Qualcomm begins shipping Centriq 2400 family of ARM server SoCs

Qualcomm announced that it's begun commercial shipments of its Centriq 2400 family of server CPUs today. Fabricated on Samsung's 10-nm FinFET process, the Centriq 2400 die crams a whopping 18 billion transistors into an area of 398 mm². Centriq CPUs will have as many as 48 single-thread, 64-bit-only ARM v8-compliant cores (a custom design that Qualcomm calls Falkor). Those cores will have 2.2 GHz base frequencies and peak frequencies of up to 2.6 GHz on the Centriq 2460 product.

Each Falkor core has a 64 KB L1 instruction cache paired with what Qualcomm calls a "24 KB single-cycle L0 cache" designed for low-power operation, for a total of 88 KB of I-cache per core. Those instruction caches sit alongside a 32-KB L1 data cache with a three-cycle load-use latency. Qualcomm pairs two of these cores in 24 "duplexes" across the chip. Each duplex has a shared 512 KB of L2 cache, for a total of over 12 MB of L2 across the chip.

The Centriq 2400 SoC makes available up to 60 MB of shared L3 cache to those cores on a fully-coherent, bi-directional multi-ring interconnect with more than 250 GB/s aggregate bandwidth, and its memory controller offers as many as six channels of DDR4 memory running at speeds up to 2667 MT/s. Maximum memory per SoC tops out at 768 GB.

Three Centriq 2400 SoCs are launching today: the Centriq 2460, the Centriq 2452, and the Centriq 2432. Clock speeds will remain largely the same across these three chips; Qualcomm is instead relying on core count and L3 cache tweaks to set these products apart from one another. The two highest-core-count Centriq CPUs will carry 120W TDPs, while the Centriq 2432 and its 40 cores will have a 110W TDP. Perhaps even more interesting is how Qualcomm is positioning these chips: the Centriq 2460 against the 205W Xeon Platinum 8180, the Centriq 2452 against the 140W Xeon Gold 6152, and the Centriq 2434 against the 85W Xeon Silver 4116.

Qualcomm claims a number of impressive-sounding performance wins for the Centriq 2400 family in its press materials compared to those Xeon processors, but as usual, I'm skeptical of any estimated or otherwise projected performance results in that context, especially because ARM's own Drew Henry touts the ISA's future in the data center as something other than a mere alternative to x86 CPUs. My sense is that Xeons and their high-performance computing DNA are not natural comparisons for ARMv8 cores.

Instead, my gut tells me that putting a ton of ARM v8-compatible cores on a single SoC will be good for both absolute performance and performance-per-watt for the exact type of workloads where Qualcomm expects the Centriq 2400 family to excel: highly-threaded applications, microservices and containers, and any application that needs a "scale-out" platform, i.e. lots and lots of cores per node and per rack. 

Qualcomm is positioning the Centriq 2400 as a key part of the 5G equation, where a titanic number of high-speed wireless devices will need to communicate with large numbers of cloud compute resources positioned near the edge of future networks. Qualcomm claims it already has a wide range of cloud-services providers, software partners, and OEMs under its umbrella as it readies for that future. I'll be keeping a close eye on the role Centriq plays in data centers as it joins a freshly-competitive Intel and AMD in that space.

Comments closed
    • AMDisDEC
    • 2 years ago

    I predict these will be at least as successful as SPARC, or maybe not.

    • End User
    • 2 years ago

    ARM on the desktop can’t be too far behind.

      • AnotherReader
      • 2 years ago

      Using which OS?

        • End User
        • 2 years ago

        Darwin

          • AnotherReader
          • 2 years ago

          Well, Apple certainly has had the performance since at least the A9. That they haven’t done so yet might mean they don’t feel the need to.

    • Zizy
    • 2 years ago

    WTF is with their comparison? 8180 mops the floor with 4116, while their top and bottom chips should have about 20% difference. Also, competing with 8180? Seriously? I could see performance somewhere between 6152 and 4116, but where did they even imagine to fight top of the Intel chips with their first product at half the TDP?

      • AnotherReader
      • 2 years ago

      They are choosing the cheaper Xeons. It would have been interesting to see a comparison against an Epyc 7401 as that is closer in price.

    • ronch
    • 2 years ago

    If K12 was released, I wonder which would’ve ended as a better product.

      • AnotherReader
      • 2 years ago

      Given Ryzen’s competitive single thread performance, I bet that K12 would have annihilated Falkor. However, AMD was right to focus on Ryzen. For a corporation in as parlous a condition, taking a small part of an existing large market is better than being a trailblazer in a new, uncertain market.

        • ronch
        • 2 years ago

        Completely agree. And that’s exactly why Ryzen was given much fanfare while K12 was canned.

    • tootercomputer
    • 2 years ago

    18 billion transistors. Geez. I once built a “visible” transistor radio for my 8th grade science fair. It was a kit your could purchase, about 18″ by 18″ with a plastic cover over it, it had to be assembled, and we used it to illustrate the thrust of our project, which was the magic of transistors. This was in the winter of 1964 a dreary time, February in Chicago. JFK had been shot and killed just a few months earlier, the Beatles had just appeared on Ed Sullivan for the first time and helped a lot of us move past JFK’s death (I still feel tremendous gratitude to the Beatles; their presence was just wonderful and made us smile), and our science fair project with my visible radio that contained 2 (yes, 2) transistors won first place. Each transistor was the size of a pea with two wires coming out the bottom. I’m trying to remember, transistor radios back then typically had maybe 5 – 15 transistors, something like that. They were so tiny compared to tube.

    18 billion on a chip. Geez. Ain’t technology great.

      • ronch
      • 2 years ago

      Yeah. I very much appreciate what we have now when I recall what I had before. My first PC had an NEC V20 and 640KB of RAM. Now it amazes me how my PC today has a 1.3 billion-transistor FX-8350 CPU with 8 cores running at 4.0GHz supporting all sorts of advanced instruction sets and power management technologies while STILL being fully compatible with that NEC V20. The FX is/was probably much cheaper than the V20 when it came out too, accounting for inflation.

        • cygnus1
        • 2 years ago

        Think about this. Three years ago the computing power/resources on wrist watches dwarfed that first PC of yours. Wrist watches!

        And even though they’re not cheap watches, they’re probably cheaper than what that first PC of yours cost too.

        • tootercomputer
        • 2 years ago

        640k of ram. I recall those times, there were advertisements in magazines like Time and Newsweek touting computers with 640K of ram.

    • ronch
    • 2 years ago

    Well, I hope this thing doesn’t end up like Artax.

    • tsk
    • 2 years ago

    Are there any reviews for these chips yet?

    • derFunkenstein
    • 2 years ago

    Centriq sounds like an old Compaq desktop model.

      • Alexko
      • 2 years ago

      Or a Compaq laptop built upon a Centrino platform.

      • ronch
      • 2 years ago

      Or Centrino.

Pin It on Pinterest

Share This