With pretty much the only difference being the core i3 doesn’t support turbo (so you got quite a bit less single-thread performance, multi-thread performance the difference was smaller) compared to the core i5, and on top of that the core i7 have a bit more cache and a bit higher frequencies (neither of which really made much of a difference).
    But with KBL-R, both the i5 and i7 now have 4 cores / 8 threads. Again, there’s barely any difference between them (6/8MB L3, minimal max clock difference). However, it would not make sense for the i3 to also have 4 cores but not turbo (the base clock needs to be very low to fit into 15W with 4 cores). Hence instead it’s 2 cores / 4 threads but this time with turbo – without turbo it would be really unattractive (mobile i3 seem to be quite rare in any case already). So, this time around, you get similar single thread performance compared to i5, but lower multithread performance (and apparently you also get 4 MB of L3 – I’m not even sure if it’s a separate die or just the same as the i5/i7 with half the cores disabled).

    • DPete27
    • 2 years ago

    Availability is always the most important keyword when talking about mobile SKUs. It doesn’t benefit anyone if a CPU doesn’t actually show up in any products.

      • MOSFET
      • 2 years ago

      This could be nice in a NUC. The first, and last, i3 NUC I bought was a Broadwell with i3-5010U at 2.1 GHz. 2C/4T isn’t a problem (especially not 3 years ago when purchased) and the IGP isn’t a problem, and the DDR3-1600-LP isn’t even a problem, but the lack of turbo (bursts over 2.1 GHz) is a real killer to system smoothness. Since then I’ve gone exclusively i5 and i7 NUC.

    • derFunkenstein
    • 2 years ago

    The conclusion drawn by InstLatX64 seems kind of hokey. It’s more likely a Kaby Lake refresh CPU in the same vein as the Core i7-7600U, which also had 4MB of cache. Seems like quite a stretch to claim this is a 10nm part.

      • chuckula
      • 2 years ago

      I also like how the Model number on the “new” part is noticeably lower than the one of the Kaby Lake part.

        • derFunkenstein
        • 2 years ago

        That’s something I found weird in this article.
        [quote=”Stately Wayne Manion”<] an unusual CPUID string[/quote<] The string didn't seem all that unusual to me. [quote="normal CPUID"<]GenuineIntel Family 6 Model 142 Stepping 9[/quote<] [quote="quirks mode CPUID"<]GenuineIntel Family 6 Model 102 Stepping 3[/quote<] I guess it'd be downright weird for a 10nm CPU, though.

      • Jeff Kampman
      • 2 years ago

      If you dig into the Geekbench results there is something going on that suggests this isn’t just a respin of a Kaby part. Accelerated AES, matrix math, and especially fast Fourier transform results are all in line with a more capable core in some ways. It’s not much but it’s something.

        • chuckula
        • 2 years ago

        Or it’s just a higher-clocked Kaby/Covfefe lake part running a microbenchmark.
        The twitter link you have claims the “new” part has a turbo-boost of 3.4GHz that’s about 25% faster than the 2.7GHz non-turbo Kaby Lake part.

        Interestingly enough, the “new” processor posts a single-core Geekbench score is almost exactly 25% higher than for the old Kaby Lake part. Ignore the lower 2.2GHz base clock because in a microbenchmark the CPU is running flat-out practically the whole time.

        Or the results are faked.
        Geekbench is definitely capable of being fooled.

        • derFunkenstein
        • 2 years ago

        Yeah, maybe. Here’s a different comparison: the Latitude E5570 is my work PC, with the mystery PC as the comparison.

        [url<]https://browser.geekbench.com/v4/cpu/compare/6778890?baseline=6246666[/url<] My Dell gets thumped in single-threaded tests, pointing to much higher turbo clocks (which I don't dispute, since I already think this is a Kaby Lake part). For the most part the multi-threaded tests are a wash. I can't explain the two tests you pointed out, though. [url<]https://browser.geekbench.com/v4/cpu/search?q=Intel+Core+i3-8121U[/url<] Bunch of results. SFFT and SEGMM are always abnormally high. If it's getting faked (as chuckula suggested), it's getting faked consistently.

          • Zizy
          • 2 years ago

          It isn’t *that* abnormal. These comparisons are of laptops after all. Add better cooling and you have one winning over the other, especially in the heavy tests that are CPU and TDP limited.

          Compare this chip to the i5-8250U in the first laptop you can find:
          [url<]https://browser.geekbench.com/v4/cpu/compare/6475459?baseline=6246745[/url<] This i3 wins ST AES and FFT. It gets relatively close in some MT tests including AES and FFT, but there is nothing way out of ordinary. Now, compare two laptops with i5-8250U. The first two on the list. [url<]https://browser.geekbench.com/v4/cpu/compare/6446133?baseline=6475459[/url<] This time, Acer (comparison above) shows suspiciously large lead in MT DGEMM, as if the CPU wasn't the same 🙂 So the conclusion is simple - nothing is weird enough to warrant guessing this is a mysterious 10nm CPU. If the "mysterious" i3 CPU is capable of turboing higher in ST and has a bit better memory, perhaps it even has better cooling able to run at 25W unlike other laptops where it gets throttled to 15W faster... all these things will show in results, especially on the DGEMM and FFT tests, which stress the CPU and push it to the TDP limits. And hammer the memory as well, so faster memory helps too. Not sure what AES stresses, likely the same things.

            • derFunkenstein
            • 2 years ago

            That’s the reason I linked all the “i3-8121” results. They’re quite consistent. The SGEMM and SFFT results are all basically within +/- 2% of each other.

            It’s almost like it’s the same core but with some extra AVX throughput, or maybe AVX-512. I’m still not willing to say this is a 10nm CPU (though I’d love to be proven wrong), but those results are a little weird and consistently so.

        • mczak
        • 2 years ago

        Yes, the assumption this is cannonlake looks reasonable to me. geekbench supports avx-512 for two subtests – gemm and fft, and these are exactly the two which show the largest gains outside aes (forget the memory scores, which seems to be the result of single channel vs. dual channel). And Cannon Lake cpus (even the non-server versions) are supposed to support avx-512.
        As for AES, we know Cannon Lake supports a couple new instructions for encryption (SHA) – albeit these won’t get used by geekbench, it’s quite reasonable to assume AES got beefed up as well (probably not so coincidentally, ryzen cpus blow away kaby lake there on a per-core and per-frequency basis…).

    • chuckula
    • 2 years ago

    You know what else is in every Cannon Lake i3?

    Unicorn Tears.

    And $500 worth of Unobtanium.

      • Wirko
      • 2 years ago

      Do you often confuse Cannon Lake for BMW?

        • chuckula
        • 2 years ago

        No. I can buy a BMW*.
        I can’t buy CannonLake.

        * After robbing a sufficient number of banks.

          • christos_thski
          • 2 years ago

          It’s kind of funny how expensive beamers are in the USA. In europe many beamers are within the range of a middle-class household.

          And they’re not even worth that. German cars are overrated. My girlfriend drives a Mercedes and my brother drives a Ford Focus. I’d pick the Focus any time of the week.

            • willmore
            • 2 years ago

            BMWNA is a bunch of jerks. Long time BMW owner in NA.

            • bthylafh
            • 2 years ago

            Is the stereotypical BMW driver in Europe a wanker? Seems to be so here in the States.

            • smilingcrow
            • 2 years ago

            Seems that way in the UK according to folklore but can’t say I’ve noticed myself.

            • cygnus1
            • 2 years ago

            They’re not wankers. People just forget that German cars come with a built in right of way and a drivers license exemption from any blinker usage.

            • DrCR
            • 2 years ago

            [url<]https://c1.staticflickr.com/3/2334/2136703337_f303dd017e.jpg[/url<]

            • christos_thski
            • 2 years ago

            I wouldn’t go so far as to call them flaming assholes, but yes, the average bmw driver does have a reputation for aggressive driving and jerkish behaviour in europe as well. Having said that, I have known very responsible bmw drivers. I do think that on average they’re somewhat worse, though ; ie the stereotype is not completely unfounded..

            • Wirko
            • 2 years ago

            Just google “bmw drivers”. It’s telling enough.

            • Waco
            • 2 years ago

            “Audi drivers” are worse. 😛

            • kuraegomon
            • 2 years ago

            Which Mercedes? And which Focus? What do you dislike about said Mercedes? What do you prefer about said Focus? If you’re going to state a strong opinion like that, it would be helpful to provide some context.

            • christos_thski
            • 2 years ago

            It’s the b class 1700cc mercedes, and any car manufacturer that respects his brand would never put such an underpowered engine under the hood of this heavy car. It has the lousiest acceleration ever, for a car of this price range, and I’m no speed demon either. It just handles worse than the Focus, which is a 1600 engine European sedan 2009 model. The Focus is nimbler, has a lot more extras and -amazingly enough- seems like a more “premium” car all around. The Mercedes is bog slow, lacks a lot of accessories (the hell… this thing didn’t even have bluetooth standard) and doesn’t handle as well.

            And all that is not even accounting for yearly service costs, where Mercedes robs you blind.

            Or Mercedes’ useless “start and stop” system on a stick shift. Only use of this thing? Paying double the price for a new “start and stop compatible” battery when the existent one peters out.

            • kuraegomon
            • 2 years ago

            Right. That definitely sounds like a European experience alright – where the Fords are (often) better than in North America, because competition – and the Mercedes/BMW/Audi hydra are often worse, because they have to fill the entire spectrum, including entry-level. Their base chassis, body panels and sound-deadening have to be sufficient for the up-market models, but their engines and interior trim are well, inferior at the low end. The low-end models don’t really make it across the water to North America…

            At any given price-point on the low end, the Germans have a tough time competing because of their base platform cost.
            Once you get to a price point where they can put the engine, suspension and interior upgrades in, the experience gets _much_ better.

            • cygnus1
            • 2 years ago

            Yep. There’s a reason I saw BMWs and Mercede’s as commonly taxi’s in Germany when I was there. They were honestly not even as good as the Ford and Toyota taxi’s I’ve been in in the US.

            • frogg
            • 2 years ago

            “German cars are overrated” … exactly ; And “Made in Germany” is highly overrated nowadays ; it is outsourced in other parts of Europe, and German workers are replaced by fresh migrants. I’ve had bad experience with Audis recently. Now i’m buying Toyota and i’m very satisfied.

        • Pancake
        • 2 years ago

        No, I think you’re confusing unicorn tears for monkey tears.

Pin It on Pinterest

Share This