Samsung's Exynos 1000 might be faster than the Snapdragon 875

19 September 2020
According to alleged Geekbench scores.

Sort by:

  • ?
  • Anonymous
  • IJG
  • 18 Apr 2021

Jewel, 10 Jan 2021The article has shown only performance rating but it has no... morePersonally in my. Eyes Samsung owes me a solar power station simply because of taking 23 .00 more dollars out of my account than they were suppose to

    • J
    • Jewel
    • 6p}
    • 10 Jan 2021

    The article has shown only performance rating but it has no information about heating issue and battery efficiency so it should be mentioned here.

      • D
      • AnonD-754814
      • 6p}
      • 26 Sep 2020

      Anonymous, 26 Sep 2020exynos 8895 not worst for throttling gpu throttles after a... moreEvery GPU throttles(some just throttle too much). The Exynos 8895 review was little weird in AnandTech's website as the GPU was drawing almost same voltage all the time. This might be the reason to throttle that much. I guess that issue was fixed with a firmware update.
      But even after throttling exynos 8895 GPU was on par with Adreno that year. So, throttling didn't make the exynos bad that year as it is this year with Exynos 990.

      Other info are right BTW. The improvement after M2 core wasn't very fluid and not as much as it should have been.
      M3 was a complete failure.
      M4 was good but putting it with A75 cores in Exynos 9820 was a bad idea. AnandTech says the middle A75 core was even more power hungry than custom M4 core.
      Can't comment about M5 core because I think Samsung just disbanded the custom core division too early. That's why they never got any chance to optimize the M5 in Exynos 990 at hardware level. That resulted in historically worst Exynos SoC.

        • ?
        • Anonymous
        • n5a
        • 26 Sep 2020

        AnonD-754814, 22 Sep 2020Smaller node is always efficient unless there is a design f... moreexynos 8895 not worst for throttling gpu throttles after about 10 15 min
        20% improvement
        exynos 9810 cpu throttled more than gpu from max 572mhz to around 266 320mhz cpu throttled significantly battery drain bad
        30% improvement
        exynos 9820 cpu and gpu throttles and runs very bad
        from the max of 702mhz to 166mhz after 5 min gaming or in hot weather cpu seems throttled instantly
        never using 100% utilisation more dvfs issue where not running efficiently battery drain bad
        45% improment
        exynos 990 cpu gpu throttling to about 70% max performance which better than all previous generation
        much better battery drain but still but

        but the upgrades from each year have been so minimal that they 2 generation behind in gpu performance and cpu they used old tech and not optimised correctly

          • D
          • AnonD-731363
          • SH3
          • 23 Sep 2020

          As long as it cant handle overheating it wont be interesting for consumers and manufacturers.

            • D
            • AnonD-754814
            • 6p}
            • 22 Sep 2020

            Mediatek sux, 22 Sep 2020Look at this - 'After factoring in overall PCMark perf... moreYou're right. No different opinion here.
            But I think I've made myself clear where did I get the 15% less efficient chart.

            And most importantly I've got the info about 8895's GPU problem.
            It's in AnadTech's review of S9 exynos SD.
            The problem is indeed with frequency but not the way I was saying, slightly different.
            The problem is,
            No matter what frequency the Mali GPU was running it was drawing almost the same voltage. So, I think it was a software issue, can be hardware too.
            But if it's software then It had been definitely fixed with a later update.

              • D
              • AnonD-754814
              • 6p}
              • 22 Sep 2020

              Mediatek sux, 22 Sep 2020About speedtest g, i have one doubt, why do 1440p and 1080 ... moreYeah.
              I had the same doubt. I think the fault is within the 3D Engine or Android.
              For Example benchmark Score for AntuTutu or other 3D test also doesn't improve as much as it should be going to 1080p from 1440p.
              I think(most possibly I'm wrong)
              For these type of test all phone have to render the 3D scene as it's made but show it at lower resolution. This is why there is small difference but not as much as it should be as they need to do most of the task at native resolution.

                AnonD-754814, 22 Sep 2020I kinda like Gary's Speed Tests, He's making a b... moreAbout speedtest g, i have one doubt, why do 1440p and 1080 phones have so little difference in gpu test like poco x2 pro (27.2) vs s20 ultra (29.6). Other than than that its decent.

                  AnonD-754814, 22 Sep 2020You're counting it wrong. 3hrs vs 4hrs means 25% les... moreLook at this - 'After factoring in overall PCMark performance, the S8 (S835) has the highest average efficiency of any device in this chart—about 24% better than the S8 (E8895). In the Wi-Fi Web browsing test, we saw very little difference between the two SoCs under a relatively light CPU load. The S8 (S835) manages to distinguish itself in PCMark primarily because of its more efficient GPU (Photo/Video Editing tests), which did not factor into the Wi-Fi browsing results.'

                  Also the gpu effiency gap is closed since they both throttle and at lower power exynos efficiency improved a lot. But that should be what a real use case is like, unless you attach those gaming cooling fans.

                  So for normal usage exynos is better, but depends on how much gpu work your normal usage needs.

                    • D
                    • AnonD-754814
                    • 6p}
                    • 22 Sep 2020

                    Mediatek sux, 22 Sep 2020The 8895 gpu outperforms adreno 540 overall, but for whatev... moreI kinda like Gary's Speed Tests,
                    He's making a battery endurance list based on Speed Test G.
                    He's probably making a new battery test app too.

                      • D
                      • AnonD-754814
                      • 6p}
                      • 22 Sep 2020

                      [deleted post]We weren't taking about regular usage. So, that doesn't count.
                      We were talking about benchmark battery lasting tests and benchmark efficiency.
                      I real world Exynos 8895 would be more efficient because of it's efficient CPU.

                        • D
                        • AnonD-754814
                        • 6p}
                        • 22 Sep 2020

                        [deleted post]Smaller node is always efficient unless there is a design fault.
                        What I can't make you people understand is,
                        For regular usage S855 will outlast Exynos 7904.

                        If the peak performance battery drain were the accurate way to measure the efficiency then Exynos 990 would be more efficient than than SD865.
                        Because Exynos throttles under heavy load and doesn't use it's full power.

                        You can call Exynos 7904 low power or low power consumption SoC but It isn't efficient.
                        Efficiency is totally different thing.

                        There is a thing what others don't understand. Samsung just scaled the Exynos 9820 down to Exynos 9825. It's just the node efficiency. No other major improvement was done.
                        But you can see almost 15% efficiency overall, That doesn't mean the SoC is 15% more efficient but it was efficient enough to post a endurance rating which no so called efficient Huawei devices have surpassed.

                          • D
                          • AnonD-754814
                          • 6p}
                          • 22 Sep 2020

                          Mediatek sux, 22 Sep 2020The 8895 gpu outperforms adreno 540 overall, but for whatev... moreYou're counting it wrong.
                          3hrs vs 4hrs means 25% less efficient.

                          I have summarized all those test results.
                          Here,
                          https://www.anandtech.com/show/11540/samsung-galaxy-s8-exynos-versus-snapdragon/5
                          For web browsing exynos has more battery life which indicates it's efficient CPU.

                          Why are you calculating it wrong way ?
                          In PCMARK work 2
                          SD 835 is less than 5% more efficient.
                          GFX Mannhattan ES 3.1 Metal
                          SD 835 is around 23% more efficient.

                          That summarizes to 15% .

                            • V
                            • Viper
                            • U{$
                            • 22 Sep 2020

                            Mediatek sux, 21 Sep 2020Yeah my bad, but i dont think i was rude tho. Anyway the pe... moreI actually said about performance and efficiency because in august i saw a wccftech article where it was written that exynos 1000 will be slower than sd875 but it will be highly efficient so if i go with that that means samsung will underclock the x1 cores to reach the efficiency but in this gsmarena article they are saying that exynoa 1000 might be faster so its really confusing.

                              AnonD-754814, 21 Sep 20205W is true for 8890. And before 7420 the TDP was usually ar... moreThe 8895 gpu outperforms adreno 540 overall, but for whatever reason on gfxbench test adreno is vastly efficient while being just a bit slower, and after some throttling both have same fps. And anandtech does the battery test based on gfxbench test too where exynos lasts 3 hrs and sd 4hrs thats what a 40% difference?, And the 80% is when they are their max perf, so it would be only for a few seconds i guess. Also on pcmark we see 24% more efficiency on sd, looking at both performance and battery life tests. Maybe this test also is more gpu oriented since exynos certainly is better at browsing battery life, or maybe the modem differences matter since theres only 4% difference.

                              Also where did you get the 15% more power hungry on full load? The 80% is at their full load , but its not sustainable, but it still is 40% according to gfxbench battery test. So exynos gpu does clock down. Anyway i dont think these benchmarks are that reliable, especially pcmark since it doesnt even use big cores during its test, and its not even been updated in years. And anandtech suspected qualcomm of low level cheating on 3dmark, now they dont do 3dmark tests. I just hope they do efficiency test with something other than gfxbench.

                                AnonD-754814, 21 Sep 2020Emulators aren't really the best thing for CPU. For ... moreThats true, they arent best example, but they do use all the cpu you throw at them. And there arent other ways to get some specific thing running other than emulators.

                                  • D
                                  • AnonD-754814
                                  • 6p}
                                  • 21 Sep 2020

                                  AnonD-754814, 21 Sep 20205W is true for 8890. And before 7420 the TDP was usually ar... moreI meant to say 5W for exynos 7420,
                                  Mistakenly said 8890.

                                    • D
                                    • AnonD-754814
                                    • 6p}
                                    • 21 Sep 2020

                                    Mediatek sux, 21 Sep 2020Yeah my bad, but i dont think i was rude tho. Anyway the pe... moreEmulators aren't really the best thing for CPU.
                                    For example, Even if I give all 8 core to my android emulator in PC it doesn't work fluently given the power it's drawing from CPU.
                                    Emulators waste more than half of the speed.

                                      • D
                                      • AnonD-754814
                                      • 6p}
                                      • 21 Sep 2020

                                      Mediatek sux, 21 Sep 2020Yeah, gadgetversus, i dont think its a reliable source, but... more5W is true for 8890. And before 7420 the TDP was usually around 3-4W.

                                      I don't understand one thing, AnandTech's review for 8895 shows around 15% average more power hungry in 3D-GPU test(not more than 5-10% CPU is used here). But why the power draw is 80% more ?
                                      I mean in that case the 3D test would have been worse. It's fishy.
                                      I guess AnadTech's unit might have firmware issue. Probably not clocking down the 8895 GPU even for low task.
                                      I mean only this explains the fact that at full load it's only 15% power hungry but average it's 80%. I don't see any other reason. Do you have any explanation ?

                                        Viper, 21 Sep 2020The fact is that samsung itself said that amd rdna will be ... moreYeah my bad, but i dont think i was rude tho. Anyway the performance isnt enough, emulators still use 100% cpu and native games with high refresh rates are gonna be using 100% cpu and gpu. We need both perf and efficiecy, and most of the time one comes with another.