Such Power, Much Heat. Xiaomi Mi 11 Snapdragon 888 Performance/Efficiency Analysis vs 865 Kirin 9000 By Golden Reviewer

By Golden Reviewer
Aug 14, 2021
0 Comments
Such Power, Much Heat. Xiaomi Mi 11 Snapdragon 888 Performance/Efficiency Analysis vs 865 Kirin 9000

Hi guys welcome back to golden reviewer. We all know the hot news recently is the Xiaomi mi 11 and the snapdragon 888. After all, it will be the most popular android flagship soc next year I mean this year and of course I tried my best to get this device as fast as possible, although I had to source it from China and pay a hefty shipping and improve fee, but I just wanted to test this as soon as possible and give you the results, because I know all of you are just waiting. So if you have seen my unboxing video and some previous testing videos, you've known this is the Xiaomi mi 11, with snapdragon 888 and of course I have the Xiaomi mi 20 pro with snapdragon 865 here and also the Huawei Mate 40 pro with the newest Karin 9000. But from my testing I was actually very disappointed with how the snapdragon 888 performs. Why am I disappointed? Why is snapdragon 888 a bad soc or a failed soc? Why would I say that Qualcomm never deliver what they promised? I think it's quite hard to answer all these questions.

From my normal gaming fps test, because I think we need a more rigorous, more scientific and more systematic test to show you what I mean, so this test will not be about fps about battery life. How many hours of battery life whether the game is smooth or not? Can you run PUBG at HDR settings? Does it run gangster impact smoothly? No, that are just hard waving and people will question they'll, say: oh it's about the display resolution. Uh! Oh, it's about optimization! So I want to isolate all these and let the numbers talk. You might be confused about what I mean but never mind. Let me show you so without further ado.

Let's start our test, so the first test we are going to run is geek bench 5 and no it's not like your geek bench 5 benchmarks. You see all day this time. We are also going to record the power consumption of the device while running geek bench 5. Furthermore, I want to make the results more accurate and representative of the soc itself instead of the device, so I will actually leave the device idle for one minute and record the idle power consumption. First, then, when we run the benchmark and record the active power consumption, we can actually subtract the idle power consumption to know roughly what is the real power consumption of the soc itself and as we talk, we can see that this Xiaomi mi 11 has been idle for about one minute, and we know that the idle power is around 1.3 watt. Now, let's start to run the benchmark.

Now the benchmark has finished and let's take a look at the power record, we can see that there are about two parts to this power record. The first half has a bit lower power consumption and the second half has a bit higher power consumption. This makes sense is because geek bench 5 actually run the single car benchmark first and then multi-core benchmark later now, let's look at the data from the single core benchmark and multiple benchmarks separately. So let's look at the single core part. First, we see that the average power is around 3.1 watt and subtract the idle power we just got. We know that the average single car power consumption is about 1.81, but I need to give a disclaimer here, because this number is not 100 accurate. That's because geek bench 5 actually performs the test at intervals, so the core is not working 100 of the time.

Nevertheless, I think the average power consumption is still meaningful when we compare it across different socs from here. We also see that the single car power consumption as its peak is about 5.9 watts, so subtract, the idle we got about 4.6 watts. Next, we look at the multi-core part. In the same way, we see that the average power consumption is about 6.4 watts, subtract idle. We got about 5.1 watt and then, similarly, the peak is around 8.6 watt. Okay, that's it for the geek bench 5 for snapdragon 888, and you should have already understood how I am going to do all these tests next I'll follow the same way to test out the snapdragon 865 and the Karin 9000, and I won't explain these methods all over again.

Instead, I'll just show the footage and show the numbers on the screen right, but of course you don't have to worry about remembering all these numbers at the end, I'll, actually consolidate all the data into an Excel form and make some in-depth analysis with you right. So now I've done the geek bench test on all three devices. I have consolidated all the data I've gathered here, and I've also done some simple calculations to actually derive the score per watt number, which is also known as the energy efficiency, and I also added color code to let you know how well or how badly the soc performs. So green color means is the best of all three yellow means is the second and the red means it's the worst, and now it's very easy for you to tell what's the problem with the snapdragon 888, so it has two green boxes which are the performance. The single core and multiple performance are both the best of the class right, but they also have has four red boxes, which is power, consumption and score per watt or energy efficiency.

That means the snapdragon 888 has the best performance, but the worst power, efficiency and worst power consumption, and for this CPU part, I think the carrying 9000 actually is the best balance here, because if you, if you see the snapdragon 865, apparently the performance is too far behind the newer socs, so the Karen 9000 actually achieves a performance which is very close to that of snapdragon 888, while it still have very good power efficiency, but we also have to know that the snapdragon 865 is almost one year old right. It's the previous generation flagship soc, so I think it's still holding up pretty well here, so the only loser is the snapdragon 888. Okay. Now, let's use the Manhattan 3.1 test in GFX bench to test out the GPU performance and power efficiency and power consumption using very similar way. I think this also does not need much explanation.

I'll just play the footage and display the results. Okay, now we finish the test, and we add the data to our Excel form. We have three more rows now. If you look at the pure performance, the snapdragon 888 and carrying 9000 are both very good. Apparently they beat the snapdragon 865 by a large margin.

However, did you notice that the snapdragon 865 is so, so power efficient? It uses only half the power of the Karin 9000 and even less than half of the power of the snapdragon 888, so its power efficiency is much, much better than that of the snapdragon 888 and Karen 900. So I want you to think about this. Is it really that meaningful to have this high performance, while your power figures shoot through the roof? Okay? Now let me show you more other tests. Let's do an tutu first, I think we are all quite familiar with an tutu, so this benchmark actually covers CPU, GPU ram and disk. Almost everything you know a smartphone.

Now they have done the an tutu benchmark and, let's just take a look at the result, I think the score uh we are all familiar with. We see the leaks, we see all the reviews, so that's not interesting to us. What is interesting to us is the battery and temperature figures. So do you notice that the temperature rise for the snapdragon 888 is actually 12 degrees, which is much higher than the other two all three devices start with a battery temperature of around 31 or 32 degrees, and the snapdragon 88 actually reaches 44 by the end of the test, also notice that the snapdragon 888 used 5 of battery, while the other two use less than it. So you see this corresponds to our previous test.

The snapdragon 88, although it has the highest performance, is not power efficient at all, so the devices will consume more power and be hotter. Okay, next up is 3dmark stress test. This is more of a GPU focused test, rather than a comprehensive one like an tutu. So if you play a lot of GPU hungry games, this will be more interesting to you. What you're kidding me, the me11 is actually giving me a warning saying that your device has overheated, and I don't know it just kicked me out of the benchmark: okay, okay, so the me11 with snapdragon 888, couldn't even finish.

The 3d mark stress test after around 18, runs out of 20. It overheated and kicked me out of the benchmark notice. The battery temperature is 52 degrees. This is crazy. I'm really shocked by this, because I've never seen anything like this before.

Okay, anyway, let's just let the other two devices finish the test, and I'll get another test for the Xiaomi device to compare the results. Okay, let's take a look at the scores first, so if you look at the best loop score, apparently the carrying and the 888 will be better than the a65, but both of them will throttle, especially for the carrying 9 000. You see it reduced to only 60 percent of its peak performance and after throttle, the performance is actually more or less the same as the snapdragon 865. The snapdragon 888 does not shuttle as much is only 88, but that's at the expense of extremely high temperatures, which I'm not I'm really unsure about. But if you look at the snapdragon 865, it does not throttle at all right from the beginning.

Until the end, the performance is 100 the same, and if we look at the more interesting temperature and power data, we see that the snapdragon 865 actually consumes only seven percent of battery, and it has a temperature rise of 12 degrees. While for the snapdragon 888 it uses 15 of battery, that's more than double that of the snapdragon 865 all right, I know the battery capacities are different. All right I'll, do a calculation later and put the exact energy numbers in midi mp hours here, so that you can compare them. Okay and the temperature rise for snapdragon 888 is also a bit ridiculous. It's 19 degrees right.

If you see just now, the test ended when the temperature reaches 52 degrees. So for this another test, the temperature reached 51 degrees is about only one degree to the threshold where the OS will actually kick me out of the benchmark. So maybe for this particular time I was lucky. I never hit 52 and then, if we look at the Karen 9000 is somewhere in between uh, it's not as hot or power hungry as the snapdragon 888, but it's still not not as efficient as the 865. Now we've finished all the tests.

I want to do, and we have all the data we need, and we can complete our table and I think most of you can already see exactly what the issue is with snapdragon 888. The performance is not the issue you see. The four green boxes are all for performance, so it is the performance, is amazing, no question there. It's the strongest soc on any android phone period, but we'll also notice that it also has the most red boxes almost for all energy numbers and efficiency numbers. The snapdragon 88 is the worst of all.

This clearly shows that the snapdragon 888 is a terribly inefficient chip. Well, if we look at the snapdragon 865, it's not amazing in terms of performance, but it's so good for power efficiency, and it beats the snapdragon 888 and Karin 9000 in almost all power efficiency figures and now Qualcomm. You are telling us that the snapdragon 888 actually improved the power efficiency by 25 over the 865, and that is the funniest joke. I've heard this year, maybe just keep that joke to yourself, Qualcomm and then the Karen knight Norton is somewhere in between the snapdragon 888 at 865. Its performance is better than the 865, but if it is also more efficient than the 888.

So I think maybe Karen also strikes. Some violence in between is also a pretty good soc I'll say and of course I also run the spec int for snapdragon 888, but this is more on the hardcore side. I've also consolidated the data in this table. Some of this data is from another friend, so sometimes we do this together, because uh one person cannot have all the devices in the world right, so we share data with each other, and we do the test uh together. This spec in test is mainly about CPU performance and efficiency, and if you look at it, the snapdragon 888, of course, has the best performance, but the power efficiency is not good.

If you compare it to Karen 9000 and snapdragon 865 is clearly lacking in terms of efficiency, to perform performance per watt column- and it is even not as good as the snapdragon 865 plus, which I already say, is a meaningless product. It's a gimmick. It's only for numbers right, it's not really useful. All right, that's all the tests and data. I want to show you today.

In conclusion, I think snapdragon 888 is indeed a failure. We see some healthy improvement in terms of performance, but at the expense of awful power efficiency, which is never a good thing for mobile devices, and it's the false claims from Qualcomm. That frustrated me. The most I mean it's okay, if your product is not good, but why would you make false claims and make us believe that it's better right, it's straight out a lie to say that the 888 is more power efficient than a65, because it's simply not the numbers do not lie so for how bad the snapdragon 888 is. I wouldn't recommend any of you to buy any device with snapdragon 88 that includes the mi 11 and the upcoming s21 series.

Some people ask me whether this will be fixed by optimization I'm. I won't hold my breath because all the tests we do today are pretty low level stuff. You can optimize a game, you can optimize certain benchmarks, but you are usually not able to optimize the uh pure power efficiency. Look at the Enos 990. It has been one year.

I don't see any concrete improvement uh, but we are still interested to see uh when more manufacturers uh with small devices coming out. How will snapdragon 88 evolve in the future right, so uh stay tuned to my channel and subscribe if you haven't uh because I'll always bring the newest stuff to you? Okay, thank you for watching and see you next time.


Source : Golden Reviewer

Phones In This Article


Related Articles

Comments are disabled

Our Newsletter

Phasellus eleifend sapien felis, at sollicitudin arcu semper mattis. Mauris quis mi quis ipsum tristique lobortis. Nulla vitae est blandit rutrum.
Menu