Leakster: the Snapdragon 898 will run its Cortex-X2 prime core at 3.09 GHz

29 July 2021
The successor to the Snapdragon 888 is expected to be fabbed at Samsung's 4 nm foundries and the new process will allow for a clock speed boost.

Sort by:

  • ?
  • Anonymous
  • vIg
  • 30 Jul 2021

6695385, 30 Jul 2021Wonder what exynos amd going to do with this structure over... moreSame set up probably 1+3+2+2(LP)

    • ?
    • Anonymous
    • vIg
    • 30 Jul 2021

    TheLastOracle, 29 Jul 2021The potential of the V9 architecture will take more time to... moreV9 is nothing they just ad SVE2

    Just think of it as 8.3

      • ?
      • Anonymous
      • vIg
      • 30 Jul 2021

      AnonD-1003038, 29 Jul 2021The point of caches is to have data closer to compute cores... moreIf so why don't they have 1gb cacahe and call it a day??

      Why only 8mb cache.

        • 6
        • 6695385
        • n5a
        • 30 Jul 2021

        Wonder what exynos amd going to do with this structure overclock further like exynks 2100 or amd improvements

          • D
          • AnonD-1003038
          • M}3
          • 30 Jul 2021

          Shanti Dope, 30 Jul 2021Tbh, I don't even think there's a need for A15 Bi... moreThe A14 is technically just a cut down version of M1. Thing is, while performance boost is always nice, we're at point where top end chipsets are ridiculously fast. Why not focus on efficiency more? Why not make A15 as fast as A14 is, but lets cut power consumption by 20%. Same can be said for SD 888. The chipset is stupid fast. Why not focus on making it cooler and consume less power for 898 or 888+? I honestly don't think we need more performance just yet, but longer battery life and cooler running, why not. And with mobile devices, that's also a very important factor.

            Davy Jones, 30 Jul 2021Enough of this performance madness Qualcomm need to shift t... moreExactly!
            It's Intel all over again, they've forgotten these are SoCs for mobile devices.

              Anonymous, 29 Jul 2021Nothing too do with Snapdragon, blame the OEMs, Sony Pro is... moreOf course it has everything to do with Qualcomm, they keep promising better efficiency which should result in cooler running. What we actually get are portable heaters that are forced to throttle for survival, only Huawei ever got this right, the Kirins were fast and cool may they rest in peace.

                • W
                • Wayne Morellini
                • 7tm
                • 30 Jul 2021

                Let's take a look at how simple it is to include a Black Magic Design Braw like professional video codec. From what we can tell, this is basically a simple mpeg that's super accurate, with some leeks control profiling, and LUTS and meta data. Basically, similar to a series of stills pictures put in a file. It runs at 27MB/s+ variable bitrate, 4k 12 bit, but a version based on the latest stills standard like, heif might be half that. It can process images in editing quickly on more basic hardware than h264/5/6 or wavelet codecs like Red's. I won't say it's the absolute best, but professionals love it, and can even edit the 12k camera footage.

                It maintains a close relationship to the underlying sensor pixel color value, allowing for simple reconstruction of the Bayer value, bypassing the Red patent, giving a simple recording that can be played back very quickly in regular photo display hardware.

                Now, Ambarella compression circuites, most Likely used on many phones, could be made to simply handle this, as it's basically very close to their capability, you just have to have a big enough version of the circuit on chip.

                The highest end sensors, have 12-14bit modes, can handle 4:2:2 and 4:4:4. 14 bits is preferable to 12 bits when recording pro HDR. Extra bits are easier to compress, as they are no or adjustments in precision to compression data components. So, 12 bits is not 50% more than 8 bits in compression.

                Pro quality video is going look much better on your big big screen/wall TV/projector at 4k+, and full HD worse. More detail, less contour gradient lines from a mis-match in perception to bit depth levels of current systems, and better colour detail (4:2:0 video is a joke in a very large screen, producing rendering errors, with the way photos do it better and more compressible). The Japanese aim for 16 bit tv with rec2020 colour in HDR. Our phones, are a nothing much.

                The final issue is, and external card format. Basically to use USB-C 3.2+ to record to external card or add, or by wifi direct, and carry it in your pocket, or off load and back up files to such a drive when you get home or to the car.

                The think about this sort of pro format. Is you can accurately edit footage frames on a tablet or your phone. Much easier and faster.

                The Braw video codec follows suggestions by a user on how to make a consumer like pro video codec, and how to treat the levels for best pro look, which we see being used more and more in the industry. The same user who published image break down proposals for h264/mpeg2 to retain the best image quality for the data, by profiling it based on what was more noticeable, and tolerable, in human perception. H264 etc, compression improved greatly after that. It also means, that consumer hardware is close to being able to do it at higher resolutions.

                This can be applied to full HD video, 4k, 8k, and whatever 16k+ Samsung is planning. So, user who wants can just use fullhd at better quality. The algorithm can match the group of underlying pixels.

                16k would be less than 13.5/27MB/s+ 4k 12 bits X 16. This is because, increased resolution, but depth and framerate is more compressible due the same quality, but worse comes to worse shooting image quantity matters. It is the only way for them to record their 16k at descent quality without sacrificing significant chip space. Every step between h264/h265/h266/h267, demands increasing multiples of circuits complexity to achieve the same results, plus 16 times the resolution. This means hundreds of times more complexity than h264 with these codecs. Versus, maybe something closer to present day video encoding circuites.

                Plus, you can extract descent stills from it, and do descent action sports video with 180 degree surround lens attachment in 8k-16k plus, to make 2.66-5.33k non VR windowed views of the action to show friends.



                  • W
                  • Wayne Morellini
                  • 7sq
                  • 30 Jul 2021

                  xc, 29 Jul 2021another pocket heater, waiting for next gen node 2nm and armv9 Yep, as other are pointing out. They need different versions. Ones that are ultra performance, ones that are ultra efficiency of you choose to use it that way. Load in a game, it tries to overclock as much as needed.

                  But what they do need, is a mod range Phone chipset with outstanding pro quality video photo and audio artist. If we face reality, a mid range phone has enough performance for most of us and most regular applications band games. High end is unneeded unless for games, or running some science like simulation etc. But no high end phone has really good pro level video, or mostly even video to match the best quality consumer Bluray on a pixel to pixel basis (meaning 144mb/s h264 4k, where pro capture often is much higher). Pro video is often 4:2:2 or raw Bayer 12-16 bit. Let alone 8k, or what ever that 16k Samsung chip is going do. 8k often looks not that great. But, the circuitry to do pro video is a lot less than that to do a high end mobile chipset. So, is a simple addition to an midrange chipset in place of a core or a little part of the gpu. Visual and audio artist, and reporters, and trades with less enthusiasm, will have a field day. There are millions of sales there, on top of regular sales. The hay day of the high end chipset, is drawing to a close, as most people simply don't need that much power on a mobile phone, except for highest quality video, which can be done on a midrange phone, with a codec like back magic designs Braw!

                    C3H844g, 30 Jul 2021At least it's a theoretical performance boost, I estim... moreTbh, I don't even think there's a need for A15 Bionic.
                    Apple can simply put the M1 chip inside the iPhones this year and they wouldn't have to worry about being outperformed by any of their Android competitors at all.
                    The difference is just way too much, that Apple should actually stop producing a new Apple A series chip just to give some headroom for Android chipmakers to catch up.
                    By using the M1, they are saving more money by not having to develop a new SoC that's essentially just the same, if not much better than the M1 and exclusively on iPhones only. Additionally, app developers can easily optimize their apps on this M1 SoC more easily since a wide variety of Apple products is using this.

                    Apple can just relax this year and not bother about making a new SoC, and they would still remain on top. This gives them a lot of room for making their next generation silicons to be even more powerful, so much so that it would take other manufacturers another 2 years to catch up.

                      [deleted post]He is talking about the frequency of the M1's CPU cores, which are not that high, at least when compared to what competing AMD and Intel CPUs are capable of.
                      Still, that SoC is so much more powerful than almost anything else while being relatively low in terms of frequencies (only the Ryzen 9 5900x and more powerful CPUs can beat it).

                      The same applies to benchmarks of the M1, where it doesn't matter even if it's actually having some of the highest benchmark scores among all so many SoCs today, as the performance of this chip in reality is evident that it is indeed a truly powerful SoC that trashes the latest and greatest in both the Windows and Android platforms. It is so good, in fact, that Apple managed to urge AMD and Intel to invest in producing ARM-based CPUs as well like what Qualcomm is doing now. Even Qualcomm is trying to produce an ARM-based laptop CPU to compete with the M1, and that simply goes to show how superior Apple is when it comes to SoCs, regardless of whether we like them or not.

                        • ?
                        • Anonymous
                        • 6QK
                        • 30 Jul 2021

                        Mediatek is better suited for daily drover tasks, and performs much better wjth regards to throttling.
                        Qualcomm is becoming more and more like Intel with each iteration.

                          • ?
                          • Anonymous
                          • U}}
                          • 30 Jul 2021

                          hmm... its samsung node... again i ll skip this one too

                          enaugh with sd 888 fiasco

                            Enough of this performance madness Qualcomm need to shift thier focus to efficiency and thermal throttling so do the smartphone brands

                              [deleted post]Are you dyslexic?

                                Solving the overheating issue with more gigahertz. 69. Nice.

                                  Wgtr, 29 Jul 2021How many cores does Apple have with a frequency of 3 GHz?The 2 "Firestorm" cores in the A14 Bionic reach a maximum clock frequency of 3.1 GHz, also in the high-performance cores of the Apple M1.

                                    AnonD-936656, 30 Jul 2021If they use 8Mb of L3 cache, 4nm Samsung, 3.09 GHz would sc... moreAt least it's a theoretical performance boost, I estimate that the A15 Bionic will score between 1700 to 1800 points in the single-core Geekbench 5 score.

                                      The element needed to cool a Snapdragon 898 as much as possible is very common, liquid helium. ˣᴰ

                                        I think SD 888 is similar to SD 810 where there are major issues like overheating and battery drainage. Personally, Qualcomm should make 2 versions of chipsets for separate audiences(Plus chipsets cater for gaming). For example, SD 898 with slight lower in power but higher in overall efficiency state for flagship phones and SD 898+ with higher in power for gaming phones since gaming phones have cooling system in-built. I could be wrong here but the idea sounds good though. Don't you guys agree?