Intel NUCs with Cannon Lake inside pop up at SimplyNUC

Intel's Core i3-8121U is the one and only Cannon Lake CPU to exit the company's foundries so far, and its implementation in actual products has been confined to a single Chinese-market Lenovo laptop. That could change soon with the NUC8i3CYSM and NUC8i3CYSN, two NUCs widely referred to as “Crimson Canyon” in rumors and leaks.

According to new product listings from SimplyNUC, the NUC8i3CYSN will come with 4 GB of soldered-down LPDDR4-2666 memory in a dual-channel configuration, while the NUC8i3CYSM will bump the memory to 8 GB. The systems are otherwise identically configured around a Core i3-8121U CPU.

Since Cannon Lake doesn't include an active integrated graphics processor, the Crimson Canyon NUCs will rely on a Radeon 540 discrete graphics processor with 2 GB of its own GDDR5 RAM. SimplyNUC will include a 1-TB mechanical hard drive with both NUCs in their base configurations. That drive occupies the NUCs' sole SATA 6 Gbps port. Builders and reviewers who want to make an expedition to Cannon Lake will also find an M.2 2280 slot for SATA or NVMe gumsticks.

On the outside, the NUCs have two HDMI 2.0b ports capable of driving two 4K displays at 60 Hz. The systems have two USB 3.0 Type-A ports on their front panels and two more of those ports around back. Intel includes a Gigabit Ethernet jack driven by one of its own i219V controllers, as well. A UHS-I SD card reader and an integrated headphone-microphone jack round out the systems' connectivity options.

SimplyNUC will sell the base NUC8i3CYSM for $574 with the aforementioned 1-TB hard drive and Windows 10 Home included, while the NUC8i3CYSN will run $529 with that same configuration. The company estimates the NUC8i3CYSM will ship in mid-September, while the lower-end NUC8i3CYSN will begin shipping in late October.

Comments closed
    • Anonymous Coward
    • 1 year ago

    Seeing soldered RAM and a spinning disk is only slightly more surprising than seeing the CPU & GPU. The only example in recent history of a discrete GPU being chosen as the cheapest option.

      • NovusBogus
      • 1 year ago

      It’s where it makes the most sense, though. A configuration like this would be ideal for a college student who wants some light gaming in a very small, suitcase friendly box. The spinning rust is dubious but whatever, it has room for an SSD for those interested.

      Soldered RAM sucks but RAM prices are goofy enough that I’ll give them a pass on it.

        • Anonymous Coward
        • 1 year ago

        I’d argue this makes no sense for college students, they should rather look at a laptop. There are plenty of them, at reasonable prices, that would eat this thing up. If you’re hauling around a big fat screen, pair it with something worthwhile. Even an AIO would destroy this, should it have a discrete GPU, which many do.

        • NTMBK
        • 1 year ago

        College students usually want a laptop, so that they can take it to lectures/lab/the library/tutorials/LAN parties.

    • Krogoth
    • 1 year ago

    Intel just give it up.

    You know that Covfefe Lake doesn’t stand a brew of a chance against AMD’s Ripmaker. Dud Lake will just misfire from a premature detonation!

      • chuckula
      • 1 year ago

      Nice try, but it would have been much better if you had just called this NUC proof that discrete GPUs are dying. Because it uses one.

        • Krogoth
        • 1 year ago

        Intergrated GPUs in small form factors are too mainstream!

    • Chrispy_
    • 1 year ago

    Radeon 540 – does that mean the RX 540 or something new for laptops and embedded GPUs?

    IRC the RX550 is Polaris 12 with 512 of a potential 640 shaders used – presumably for yield reasons. Perhaps the 540 is a 384-shader part to fit lower TDP requirements?

      • DancinJack
      • 1 year ago

      Compute Units – 8
      Boost Frequency – Up to 1124 MHz
      Max Performance – Up to 1.2 TFLOPs
      ROPs – 16
      Stream Processors – 512
      Texture Units – 32

      GDDR5 64-bit

      [url<]https://www.amd.com/en/products/graphics/radeon-540[/url<] No idea how accurate this is, but - [url<]http://gpu.userbenchmark.com/Compare/Nvidia-MX150-0-vs-AMD-Radeon-540/m332822vsm316090[/url<]

        • Chrispy_
        • 1 year ago

        Okay so it’s an RX 550 with a 59MHz clockspeed downgrade.

        Shame, since the RX 550 is already too slow to be worth bothering with. It’s a 60fps card in the lightest of 2015 esports titles. I know from experience that it can’t cope with anything newer unless you drop everything to sadness-inducing resolution and detail levels :'(

          • NovusBogus
          • 1 year ago

          It’s still gonna be a lot better than the low-power integrated you’d otherwise be getting with a system like this.

            • Chrispy_
            • 1 year ago

            Agreed, it just doesn’t change this to something that will handle AAA titles.

            It’s still going to be limited to very old games or lightweight casual fun – they’ll just run a bit better than the IGP would have done.

    • ronch
    • 1 year ago

    SimplyNUC. Hmm. Seems to me they’re trying to sound witty there.

    They’re not.

    • NTMBK
    • 1 year ago

    Buy one to benchmark, you know you want to!

      • chuckula
      • 1 year ago

      Interesting blurb from Anandtech about Cannonlake in a Lenovo.

      Here’s a highly parallelizable particle motion benchmark that Ian Cutress at Anandtech wrote himself [url=https://images.anandtech.com/graphs/graph11839/91507.png<]taken from last year[/url<] Here's what happened when AVX-512 support got added and [url=https://twitter.com/IanCutress/status/1021478710690623489<]it was tested with a Cannonlake notebook.[/url<] Y-cruncher also has some extremely preliminary Cannonlake information: [url<]http://www.numberworld.org/y-cruncher/news.html#2018_7_2[/url<] Obviously a two-core part won't dominate in something like y-cruncher but it's interesting to see normalized per-core performance values.

        • NTMBK
        • 1 year ago

        Wow, nice. AVX-512 adds a bunch of features to make vectorization easier, looks like the Intel compiler fully vectorized his algorithm.

        • tipoo
        • 1 year ago

        [quote<]To confirm, AVX512-enabled code on a 2.2GHz 2C outperformed non-AVX code on a 3.4GHz 18C.[/quote<] Crazy.

          • uwsalt
          • 1 year ago

          Yes and no. In addition to the expanded register space, I speculate this is partly attributable to the availability of fused multiply-add and vector permute instructions. Similar high-efficiency instructions for these operations (with a then-large 128 bit registry) were implemented in the Motorola / IBM G4 and G5 chips used by Apple, and were one of the primary reasons why those chips performed so well on certain highly vectorized code. If you remember Steve Jobs’ (in)famous Photoshop bakeoffs, this is why.

          I note that the results on the Skylake-X were not tested with code compiled for AVX512, which is implemented on that chip. As Cutress noted in the initial tweet, the Skylake-X scored 30000 (a more than 700% increase) in the same test with AVX enabled, dwarfing the other scores.

          Again, not surprising. That’s the magic of running vector-optimized code on hardware that supports it.

            • tipoo
            • 1 year ago

            So those G4/G5 era features havn’t been implemented in consumer grade x86 until now?

            I keep getting very interested, if cautious, what Apple going their own route on larger laptop/desktop CPUs would look like. Though it would basically spell a no go for any Boot Camp gaming on the side.

            • uwsalt
            • 1 year ago

            Some of those vector features were. They’ve been added, in fits and starts, with SSE, SSE2, and so forth. But those instruction sets were nowhere near as complete or robust as AltiVec or AVX.

            I was always puzzled as to why Intel didn’t implement something like this sooner, especially on the enterprise and high-end workstation side, where applications can make extensive use of these features (and in, the cases of many scientific computing applications, are custom written for the purpose). For example, looking at the November 2003 edition of the Top 500 supercomputer list, you’ll find at #3 a system built by Virginia Tech out of 2,200 2GHz G5 cores (by integrating 1,1100 dual core Apple PowerMac G5s) systems from Apple. Just below is a custom-built Dell system that used 2,500 3.06 GHz P4 Xeons.

            • Srsly_Bro
            • 1 year ago

            I had a 3.06 GHz P4 Northwood in a laptop. It was greatest at killing battery life that may have been 90 minutes.

            • tipoo
            • 1 year ago

            Same, this 7.5 pound absolute unit, and you couldn’t exactly get away with taking it anywhere without the 2 pound power brick:

            [url<]https://www.youtube.com/watch?v=gxhx3O5mH7I[/url<] It's why I guffaw really hard when modern reviews call 4 pound 15" laptops too un-portable!

            • not_a_gerbil
            • 1 year ago

            No he is comparing current 256 AVX vs 512 avx to past 64 bit Intel mmx vs 128 bit Motorola G4.

            A few specialized programs will benefit.

        • Sahrin
        • 1 year ago

        The problem with the Intel AVX512 unit is not that it doesn’t accelerate FP code, it’s that it slows down everything else.

        A benchmark on a system taken out of that system’s context makes about as much sense as comparing a program written in assembly to one that is running on a software emulator. This CPU is not going to be running physics modeling code, at least not while running without something else.

        Also, why the hell are we comparing Cannon Lake to Zen. That’s a prior generation chip, it should be compared to Zen+.

          • chuckula
          • 1 year ago

          For somebody who should be celebrating the fact that Threadripper 2 is launching on Monday you sure seem extremely defensive and butthurt about a two-core product that you’ve been calling a failure for over a year now.

          Incidentally, if it’s “unfair” to compare Cannonlake in 2018 to the same CPU architecture that AMD is launching next week, I’ll be sure to remember that [b<]any and all claims of Zen 2 being superior to Coffee Lake are wrong[/b<] because of course Zen 2 didn't launch until 2019 so it's only "fair" to compare it to Intel's most advanced parts from 2020.

            • Gadoran
            • 1 year ago

            I don t think increases in IPC is a priority at Intel, unfortunately new nodes give very little in clock speed and power reduction.
            Any increase in IPC of 1% means 1%more power, so Zen 2 will be hotter and AMD will be costrainad to run it at a lower clock speed, nullifing advantages.

            Old days of 30% power reduction with node shift are gone……at least on high power cpus.

            Look Intel, it is more profitable do an MCM with two or three high core counts dies on 14nm, to give a winner solution vs AMD Rome.

            Xeons MCM will have great single cores turbos, AMD Rome nope, being TSMC 7nm Early a little crap on absolute clock performance.

            • Anonymous Coward
            • 1 year ago

            I’m don’t agree that “any” increase in IPC will be tied to an increase in power usage, though clearly some are. I suspect Zen2 has a good shot at increased IPC with little or no increased power usage, but also I suspect Intel would have a harder time pulling that off on their implementation, as it is quite mature. I don’t see why a design can’t be “perfected” for a given workload, the question must be of changing workloads and/or increasing specialization.

Pin It on Pinterest

Share This