There is a difference between CPU temperature vs device temperature. CPU temperature is normally measured on the SoC or the die itself. Device temperature is how it feels on your hand. 52°C for CPU temp is perfectly normal because the phone is on passive cooling (no fan to dissipate heat quicker). So the app temperature reporting is not false positive. The reason it stays at 50°C even after put in the fridge for 10 mins is because what you just did is only cool the surface of the phone, not inside the CPU die. You need to put way longer time if you wanna cool down that CPU die, 10 mins definitely is not enough. Even if you put it for 2 hours, I can bet the CPU temp will still hover around 40°C since the CPU still generate some heat even when idle (due to heat cannot escape fast enough without a active cooling fan).

Also this, as an update to the video above, I guess Samsung made an official response?

My two cents on this. It depends on how you evaluate the benchmark scores. To me benchmarking apps is just another supporting evidence to indicate the overall performance of a device. It does not paint the complete picture as the benchmarking app itself could be flawed or skewed whether intentionally (to get great score to show off) or unintentionally (due to new SoC hence less optimization or just pure buggy code that leads in inaccurate score).

Using real apps to test device performance is still considered as benchmark if you are using it to test and compare different devices using the same app while trying to reproduce the result you want using the same task or condition set by either yourself or other people.

It is best you use other people's review as just a standard reference point and you do the benchmark again yourself to see if it does reach to your expectation. But then again, benchmark results can vary due to various condition or circumstances beyond your control. Even yourself testing it could be flawed as you might not have realised some mistakes you do before you start the process.

Conditions such as:
1. The room temperature you used to test which can affect your device initial temperature (whether you test in an aircond vs normal temp room)
2. Apps installed/running in background (The best way to use a benchmarking app is with its stock configuration aka only benchmark it when the phone is just newly setup finished after factory reset)
Even then we can argued that, but hey Samsung came preloaded with lots of bloatwares which can skewed benchmark score in some ways. If that's the case might as well strip and disable all unnecessary services and apps but that requires the devices to be rooted if you wanna go for the extreme to get the best out of your devices

Circumstances such as:
1. The SoC itself (whether you get the golden sample or not aka silicone lottery) eg. like how some Intel i9-10900K series (unlocked) can overclock beyond 5.3GHz stable with lower voltages required.
2. SoC manufacturing flaw (again based on 1st point but you get the worse silicone which performs below average standard)

Antutu will give greater scores if the device have larger RAM and storage size even if the speed/bandwidth is not increase, which means that capacity alone contribute to more score, did you know that?
To me, 3DMark is quite an accurate bencmark to be used to evaluate GPU performance only since it does show real time FPS while it performs benchmarking.
Geekbench, I can't comment this much as I did not really use it plus some people say it is not accurate or skewed in some ways.

For me personally, I will benchmark using both synthetic and real world apps to gauge the device performance.

Ok, back to the topic that I wish to discuss. Mr.whosetheboss phone could be getting the flawed SoC phone which result in below average performance, we can't say for sure until we have further validation. Plus I don't agree with the way he tested sustained performance, no real world apps will stress both CPU and GPU to the max at all times, I mean literally NEVER so do not take this as a good indicator of real world performance. Those test are on the extreme side which means the worst possible scenario. It is good when you use it to test stability NOT performance. What I'm trying to emphasize here is his testing methodology is not wrong by any means, it is just that do not put this type of test as a very important metric when compared to other types of testing such as real world sustained performance which is by far more important aspect.

For Booredatwork.com, his testing methodology seems fine for a normal average consumer that is, personally I will not use his because his testing lack of details I want to know such as frame pacing (does it micro stutter?), power consumption used (an important factor to estimate battery life while gaming) and etc. There is only one thing I wanna highlight here, Genshin Impact at Max settings can maintain 60 FPS with 99% stability with SD 888? I call that BS. Not blaming the reviewer here as he himself might not have realise this but I think the app he use to test Genshin Impact is bugged for unknown reason.

"But in real life performance, both of them are almost the same." I have to disagree on this. It looks almost the same to you because you have not use any real world heavy apps that can enable you to see the real life performance differences. I do not have SD 888 devices with me so I can't test it but I can say one thing for sure. In gaming alone, games is always more optimized in SD counterpart compared to Exynos. Also, Snapdragon 888 introduce a new feature call as Variable Rate Shading (VRS), that can impact the game FPS quite significantly along with extra battery life due to the said optimization. Unfortunately for Mali GPU, VRS is still not present yet.

To be honest Mali G78 MP14 GPU still does not meet my performance expectation. Let's take Asphalt 9 as an example. A 2.5 years old game released around July 2018 still cannot maintained lock 60 FPS at High settings on my S21 Ultra. It is much better than my predecessor phone though S10 and S20+ (which can drop to below 40 FPS a lot of times). This is still not so bad. The game I also play which is Pokemon Masters EX. This game's performance experience is so atrocious, can say it is much more worse than Genshin Impact. In battles average around 22 FPS (S10 average around 17 FPS). The UI with a lot of characters can drop to as low as 8 FPS (mind you this is the lowest graphic setting you can go), my God can you imagine that even on S21 Ultra (S10 was 5 FPS ...). I saw so many YouTube game play and their phone seems to be smooth at around 30 FPS, I have no idea if they used the SD counterpart phone, but I believe so.

Sorry for the super long text. I might change or add new info if I sees fit.

I have never said that benchmarking app represent actual usage. Benchmarking app is created to mimic real world usage to test performance but you should never rely too much or put a lot of importance on it. It is just another tool to measure a standard reference point of view of how it compared to another device but never to judge the experience you use on a single device. You already understand that so I don't need to explain further.

I would not say little improvement, it is a massive improvement compared to Exynos 990 in terms of CPU performance and energy efficiency. The only inferior part is the GPU which still uses Mali GPU. You need to know that Exynos is an SoC not just one system component. SoC includes the CPU, GPU, AI, DSP, ISP and also the modem(4G, 5G signal receiver and etc). The only part it cannot keep up to Snapdragon is the GPU part where SD uses Adreno GPU. You need to get this fact right 1st. Wait for next flagship that comes with AMD GPU, then we can start bashing whether Exynos is truly trash or wonderful SoC.

Your statement can be applied on Exynos 990 as it was truly an awful chip. It lose in all aspect to Snapdragon whether in CPU/GPU performance, battery life and picture quality computation by ISP was a massive letdown.
If you still insist this year Exynos has little improvement, then you have a problem with yourself, being blinded by the facts with lots of hatred.

Arun testing is not wrong, just another method to test sustained performance. It just test the worst case scenario as it test on the extreme side of things.
Yeap, I agree nobody use it like that in real life. It just showcase the capability of the hardware and how it handles if push to the max.

It is like comparing a drag race between Perodua Myvi vs Proton Iriz on which car can cross the finish line 1st, which is also consider a benchmark.
But why do it you may ask, they are not racing cars, so whats the point. We don't race on public roads under normal circumstances so this so call benchmark is useless or stupid.
They do it because they can while also doing it for fun or we can call it in the name of science.

Is it an important test? Maybe not. But can you get any meaningful information out of it? Absolutely.

By knowing which car can cross the finish line 1st, we can get to know how fast the car shifts its gear and also how fast it can accelerate. This metric can be used to determine the time it is needed to accelerate to a certain up to speed which can be useful if you wanna overtake a slower car, whether you car can cut it fast enough or not when switching lanes or even when making a turn on fast moving public road within the split second (whether you successfully turn and accelerate on time without crashing on another car or if your car too slow too react and it cause a collision with another car).

What I'm trying to convery is we should not take the result of the drag race as an important metric to measure speed, but we can look it at another perspective as in we can learn how the car behaves during the drag race test and use those information and apply it in real world application.

I'm not sure if my car analogy make sense here but I hope it does and you get to see the bigger picture out of this using a so call benchmarking app.

Well, I did not say it transform into real world performance. I just said that you can use the results as an additional knowledge for you to be applied in real world scenarios. Benchmark result can transform into a real world performance if the benchmark itself is flawless and accurate to real world usage, the possibility is there just not there yet on smartphone space. Imagine they put clips of heavy team fights occur in Mobile Legend and it is used as the GPU benchmark, how ironic would that be, definitely can transform into real world performance, sort of. Or maybe Genshin Impact instead, since this game is the talk of the town when it comes to performance testing

It just means that those so call race simulators is far from perfect and requires more tweaking lol.

One simulator that can be fully used and applied on real world would be the Flight Simulator (not the one came with computer program only but I'm talking the one built with proper hardware that can simulate physics and you feel just like the real thing, you know how prepilot take SIM test before becoming a real pilot). Of course I have to put disclaimer here, no simulator can exactly have 100% accuracy to mimic real world scenarios no matter how good it is, you can get close enough as simulator is build when someone experience that in real life, so if one never experience such mishap before, the simulator cannot reproduce such incident as it was not written to produce such result since it did not happen before. The best part is it will improve over time as people discover/learn new information about it.

If that is the case only God can create the perfect simulator

Also have you seen this video? The future of Racing Simulators, it is not ready yet but maybe by this year it should.

The current simulator


Damn anybody here can guide me how to properly embed YouTube videos in LYN forum?

Which is why on my previous long post I said that take reviewers review as just a guideline or a standard reference point. You should do your own homework and test it again yourself and see if it matches the reviewers claim. Because there are too many configuration that affects battery life, no single reviewer can test all those aspect.

So you did make a right choice, to digest reviewers info and still make your own decision whether it is worth it or not. If only Samsung allows you to use your phone as trial for at least one week, then at least you can compare yourself between different models before you made the final purchase decision.

Samsung offers free 3-day trial of Galaxy S21
You will also be rewarded with free gifts after the trial period is over. Too bad only in South Korea, I wish Malaysia have this too.

I see, just needed the video ID code huh. I thought need a complete full link...
Still, thank you very much for letting me know

I dunno mah, silly me, still new to this forum. Well at least I learn something new everyday.

Actually the order does not matter at all, whether to remove fingerprint 1st before installing new screen protector or if you choose to remove fingerprint after installing the new screen protector lol, your choice which one you prefer to do 1st. As long you use brand new fingerprint data after installing the tempered glass is the most important matter here.

This post has been edited by Josh95: Feb 8 2021, 07:43 PM

The unique part for this reviewer is having two same SoC phone, both are SD888 and yet it shows some slight performance difference even under same condition. All this comes down to silicone/binning quality which I just mentioned in my previous long post. All reviewers result is just inconsistent, which is why I have said before to use it as a standard guideline and do your own homework to test it yourself once you owned those devices.

What I can say is Snapdragon result seems to be more consistent compared to the Exynos counterpart which was all over the place among all the reviewers I have watched.

This post has been edited by Josh95: Feb 9 2021, 12:17 AM

Lol, come on don't get confused. Antutu reports CPU temperature, of course it will be much higher, whether the reporting is accurate for the latest Exynos 2100 (I can't confirm this). Temperature tested from outside is device temperature. You cannot get CPU temperature reading just by using that temperature measuring device ...., it can only be measured by software or if you disassembled the phone and measure the temp on the SoC itself.

That is possible actually, the heat output was too small hence the heat did not get transfer much to the body of the phone just yet. If you repeat the test and the heat pipe of the smartphone gets overloaded, then the heat will be forced to escape to the body of your phone. It just means that the heat pipe is doing its excellent job here, it only fails or cannot do properly when heat produce more than the heat pipe can handle, this is where the phone's body will start to heat up. To offset this heat, people install fan so that heat will not trap in the smartphone for too long and accumulate in a heat pipe.

An example I can give you. Imagine you cook noodle soup with a pot. The pot definitely will be too hot for you to handle and you cannot touch it else you risk burning your hand. But once you submerged the pot into water, the pot will be cool down instantly and yet the water do not feel too hot to touch, just warm at most. I'm not sure how hot is the metal but it could be more than 100°C but when the heat transfer to water it is just maybe 40°C.
This is because when heat accumulate in a small area, it has high temperature, but when you disperse into larger area, the temperature will be lower. This is what we call heat capacity. So the heat pipe in smartphone also has its heat capacity tolerance level before it gets overloaded.

I think he misunderstand your context or did not quite get your question properly (I bet it even until now he could be still clueless ). I have read the reply hence my answer.
Which is why you need to explain more thoroughly next time if you don't quite get the answer you wanted the 1st time.

I saw you did explain again but he still did not get the context. Could be language barrier or he just misunderstand the context, quite a common issue on today's society hahaha

This post has been edited by Josh95: Feb 9 2021, 11:33 AM

Did you increase the touch sensitivity? It is a must for Tempered Glass installation.