Samsung 9820 subjective cpu perf/efficiency impression vs prior gen

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
Have had s10+ for a few days as a replacement for the s9+ - both exynos versions
I have noticed the preliminary test on AT from Andrei and also a few other online and even if my usecase is very personal and its subjective results i think they are important to add to the picture.

During installation procedure and updating the phone it got warm and slowed down and in that sense it performs as you can tell from many test, but for normal light workload i think its an entirely different picture that emerges.

My usecase is comprising of www surfing for typically several hrs, a few light apps like trello/upwork/reddit and stuff like that and sometimes watching a video for an hour or two. No gaming og heavy video rendering what not.

Besides from the surfing heavy pages, the load is light but i use it way to many hrs, so the s9+ seldom lasted all the day and almost each day had to recharge. And i even ran it in power saving mode because it just blow the battery to fast in normal mode.
For surfing the s9+ often got warm and even light apps seemed to hit battery life.
Even without power saving mode the S10+ last an entire day without the slightest problems. Its not comparable.
I dont know power usage for my load, and the s10+ battery is bigger, but my guess is the cpu draw is next to half if not far less. In practical use for my usecase its a radical change.
For speed the phone is much faster. Its faster all over and when you give it a heavy www page to load it just hammers it nearly instantly.

I dont know how others usecase mirrors mine, but its interesting for me to know how often those M3 cores must have booted up on the S9. It seems like the M4 are standing idle much of the time and even the a55 is taking a bigger toll and leaving most of the rest to the a75. Then the M4 can stay idle and be cool and ready when they are asked to deliver.

For my usecase i am convinced. Its the biggest yearly generational cpu upgrade in a phone i have experienced. Granted the 9810 was lousy but still the 9810 just delivers in a way the bm numbers dont seem to indicate. I am sure if you use it for video editing and some heavy loads you will get the same old battery buring M core results, but how much real world workloads for most users is like that?
 

dark zero

Platinum Member
Jun 2, 2015
2,655
138
106
The problem of the Exynos 9820 is that needs thermal dissipation on another grade. A superior one. And the A55 needs to be custom on the next gen. They are not efficient enough.
 
  • Like
Reactions: krumme

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
The problem of the Exynos 9820 is that needs thermal dissipation on another grade. A superior one. And the A55 needs to be custom on the next gen. They are not efficient enough.
Nope. The 10+ with the 9820 have fine dissipation. It's clearly spread better than the s9. Eg. Go look gsmarena review of the 9820 in s10 plus.
I will use that dissipation when I install my phone and then never again. This need for all out multicore punch for prolonged time is way overblown. So it really doesn't matter in real wolrd it's fine.
What matters is;
The sky high GB single core score of the cpu is felt each time you load a heavy www page. Its Apple speed. And its felt and easily recognized and it shows why single core perf does still matter a whole lot.
The speed is there in spades unless those m4 heats up and they never do in practice.
Battery efficiency seems fine to me but as said it's just normal usage.
I am sure this can be better next year. But its super fine implementation for normal usage and a heads up for the tri cluster solution.
I will bet its here to stay and others will follow.
 

insertcarehere

Senior member
Jan 17, 2013
639
607
136
Nope. The 10+ with the 9820 have fine dissipation. It's clearly spread better than the s9. Eg. Go look gsmarena review of the 9820 in s10 plus.
I will use that dissipation when I install my phone and then never again. This need for all out multicore punch for prolonged time is way overblown. So it really doesn't matter in real wolrd it's fine.
What matters is;
The sky high GB single core score of the cpu is felt each time you load a heavy www page. Its Apple speed. And its felt and easily recognized and it shows why single core perf does still matter a whole lot.
The speed is there in spades unless those m4 heats up and they never do in practice.
Battery efficiency seems fine to me but as said it's just normal usage.
I am sure this can be better next year. But its super fine implementation for normal usage and a heads up for the tri cluster solution.
I will bet its here to stay and others will follow.

Uhh alright?

Given that your point of comparison is the terrible 9810, the 9820 ought to be substantially better than that baseline. The problem is the competition isn't the 9810, its the SD855.

It's not exactly demanding to run most webpages these days, my snapdragon S10e loads them as fast as the network bandwidth allows and my Oneplus 5 (SD835) is barely slower, neither of them even think about heating up and can do it for hours on end (~8+hrs SOT for S10e on normal mode with this type of usage), maybe Exynos has caught up, but this is nothing to boast about.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
You guys just run with the established narrative. No one is expecting a 10nm derived process to beat 7nm for multicore perf and efficiency in sustained load.

Get some proper bandwidth and a strong cpu with strong single core perf shows on many pages.

The GB 4.1 single ore score for the s10plus exynos is 4522, while the apple xc is 4777 and the 855 in the mi 9 is 3503 (gsmarena bm). The m4 is impressive in -performance-.

The processor is just flat out excellent from a user perspective unless you regulary edit long videos on your phone or stuff like that. The problem is Samsung's. The size of the soc.