Rumor: Info on upcoming Coffee Lake CPUs leaked

It's easy for us enthusiasts to forget that the high-powered CPUs we favor aren't even a majority of the market. Intel still has a whole bundle of desktop Coffee Lake CPUs to launch, and it looks like they're on the way sooner than later. Videocardz has a list of chips that either popped up on Amazon for pre-order or were found in listings on Asus' website.

All the new chips have TDPs of 65 W or below and fill out the remainder of Intel's eighth-generation desktop Core lineup. Like the sixth- and seventh-generation CPU series, the low-power models ending in T are rated for a 35 W TDP.

Upcoming Intel CPUs Base Clock Cores/Threads L3 Cache TDP Revision
Core i7-8700T 2.4 GHz 6/12 12MB 35W U0
Core i5-8600 3.1 GHz 6/6 9MB 65W U0
Core i5-8600T 2.3 GHz 6/6 9MB 35W U0
Core i5-8500 3.0 GHz 6/6 9MB 65W U0
Core i5-8500T 2.1 GHz 6/6 9MB 35W U0
Core i5-8400T 1.7 GHz 6/6 9MB 35W U0
Core i3-8300 3.7 GHz 4/4 8MB 62W B0
Core i3-8300T 3.2 GHz 4/4 8MB 35W B0
Core i3-8100T 3.1 GHz 4/4 6MB 35W B0
Pentium Gold G5600 3.9 GHz 2/4 4MB 54W B0
Pentium Gold G5500 3.8 GHz 2/4 4MB 54W B0
Pentium Gold G5500T 3.2 GHz 2/4 4MB 35W B0
Pentium Gold G5400 3.7 GHz 2/4 4MB 54W B0
Pentium Gold G5400 3.7 GHz 2/4 4MB 58W U0
Pentium Gold G5400T 3.1 GHz 2/4 4MB 35W B0
Pentium Gold G5400T 3.1 GHz 2/4 4MB 35W U0
Celeron G4920 3.2 GHz 2/2 2MB 54W B0
Celeron G4900 3.1 GHz 2/2 2MB 54W B0
Celeron G4900T 2.9 GHz 2/2 2MB 35W B0

Although it's good to have the full information on hand, there's not much earth-shattering news here. Possibly the most curious detail is that some of the CPUs in Asus' list are marked down as being the U0 stepping, while others are the B0 stepping. It's not clear what the difference implies given that the specifications we can compare appear to be more or less identical, save for the single exception of the U0-revision Pentium G5400 and its peculiar 58-W TDP. It seems unlikely that these chips have in-silicon mitigation for Meltdown and certain Spectre variants.

Comments closed
    • BobbinThreadbare
    • 2 years ago

    The i3-8300 looks like it might be a screamer for gaming. Almost as much cache as the i5s, 4 threads at very high clock speeds

      • MOSFET
      • 2 years ago

      I can confirm that the i3-7300 is a very quick CPU. It’s a somewhat noticeable boost over an i3-6100. I imagine the i3-8300 is an even bigger jump.

    • evilpaul
    • 2 years ago

    So when are the accompanying chipsets coming out?

    • Ninjitsu
    • 2 years ago

    Oh god the header image XD

    • Chrispy_
    • 2 years ago

    The 8700 will still be the cheapest way to get more than 6 threads, so despite the increase in core count, Intel have taken a step backwards by removing the 4C/8T options from the market. Unless people care about gaming that is highly IPC-dependent, AMD are still offering 8/12 threads for significantly less money than intel’s 4/6 thread options.

      • Anonymous Coward
      • 2 years ago

      Cheapest way I guess, but with those modest base clock speeds I’m not sure its very appealing. How many people this side of the server room are interested in throughput per watt?

      • Andrew Lauritzen
      • 2 years ago

      Even with Intel’s HT though 6c6t is still better than 4c8t unless you are intentionally doing something poor with thread pools and synchronization. You should never be getting 50% perf increase out of hyperthreading.

      And AMD’s SMT is still not quite as good as Intel’s, so you’re definitely not getting that level of scaling there.

      So sure, it would be nice if there were more 6c12t options, but 6c6t is still (quite a bit) better then what we’ve had in the past.

    • HERETIC
    • 2 years ago

    Best guestimation on that stepping-
    Intel would normally have 2 dies-with broken i7 becoming i5’s
    and the smaller die,i3 and broken bits becoming pentium and celeron.

    So my theory is those”UO”stepping pentium are broken broken i7’s………………

    • Krogoth
    • 2 years ago

    I’m curious to see if some of those lesser 4-core SKUs end-up being binned Coffee Lake silicon instead of re branded Kaby Lake silicon.

      • thecoldanddarkone
      • 2 years ago

      I personally don’t care about the lesser models. The only ones I’m interested in are the 6 core and above. It’s likely I’ll look at the 8500t / 8600t for deployment.

        • Krogoth
        • 2 years ago

        “4-Core Coffee Lake” would have dark silicon which would make it easier to manage thermals with an aggressive overclock.

    • The Egg
    • 2 years ago

    Honestly, what’s the point of having multiple SKUs with a measly 100mhz between them? When you’re up over 3.0ghz, that’s almost a rounding error.

      • DPete27
      • 2 years ago

      You are aware of Intel…right?

      GPU partners do the same thing IMO, sell 8 different GPUs that carry a <5% difference in clockspeed from highest to lowest. They change up just enough other things to segment their market (less heatsink fins, 1 less heatpipe, etc etc) instead of benefiting from economy of scale.

        • The Egg
        • 2 years ago

        Yeah…I’d argue that’s pointless too, but those are just card partners offering different levels of factory OC. It’s not the same as NVidia having a model number 1075 Ti with only a 3% clock difference.

        • jihadjoe
        • 2 years ago

        It’s just a bit funny because Intel stayed away from doing just this on their desktop line. There was little point really, because just about every 8700k will make it to 5GHz so all they needed was a locked and an unlocked SKU.

        I guess binning gets super serious when it comes to squeezing out parts that will do that extra 100MHz under 35W?

      • Krogoth
      • 2 years ago

      The nature of how yields work. It isn’t just clockspeed, it is cache size, core count as well as voltages.

        • The Egg
        • 2 years ago

        You mean except for the:
        G4900->4920
        G5400->5500 (both T and regular variants)
        G5500->5600
        i5 8500 -> 8600

        Where it’s just 100mhz and nothing else.

      • evilpaul
      • 2 years ago

      Charging $20 more times many thousands of chips sold seems like a pretty nice point.

      • psuedonymous
      • 2 years ago

      Well, you have the Ryzen series where the processors are separated by in-practice 0MHz, as all hit the turbo ceiling of 4GHz. How much separation you can reasonably achieve in SKUs depends on how hard you’re pushing the process, and what that process can actually achieve in practice. If you’re getting good yields and everything is ending up in the upper binnings, then you’re left with the choice of artificially capping performance (skewing down perf/$) or ending up with SKUs that are very close to each other. I can’t think of many consumers that would prefer the former over the latter.

    • End User
    • 2 years ago

    [quote<]It's easy for us enthusiasts to forget that the high-powered CPUs we favour aren't even a majority of the market.[/quote<] I'm schooled on that point almost every-time I post about a product that is over $5.

      • Wirko
      • 2 years ago

      Let me guess, are you a student at the Richard M. Stallman University?

        • tay
        • 2 years ago

        Where nibbling on your toe jam is just an advanced stage of eco-conservation and recycling.

    • derFunkenstein
    • 2 years ago

    The i7-8700T is interesting because that should give us a hint at what’s in store for notebooks. 6-core, 12-thread CPUs at 35W would be pretty great. Too bad there’s no turbo listed but I bet with lightly-threaded loads it’ll hit 4GHz.

      • DavidC1
      • 2 years ago

      Yup. 4GHz.

      I doubt it can do anywhere near that with all 6 cores active due to the TDP. Fortunately, the notebook ones will be available in up to 45W TDP, and that will help. Leaked performance seems to be 30-40% over current 45W quad core ones.

      The 14nm++ does a real good job.

        • Usacomp2k3
        • 2 years ago

        I’d love something to replace my i7-7700HQ. It turbo’s up to ~3.6Ghz when running all 4C/8T maxed out.

          • DPete27
          • 2 years ago

          Well, considering the currently available i7-8650U does 4.2GHz (max turbo) on 4c/8t on a 15W TDP…

            • magila
            • 2 years ago

            Turbo clocks do not guarantee that the CPU will actually run that fast with all cores under load. In fact I that 8650U is not going to be running anywhere near 4GHz will all four cores under load precisely because of that 15W TDP.

            It’s important to remember what the TDP ratings and turbo clocks mean and how they are determined. Intel sets a CPU’s TDP based mostly on what sort of cooling solution they are targeting. That TDP is then programmed into the CPU such that it will throttle the operating frequency if power draw exceeds the TDP. Turbo clocks are basically the maximum frequency the CPU can run at _if_ no other limits (power, temperature, etc) are exceeded.

            The 8650U is a pretty extreme example of this. Four fully loaded cores running at 4GHz would vastly exceed 15W, so in this situation the CPU will reduce its frequency until power draw reaches the target. In practice the frequency will depend on the workload, so the only way to know what you will actually get is the run your work and see what happens. The base clock represents the frequency of the worst case workload, which in the case of the 8650U is 1.9GHz. That gives you an idea of just how large the possible range of frequencies is.

            • smilingcrow
            • 2 years ago

            Great way to talk down to a knowledgeable audience and miss the context of the original question!

            • magila
            • 2 years ago

            Well apparently DPete27 didn’t already know that because the post he was replying to talked about frequency with all cores “maxed out” and the 4.2GHz max turbo on the 8650U is completely irrelevant in that context. So I made a post explaining why it’s irrelevant and somehow I’m the one in the wrong? Ok then I guess.

            • Anonymous Coward
            • 2 years ago

            Well, around here I give people the benefit of the doubt when they use ambiguous phrasing on a well known matter like the nature of turbo clocks.

            • jihadjoe
            • 2 years ago

            The good thing is apparently the Windows performance slider gives the user pretty good control over how much power the 8650U is allowed to use, and in this case it can draw up to 32W over 2s, then 30W for 10s, eventually scaling down to a sustained 25W. Not a big deal and most not-so-thin notebooks will happily sustain that amount of power draw.

            Sure it won’t maintain a sustained 4.0GHz across all 4 cores. It’s not supposed to anyway because the all-core turbo speed is 3.9GHz, but given 25W to play with [url=https://www.reddit.com/r/Surface/comments/7kbqhj/surface_book_2_135_i78650u_analysis_throttling/<]it does sustain 3.5GHz over an entire Cinebench run[/url<].

      • Welch
      • 2 years ago

      That is a very light base clock, 2.4ghz. it would explain the very low TDP. I’m going to guess this is your flagship chip for UltraBooks. Turbo perhaps on some but not all cores to somewhere in the ballpark of 3.6ghz. That is my guess. Unless they do 4ghz on a few, and 3.6 for all, thermal headroom allowing.

    • ronch
    • 2 years ago

    Coffee Leak.

      • dodozoid
      • 2 years ago

      Better Coffee leak on your pants than urine leak…

      • DavidC1
      • 2 years ago

      Coffee Late?

      • willmore
      • 2 years ago

      Coffee Stain

Pin It on Pinterest

Share This