ARM’s next-gen chip designs are on the way, and if the benchmark results listed in a series of leaked slides are accurate, the company’s new Mali-G72 graphics design could offer up to a 40 percent improvement in graphics performance while the new Cortex-A55 and Cortex-A75 CPU designs could should provide at least a 20 percent performance boost over their predecessors.

VideoCardz has published a set of leaked slides providing some details about the new designs. It looks like ARM plans to make an official announcement on Monday, May 29th ahead of next week’s Computex 2017 trade show.

ARM doesn’t actually manufacture any processors, but the company’s designs are licensed and used by the companies that make most of the chips used in smartphones (and some tablets, smart TVs, and other devices). That means these new designs will likely influence upcoming chips from Qualcomm, Samsung, and MediaTek, among others.

Thew new Mali-G72 graphics tech is an update to last year’s Mali-G71, and it includes improvements that ARM says should make it a better option for virtual reality and machine learning applications. The company says it offers up to a 40 percent overall performance boost, 25 percent higher energy efficiency, and 20 percent better performance density.

ARM’s Cortex-A75 CPU design is an upgrade to the Cortex-A73, and the company says it offers at least 20 percent “more mobile performance” in most tasks, with some benchmarks showing a nearly 50 percent improvement (although, to be fair, the biggest discrepancy is in the Octane 2.0 benchmark which was recently discontinued).

The new Cortex-A55 design is an update to ARM’s Cortex-A53, and it can either be used in mid-range processors or paired with a Cortex-A75 for multi-core processors that use ARM’s big.LITTLE tech to select the appropriate set of CPU cores for each task.

 

ARM says Cortex-A55 brings up to 15 percent better power efficiency and up to 2 times the performance of its predecessor, although you’re unlikely to see that much difference in most tasks: ARM provides results from 5 different benchmarks, and for the most part it looks like a 20 percent performance improvement is more typical.

You can find more slides at VideoCardz.

Support Liliputing

Liliputing's primary sources of revenue are advertising and affiliate links (if you click the "Shop" button at the top of the page and buy something on Amazon, for example, we'll get a small commission).

But there are several ways you can support the site directly even if you're using an ad blocker* and hate online shopping.

Contribute to our Patreon campaign

or...

Contribute via PayPal

* If you are using an ad blocker like uBlock Origin and seeing a pop-up message at the bottom of the screen, we have a guide that may help you disable it.

Subscribe to Liliputing via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 9,547 other subscribers

8 replies on “ARM Cortex-A75, Cortex-A55, and Mali-G72 details leaked”

  1. ARM has three major design divisions: Cambridge (main), Austin, Sophia.
    And apparently these references were designed from ARM’s Cambridge division.
    So in terms of succession:
    Cortex A7 -> Cortex A53 -> Cortex A55

    However, the Cortex A75 reference IS the Cortex A72 (Austin) but hybridised with the A73 and some other improvements:
    Cortex A15 -> Cortex A72 -> Cortex A75*

    As opposed to ARM’s Sophia (French) division:
    ARM11 -> Cortex A5 -> Cortex A35
    Cortex A9 -> Cortex A17 -> Cortex A73

    ARM’s Mali-GPU divisions makes designs in their Norwegian (Trondheim) centre, which is why they always have a Viking-themed internal codenames for the products.

  2. How’s the performance of ARM chips now? Ignoring the marketing, what’s the actual performance improvements over the generations?

    I’m actually mostly interested in performance under Linux distros (Debian, Fedora, Ubuntu, etc.).

    1. If chromebooks are anything to go by, really passable. They’re still no speed demons but can at least beat Intel’s low end range, atoms, celerons based on atoms and whatnot. Core Ms win in a fight though.

      1. Core M’s lose when the ARM chips can handle more cores/threads, and the Energy (Watts) is restricted/throttled. In the near future (2018-2019), we should see the best ARM chips which will surpass Core M’s at their peak (i7-7Y75).

        1. So Intel will just sit down and do nothing?

          Intel will have 10nm this year and Core-M will be improved (i7 8Y00).
          It’s hard to say whether ARM will outperform Core-M.

          1. Yes, Intel will sit down and do nothing.
            You do realise that they only recently introduced Kaby Lake, right?

            Intel said they are skipping 10nm and going straight to 7nm.
            They already have 7nm now, but not for consumers.
            It will take time until they can make it mass produceable and cheap-enough.

            That’s why Intel is in a bind!
            AMD’s RyZen microarchitecture is finally competitive, and AMD is free to use 3rd-party wafer fabricators. So its possible the 2nd-gen Ryzen chips might use something like Samsung’s 10nm Finfet lithography, and actually equal or surpass Intel in terms of efficiency and/or performance, the way they did in the days of the Pentium 4/Athlon 64bit, and the Pentium D/Athlon X2 days.

            So until 2020, there won’t be any (major) performance or efficiency improvements for Intel’s Core M7/Core i7-Y processors. That’s what Intel said. That means that ARM SoC’s built on newer microarchitecture, and on 10nm lithography will actually surpass Intel’s i7-7Y75 performance in 2018. Maybe in 2019.

            That’s technology for you, progress and competition is a good thing.
            Being a ARM/Intel/AMD fanboy is not.

          2. ….all this said is true.
            Unless Intel feels the pressure, and decides to cancel their current roadmap and plans.

            Things Intel might do is to firstly reduce the RRP price of their stock as an instant form of damage control. They could also, within weeks, stop using cheap thermal-goop inside the silicon, and opt for more efficient copper-metal plates. Then developing into several months, they may make the leap from Dual-Cores to Quad-Cores in their mobile chips, and also increase the cores in their desktop chips… which would boost the total performance but come at the expense of single-core performance. Also Intel could adopt using 10nm wafers from TSMC, Samsung, or their own-foundries, to boost efficiency over their 14nm products. And finally they could shift the launch of Coffee Lake from mid-2018 to late-2017 which would mean they lied to their developers and investors.

            I mean, Intel has the upper hand. They’ve had it for a while.
            In 2017 they’ve been caught out, with their pants down. Yet, they could always do something as I’ve listed… even taking drastic measures like pushing for early launch to keep their reputation strong and big investors happy, even if it means proving they’ve been stifling the competition and degrading the innovation rate.

  3. Those bar graphs showing a75 performance relative to a73 are not very proportional. Good to see ARM marketing approaching Intel levels.

Comments are closed.