The kind of optimization presented here would be awesome to be exposed in a menu somewhere. Allow the user to assign performance profiles to apps. Sometimes I'd like to be able to override the OS and decide how much performance to make available to an app.
My 8 Pro 12 / 256 on (OOS11 Beta11), using Opera Touch (adblock option On), only gets me scores of 47.5-50.3. In Firefox it gets 44.1. Edge: 61.5 (this surprised me).
Not sure what to say about the whole thing. The phone feels blazing fast all the time, except when battery mode gets enabled. I can feel every app being moved over to the slower A55 cores.
Odd, i got 66.9 on my OnePlus 8 Pro on chrome, but im running OOS 11.0.7.7 as i felt it was a bit smoother than this months beta build. My T-Mobile OnePlus 7T got 61.0
They were doing this back with the Oneplus Two. The Anandtech review at https://www.anandtech.com/show/9828/the-oneplus-2-... says "the OnePlus 2 simply does not use any of its Cortex A57 cores during web browsing. This means that you're really getting a quad core Cortex A53 CPU, and with JavaScript being heavily bound by the performance of a single core it's no wonder that the results are close to those of the Moto G (2015) which uses a quad core 1.4GHz Cortex A53 CPU."
I think this is why it’s important to remember that OnePlus is legitimately a branch of Oppo and BBK. That’s it. How else were they so cocksure to do this on their second-ever product?
No new brand, especially targeted to enthusiasts, would make such a horrific throttling OS-native. New brands don’t do this. They are reputation conscious because they need sales.
OnePlus doesn’t need sales. It can go back to Oppo or BBK anytime for a cash infusion—not unlike their literal founding of a RMB 50m investment by a single investor: Oppo.
You guys are both correct. BBK here, I mean Oppo, I mean OnePlus did this "throttling" from Day 1. It was not an update. So that's a blatant attempt to bend the truth. Secondly, Andrei is also right... why doesn't the OS do this for ALL applications instead of just a few known ones?
I believe when early reviewers get their hands on it, they're going to do their own testing. And in more cases than not, they're going to report the benchmark figures (which are full-speed) and they're going to report the battery life figures (which are boosted). So they're trying to eat their cake, and have it too.
The worst part is, even with the artificial boost, it is not a very efficient phone. The QSD 888 isn't impressive, heck there's no meaningful upgrade since the QSD 855. So it means, devices like the Sony Xperia 1.3 will be reported to have issues when it comes to battery life, whilst the OnePlus 9 Pro will get a pass. Unfair. The most frustrating part is that these companies attempt to cheat, and everytime they do, we find out, we grill them for it, then they turn around and do it again !!
The early OnePlus devices (upto the 3t) were great, especially in terms of value and software, in comparison to others. They went downhill steadily after that, copying Google Pixel at every turn. Now, I think we will see people consider other brands. Shame that we've lost LG, there's No Local Warranty for expensive grey imports so brands that have low availability are also out (eg Sony, ASUS, etc etc). Whilst the major brands are continually downgrading their phones in value every year (Apple, Samsung, Google, etc etc).
Hey Andrei, I suspected Xiaomi devices of something similar to this. My redmi note 7(sd660) throttled a lot more than my brother's asus zenfone m1(sd636). Zenfone with an inferior SoC felt quicker and performed much better in heavy workloads. Although both were midrange phones, but it felt like xiaomi's software optimization was just bad. Can you verify if Xiaomi's MiUi has any similar mechanisms in place.
Can confirm xiaomi throttles the performance on MIUI. My poco f1 ran heavy application on custom android ROMs much better than MIUI while battery usage at the end of the day was pretty much same.
Perhaps Android needs a "power slider" alongside volume and brightness sliders, instead of a simple battery saver toggle? Guessing what kind of power vs. performance balance the user wants is becoming an increasingly tricky game.
Other phones have this. OnePlus refuses to add it. I can only assume because it would require them to settle.
It’s not tricky: is Samsung doing it? Is Google? Is Apple?
If OnePlus optimizations are so useful, then let us decide between them or the full performance. They have a thousand ways to tweak the performance, but just make it consistent across all applications.
Android needs a lot of things. It's an incomplete, bloated, unusable piece of crap. The fact that nobody can see this and nobody has made an alternative mobile OS is a sad statement about the species. I've been expecting more/proper features for the last 10 years, and they never came. Bring back TempleOS...
"Flagship" phones are just depressing. They are like those ostentatious BMW convertibles you see middle-aged (and beyond) men driving around in that look like Germany's answer to Cadillac.
Thank you for posting this article. I believe you are correct that web browsing requires a high transient response performance. I recently purchased a realme X7 max, dimensity1200, and I noticed that there were occasions when page interactions and rendering showed marked latency. I had been trying to troubleshoot this for a week or so. The browser bench speedometer shows a web score of 22.8 likely due to the same reason. I would appreciate it very much if you could publish a workaround for this.
I have seen this behaviour for a few months now on my oppo devices but could not find the root cause for it. thanks for pointing in the right direction. oneplus 9 series as the original nord where just late to the party and are taking the heat for it!
1. You could include this in your future phone reviews. For example a slide of contrasting browser Vs benchmarking Java performance. Considering that we spend more time running browsers than benchmarks, and considering the response in the comments section, it should be appreciated. 2. Maybe publish a workaround, Firefox is an open source software, so perhaps we can download the APK, and apps the name in the headers, and sideload the modified APK 3 This should be relatively easy to fix on OEM side if at least one of the above 2 things occur. If some can be made to backtrack on battery-gate the oppo can change its mind about it too.
It does what it is asked to do. Could it maybe do some things faster? Maybe... But for a phone I can expect to last 2 years or more I am not disappointed at all.
Will I root and ROM it? Most likely I will eventually.
OP9 Pro already has relatively poor battery life, even with this mass throttling.
OnePlus 8 Pro (4600mAh): 103h OnePlus 7T Pro (4085mAh): 100h OnePlus 9 Pro (4500mAh): 86h
So I imagine without it, it’d be close to unacceptable for a flagship and would’ve been trashed in early reviews.
It’s not just the SoC: maybe they bought less efficient panels, maybe their 5G modem stays on extra long or has a bad 4G fallback, maybe the chassis design was hurting throttling, maybe long camera recordings on the large sensor caused overheating on a warm SoC, etc.
Whats the purpose of such approach ? Why is QComm investing in new CPUs if perf/ watt is in fact that bad ? If it cannot be done , just make 2 flagship lines . Slim and slow and bulky and high perfomance ones. Its more and more clear you cannot have 1mm thick phone with blazing fast SOC
Is the X1 core actually "bad" (in the sense that the energy cost of performance is much worse than expected)? Or is the problem a lousy scheduling system that can't appropriately figure out when it makes sense to push the X1 (and the other large cores) vs when it's OK to run them at lower frequency?
Just how smart is the generic Android scheduler? How much do vendors dick around with it? How much information does it have available to do better (eg *accurate* marking of tasks submitted by apps in terms of task Qos requirements)?
I suspect we can get a feel for the generic scheduler from Android One phones like the Nokias. My Huawei has four power modes including the default, and certainly delivers far better battery life without any noticeable slow down compared to my previous Nokia 9.
My Huawei P30 Pro is the best phone I have ever owned. Whereas I used to upgrade every year, it is still fantastic. It is a real pity Huawei has been killed outside China. I also have a Huawei smart watch, bought it to replace an Apple Watch which was such a PoS needing to be charged everyday whereas the Huawei one lasts up to 14 days
Very insightful article. OnePlus has been taking a PR beating for some time now. Google Pixel is probably going to gobble all the former OnePlus niche lovers buyers. Another thing, I know it would never happen, but I'd love to see an iPhone with Android installed over iOS to test those high benchmarks you see constantly. Something tells me they wouldn't be even close then.
Not quite sure what you mean by android installed over ios but benchmarks will inevitably be far lower. iOS and A-series chips are extremely tightly integrated, and Android isn’t. It’s as simple as that.
(That’s not to say that one is better than the other. Each has their own advantages and disadvantages when seen holistically.)
I have had 2 Pixel phones and didn't keep either over 6 months.
Better than an iPhone, but nothing as good as a Samsung or OnePlus flagship.
I have over 90 tabs open in Firefox right now, and nothing about my 9 Pro feels slow, except the OEM Gallery app @ over 5600 photos and 700 videos, some of which are 8K 30fps (which the Galaxy 22 Ultra can't take over 24fps) and even then the OEM Gallery doesn't hang or glitch...
Overall, a good experience, clocking just over 11 hours SOT.
Everything is themed black, especially the keyboard, to save the eyes more than the battery. It helps for both.
I have specific apps optimized and others not optimized, and others on intelligent control.
You could paper compare all you want... But the EXPERIENCE is a good one if you just use it.
If I'm shooting 8K 30 video it's buttery smooth perfect with no need for more CPU power. If I'm editing that video, it's buttery smooth, fast processing and encoding, that doesn't leave me wishing it could be done faster.
I have a feeling the Pixel 6 won't be any better. We'll see.
Oneplus updates this last year have been near criminally broken. It's clear they have lost their development team. I would not recommend buying them at this time.
I think the more interesting question is if OnePlus is limiting performance to improve battery life why is it still noticeably worse than a Samsung? Did they have to do something like this to even get close to the proper battery life? If so, what is the root cause for this.
I could see something like this with a nice UI on top of it being a decent feature. A per app setting for max performance, balanced or max battery life that then tied into this wouldn't necessarily be bad. For some things I may not care much about their performance and wouldn't mind a little more battery life.
As a bandaid over their otherwise crappy battery life it seems poor.
The 21 or 22 Ultras are the only ones with 1440 wide screens that are comparable when figuring battery life, right?
I can't complain when my manually optimized 9 Pro gets a little over 11 hours SOT, and over 16 hours without any little top offs of the batteries.
That will be changing now that I just got a warp charger for my car, and that means my 34 minute commute to or from work is enough time to charge it from 5% to 100%.
Smaller charge increments create less battery heat, which prolongs the battery... So, I will be treating the 9 Pro just like my 7 Pro it replaced. 2 years with no noticeable loss of battery capacity.
With a 4500mAh battery it's silly to complain about battery life.
Battery life is more than enough to survive a commute and full-time work schedule without need for interim charging, but the batteries will last longer if you do.
I record and edit 4k and 8k video with my 9 Pro and use Telegram all day.. no feeling the phone isn't enough, and I never run out of battery.
Seems a non-issue to me.
Maybe you should try one for 2 or 3 weeks so it has time to settle in to your schedule and you have time to manually choose which apps to take off optimization and which ones should be left on.
For the record, Telegram is an interesting app... In general it is best left on optimize, but, if you use calling (audio or video) you have to change it to smart mode or calls will drop if another app is on top or if the screen blanks...
Switching modes for an app is easy:slide down from the top, tap the battery icon or % at the top right, hit "view detailed usage", tap the app you want to adjust, and then select Intelligent, Optimize, or Don't Optimize.
That can be done in under 4 seconds, so it's not difficult or really inconvenient.
Not sure if it's Telegram programmers or 1+ that breaks calls in manual Optimize... And it doesn't really matter for me since I make Telegram calls maybe 5 or 6x a month. That's less than 30 seconds a month switching Telegram to Intelligent Control and less than 30 seconds a month putting it back on manual Optimize mode...
YMMV, as they say. ;)
Not seeing crappy battery life at all.
Given OLED screens require more power to display lighter colors than darker ones, your theme selection will decide how much power you waste.
If you want a lighter theme you will pay the price. This is true of every OLED screen.
Think about it. Maybe try a 9 Pro. They have become pretty cheap on Swappa, where zI bought mine new in it's sealed box.
In the article the author states that the "responsiveness" of a phone is subjective and implies that it cannot be measured. I think it can be measured, but we will need new tools. I envision a very-high-framerate camera being used in conjunction with a log of nominal touch events (or visually detected touch events, possibly from an additional camera angle). This would allow the latency between a touch event and the resulting display update to be precisely measured.
Would this not be a good "bumper to bumper" measurement of device responsiveness, and possibly a good yardstick for total performance as well?
JankBench does look useful, thanks for the link. It appears to primarily measure actual display frame render rate vs. target screen refresh rate. This will show stuttering, but will not show delay between, for example, a tap and the desired UI element rendering on screen. This is the kind of thing IMO that would be pretty tough to implement on-device without prohibitive processing overhead. Instead, external monitoring (either by video mirroring, or camera) would be better. Probably some fancypants image analysis trained on the app being tested would also be required.
When the first iPhone came out, and people were commenting about how smooth it is, a team at Nokia actually acquired a high-speed camera and quantized the responsiveness of iPhone Vs the Symbian phones of the day. Seems like that was as far as it ever went through.
But yes, high-speed camera is seemingly a valid method to test responsiveness.
OnePlus has a problem with the truth. It’s an Oppo phone, through and through.
Why couldn’t they just write “Full Performance” vs “Optimized” vs “Battery Saver”?
Because then they’d need to be consistent and internal consistency is extremely scary for a weakening brand like OnePlus.
//
I mean, even interfering with benchmarks just because they have a different name?!
The entire reason people discovered the iOS throttling was through the Geekbench peaks being bimodal on identical phones but different iOS versions. It was hidden, but any benchmark app showed the difference immediately.
In this case, if you’re not Geekbench, but some other ordinary app, OnePlus kill your performance without telling you.
Andrei et al, apparently application-specific CPU / GPU throttling are becoming a theme of the OnePlus 9 Pro,
> The company has whitelisted a very small number of games, such as Fortnite, PUBG Mobile, Pokemon Go, etc., and these are the only titles allowed to run at a maximum of 90Hz. This leaves a vast majority of games unsupported and there are also no titles that run above 90Hz. Moreover, OnePlus will also drop down the refresh rate of the display from 90Hz to 60Hz in supported games when the player stops interacting, which causes a jarring drop in fluidity because, unlike static UI screens, games tend to have moving elements on-screen animating at the display's refresh rate.
Oneplus lost its credibility as an "enthusiast brand" when they dropped the headphone jack right after Apple. These revelations are actually just more of the same.
Wow, interesting and a bit weird! Initially, I thought OnePlus had seen the light and decided to not use the high performance/high power draw X1 core much, but keeping applications from the A78 cores as well makes no sense whatsoever. In general, I still see the 870 or 865+ as the better, reasonably efficient SoCs for Android smartphones vs. the 888. That being said, that OnePlus would go out of its way to throttle the 888 in one of its flagship phones like this is simply bizarre.
And, without me having any evidence for it (!), could this be the inadvertent result of OnePlus trying to cheat on benchmarks, but making an enormous hash of it? It would be incredibly incompetent programming, but maybe not impossible?
I know that; hence the "incredibly incompetent.." part in my statement. It would/does take willfulness to do what OnePlus ended up doing; either this, or they really, really pissed off whoever wrote this code. As it is, it's bizarre.
I think it's pretty simple, that it's largely what they say it is, a battery saver, only it doesn't leave the balancing of power and efficiency to the user, and they weren't being transparent about it. Why it's needed by default and why the battery life is still subpar is another issue.
Have you tried using ADB shell. It is quite easy to uninstall Google app at least on realme (oppo). The problem is that your phone will stop working properly if you unusual some apps
I noticed the weird performance of the 9 Pro almost immediately after I started using it. I get unexpected lag and stuttery scrolling throughout the UI, and scrolling in general is weirdly slow and non-kinetic. Also, there's often (but not always) a weird delay of at least a second between tapping a notification and the relevant app actually opening. The feel of using the phone is just not great, especially when going back and forth between the 9 Pro and the S21, which is why I put the 9 Pro up for sale a few days ago. Now that I've seen this article, it pretty much seals the deal for me. It's very unlikely that I'll ever buy another OnePlus phone.
Good thing you sold it early, look out for Vivo and Oppo phones too, they're essentially the same company after all. I currently own a Vivo but I don't like the way they handle design issues.
Thanks for publishing this. It was a fascinating read. Anecdotally as a OnePlus 8 Pro 8/128 user I've never noticed anything odd, and it feels substantially faster in use than the 6 I switched from. I'd not know how to check for this sort of behaviour formally, but inertial scrolling seems fine.
I just got a oneplus 9 pro this week and can confirm having these types of performance issues especially on Google news and twitter using 120hrz scrolling very jittery and reported issues to oneplus they stated they know about issue and looking into it !!! My old oneplus 5 preforms better, feel cheated
A lot of these are because of the bad 888 design + ARMs BS marketing on the X1 core. No wonder Huawei chose to stay on A77 for the Kirin 9000. Samsung's Exynos 2100 clearly handles performance and power management of the new cores better. Something really smelled when QC used exactly the same frequncies from 865 in 888.
It now looks more and more like OnePlus throttles these "flagship" phones (with the 888) to reduce their phones' tendency to overheat even at moderate ambient temperatures (around 20 C). So, yes, that gelding of the 888 is on purpose, but the "battery saving" argument is apparently pretense. In plain English, they did a poor job on the thermal design of their phone, and are now "fixing" the problem by severely diminishing performance. This is outrageous; anyone buying an 888 equipped phone has every reason to expect the performance promised by that SoC, and now find themselves with a smartphone that performs 1-2 classes below that. Thanks to Andrei for making OnePlus owning up to their misdeed! Now, OnePlus should make those who bought their phone whole, either with a sizeable rebate or right to return for full refund. As for me, I am glad that I didn't buy one to begin with.
Forgot to add: This also means that testing phones should, once again, test their thermals; that can be as simple as running the phone full tilt, measure temperatures (IR thermometer), and repeat benchmarks several times and look for throttling (something Andrei does in his tests). The OnePlus phones here apparently also overheated when shooting video for more than a minute, so that behavior is, apparently, also worth looking at.
I would be curious if they implemented these changes for older OnePlus phones as well. The upgrade to Android 11 on my OnePlus 7 Pro has been suspicious.
1 question about browser expirience: Is this "throttling" of browser visible from the end-user? Is there subjective slowness in browser response? Who cares for benchmark score if subjectively, when a real user uses the phone its responsive enough? just a 2 cents. never had Oppo phone nor planning to have
Uh, have you read the Chromium development discussion? They purposely limit threads to running on the LITTLE CPU cores to reduce power consumption........
Quite fascinating, thank you Andrei. What's beyond me is why they didn't just make it a system wide setting. Like for example Sony and their "STAMINA" modes. Have it default to efficiency if they so wish but give users an option. They could even have it be a first setup thing, default to efficiency but let an user choose between it and high perf/power save. Then you could also have different profiles and scheduling depending on location/time etc.
Hey Andrei, May I suggest we use the terms "Blocked List" and "Allowed List" instead of "Blacklist" and "Whitelist"? We should remember we are in 21st century and we should be inclusive and diverse.
Yes, since this "feature" is implemented to the kernel. Custom ROMs still use the base kernel code which is provided by OnePlus.
NOTE: Im currently using a Nord N10 The results are (speedometer 2) Chrome: 15~ Opera touch: 45~
Sometimes Opera touch gets a score of 20 or lower (even 12) sometimes not, weird... This is not because oh high temperatures, heck I even tried to put the phone into the fridge for 40-50 minutes and re-run the tests, same results... Also tried with LineageOS, same results
Altho even if the speedometer 2 scores are low (around 15) for me, the browsing performance is so-so okay some occasional micro stutters.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
86 Comments
Back to Article
RamarC - Tuesday, July 6, 2021 - link
Fascinating... I nerded out😂YB1064 - Tuesday, July 6, 2021 - link
Excellent article! Thank you.at_clucks - Tuesday, July 6, 2021 - link
Andrei is AT's "flagship". :)The kind of optimization presented here would be awesome to be exposed in a menu somewhere. Allow the user to assign performance profiles to apps. Sometimes I'd like to be able to override the OS and decide how much performance to make available to an app.
benderw - Tuesday, July 6, 2021 - link
OnePlus 8T with SD865: Speedometer 2.0 - score 70 (firmware OP8T_O2_BETA_6).So hopefully only the new "faster" phones are 4 times slower than my old one :)
Cryio - Tuesday, July 6, 2021 - link
70 with an 8T? Interesting.My 8 Pro 12 / 256 on (OOS11 Beta11), using Opera Touch (adblock option On), only gets me scores of 47.5-50.3. In Firefox it gets 44.1. Edge: 61.5 (this surprised me).
Not sure what to say about the whole thing. The phone feels blazing fast all the time, except when battery mode gets enabled. I can feel every app being moved over to the slower A55 cores.
Seremedy - Tuesday, July 6, 2021 - link
Odd, i got 66.9 on my OnePlus 8 Pro on chrome, but im running OOS 11.0.7.7 as i felt it was a bit smoother than this months beta build. My T-Mobile OnePlus 7T got 61.0QinX - Wednesday, July 7, 2021 - link
My iPhone 6s gets 54.2. Glad to see it’s still holding up in this regard. Will be switching to the new iPhone 12s/13 Pro Max this septemberjamesindevon - Tuesday, July 6, 2021 - link
They were doing this back with the Oneplus Two. The Anandtech review at https://www.anandtech.com/show/9828/the-oneplus-2-... says "the OnePlus 2 simply does not use any of its Cortex A57 cores during web browsing. This means that you're really getting a quad core Cortex A53 CPU, and with JavaScript being heavily bound by the performance of a single core it's no wonder that the results are close to those of the Moto G (2015) which uses a quad core 1.4GHz Cortex A53 CPU."ikjadoon - Tuesday, July 6, 2021 - link
I think this is why it’s important to remember that OnePlus is legitimately a branch of Oppo and BBK. That’s it. How else were they so cocksure to do this on their second-ever product?No new brand, especially targeted to enthusiasts, would make such a horrific throttling OS-native. New brands don’t do this. They are reputation conscious because they need sales.
OnePlus doesn’t need sales. It can go back to Oppo or BBK anytime for a cash infusion—not unlike their literal founding of a RMB 50m investment by a single investor: Oppo.
https://kr-asia.com/meet-bbk-the-worlds-largest-ph...
Such a damn shame. And your article is a great reminder! That article is when I said I’d not buy another OP phone after my OnePlus One.
Kangal - Thursday, July 8, 2021 - link
You guys are both correct.BBK here, I mean Oppo, I mean OnePlus did this "throttling" from Day 1. It was not an update. So that's a blatant attempt to bend the truth. Secondly, Andrei is also right... why doesn't the OS do this for ALL applications instead of just a few known ones?
I believe when early reviewers get their hands on it, they're going to do their own testing. And in more cases than not, they're going to report the benchmark figures (which are full-speed) and they're going to report the battery life figures (which are boosted). So they're trying to eat their cake, and have it too.
The worst part is, even with the artificial boost, it is not a very efficient phone. The QSD 888 isn't impressive, heck there's no meaningful upgrade since the QSD 855. So it means, devices like the Sony Xperia 1.3 will be reported to have issues when it comes to battery life, whilst the OnePlus 9 Pro will get a pass. Unfair. The most frustrating part is that these companies attempt to cheat, and everytime they do, we find out, we grill them for it, then they turn around and do it again !!
The early OnePlus devices (upto the 3t) were great, especially in terms of value and software, in comparison to others. They went downhill steadily after that, copying Google Pixel at every turn. Now, I think we will see people consider other brands. Shame that we've lost LG, there's No Local Warranty for expensive grey imports so brands that have low availability are also out (eg Sony, ASUS, etc etc). Whilst the major brands are continually downgrading their phones in value every year (Apple, Samsung, Google, etc etc).
s.yu - Saturday, July 10, 2021 - link
Good to know, now the question is does Vivo do this...on a wide scale, especially.Kopemon - Tuesday, July 6, 2021 - link
Hey Andrei, I suspected Xiaomi devices of something similar to this. My redmi note 7(sd660) throttled a lot more than my brother's asus zenfone m1(sd636). Zenfone with an inferior SoC felt quicker and performed much better in heavy workloads. Although both were midrange phones, but it felt like xiaomi's software optimization was just bad. Can you verify if Xiaomi's MiUi has any similar mechanisms in place.doomslayer007 - Wednesday, July 7, 2021 - link
Can confirm xiaomi throttles the performance on MIUI. My poco f1 ran heavy application on custom android ROMs much better than MIUI while battery usage at the end of the day was pretty much same.brucethemoose - Tuesday, July 6, 2021 - link
Testing like this is why I come to Anandtech.Perhaps Android needs a "power slider" alongside volume and brightness sliders, instead of a simple battery saver toggle? Guessing what kind of power vs. performance balance the user wants is becoming an increasingly tricky game.
ikjadoon - Tuesday, July 6, 2021 - link
Other phones have this. OnePlus refuses to add it. I can only assume because it would require them to settle.It’s not tricky: is Samsung doing it? Is Google? Is Apple?
If OnePlus optimizations are so useful, then let us decide between them or the full performance. They have a thousand ways to tweak the performance, but just make it consistent across all applications.
flyingpants265 - Tuesday, July 6, 2021 - link
Android needs a lot of things. It's an incomplete, bloated, unusable piece of crap. The fact that nobody can see this and nobody has made an alternative mobile OS is a sad statement about the species. I've been expecting more/proper features for the last 10 years, and they never came. Bring back TempleOS...chaose - Wednesday, July 7, 2021 - link
well Amazon has a good android based OS. Huawei's new HongMeng OS is pretty good as wellPaulHoule - Tuesday, July 6, 2021 - link
"Flagship" phones are just depressing. They are like those ostentatious BMW convertibles you see middle-aged (and beyond) men driving around in that look like Germany's answer to Cadillac.Chaser - Tuesday, July 6, 2021 - link
Cognitive dissonance.ads295 - Tuesday, July 6, 2021 - link
First I read "Xi missing" on the home page and I was like whattttt... Can it be true?drajitshnew - Tuesday, July 6, 2021 - link
Thank you for posting this article. I believe you are correct that web browsing requires a high transient response performance. I recently purchased a realme X7 max, dimensity1200, and I noticed that there were occasions when page interactions and rendering showed marked latency. I had been trying to troubleshoot this for a week or so. The browser bench speedometer shows a web score of 22.8 likely due to the same reason.I would appreciate it very much if you could publish a workaround for this.
Andrei Frumusanu - Tuesday, July 6, 2021 - link
I don't have many other recent OPPO phones right now but it wouldn't surprise me if this is used wider spread among their brands.Unfortunately you can't work around it as it's baked into the OS.
testuser2020 - Wednesday, July 7, 2021 - link
I have seen this behaviour for a few months now on my oppo devices but could not find the root cause for it. thanks for pointing in the right direction.oneplus 9 series as the original nord where just late to the party and are taking the heat for it!
drajitshnew - Monday, July 19, 2021 - link
1. You could include this in your future phone reviews. For example a slide of contrasting browser Vs benchmarking Java performance. Considering that we spend more time running browsers than benchmarks, and considering the response in the comments section, it should be appreciated.2. Maybe publish a workaround, Firefox is an open source software, so perhaps we can download the APK, and apps the name in the headers, and sideload the modified APK
3 This should be relatively easy to fix on OEM side if at least one of the above 2 things occur. If some can be made to backtrack on battery-gate the oppo can change its mind about it too.
BDProductions - Sunday, August 22, 2021 - link
3rd party ROM.....You can work around this.
Root is easy on 1+ devices.
I have no issues with my 9 Pro performance.
It does what it is asked to do. Could it maybe do some things faster? Maybe... But for a phone I can expect to last 2 years or more I am not disappointed at all.
Will I root and ROM it? Most likely I will eventually.
sharathc - Tuesday, July 6, 2021 - link
Andrei, if you do like this, Google or Apple will kidnap you soon and hire you.Yours,
Huge Fan
melgross - Tuesday, July 6, 2021 - link
Might be what he’s hoping for.Leeea - Tuesday, July 6, 2021 - link
I wonder why they do not get the battery life of their competitors?( all of these phones have spyware/telemetry in them, I wonder if oneplus is worst then most )
ikjadoon - Tuesday, July 6, 2021 - link
OP9 Pro already has relatively poor battery life, even with this mass throttling.OnePlus 8 Pro (4600mAh): 103h
OnePlus 7T Pro (4085mAh): 100h
OnePlus 9 Pro (4500mAh): 86h
So I imagine without it, it’d be close to unacceptable for a flagship and would’ve been trashed in early reviews.
It’s not just the SoC: maybe they bought less efficient panels, maybe their 5G modem stays on extra long or has a bad 4G fallback, maybe the chassis design was hurting throttling, maybe long camera recordings on the large sensor caused overheating on a warm SoC, etc.
ikjadoon - Tuesday, July 6, 2021 - link
Source: https://www.gsmarena.com/oneplus_9_pro-review-2241...RaduR - Tuesday, July 6, 2021 - link
Whats the purpose of such approach ? Why is QComm investing in new CPUs if perf/ watt is in fact that bad ? If it cannot be done , just make 2 flagship lines . Slim and slow and bulky and high perfomance ones. Its more and more clear you cannot have 1mm thick phone with blazing fast SOCname99 - Tuesday, July 6, 2021 - link
Is the X1 core actually "bad" (in the sense that the energy cost of performance is much worse than expected)?Or is the problem a lousy scheduling system that can't appropriately figure out when it makes sense to push the X1 (and the other large cores) vs when it's OK to run them at lower frequency?
Just how smart is the generic Android scheduler? How much do vendors dick around with it? How much information does it have available to do better (eg *accurate* marking of tasks submitted by apps in terms of task Qos requirements)?
BedfordTim - Wednesday, July 7, 2021 - link
I suspect we can get a feel for the generic scheduler from Android One phones like the Nokias. My Huawei has four power modes including the default, and certainly delivers far better battery life without any noticeable slow down compared to my previous Nokia 9.Speedfriend - Wednesday, July 7, 2021 - link
My Huawei P30 Pro is the best phone I have ever owned. Whereas I used to upgrade every year, it is still fantastic. It is a real pity Huawei has been killed outside China. I also have a Huawei smart watch, bought it to replace an Apple Watch which was such a PoS needing to be charged everyday whereas the Huawei one lasts up to 14 daysChaser - Tuesday, July 6, 2021 - link
Very insightful article. OnePlus has been taking a PR beating for some time now. Google Pixel is probably going to gobble all the former OnePlus niche lovers buyers. Another thing, I know it would never happen, but I'd love to see an iPhone with Android installed over iOS to test those high benchmarks you see constantly. Something tells me they wouldn't be even close then.brucethemoose - Tuesday, July 6, 2021 - link
There's an effort to get Android running on iOS devices, but AFAIK it hasn't made much progress.The M1 could change that, as there's a whole lot of developer interest in running desktop linux on an M1.
Tomatotech - Wednesday, July 7, 2021 - link
Not quite sure what you mean by android installed over ios but benchmarks will inevitably be far lower. iOS and A-series chips are extremely tightly integrated, and Android isn’t. It’s as simple as that.(That’s not to say that one is better than the other. Each has their own advantages and disadvantages when seen holistically.)
Maxpower27 - Wednesday, July 7, 2021 - link
I have a feeling the Pixel 6 is going to bring a whole bunch of OnePlus people back into the fold in a way that the 5 couldn't manage to do.BDProductions - Sunday, August 22, 2021 - link
Doubt that.I have had 2 Pixel phones and didn't keep either over 6 months.
Better than an iPhone, but nothing as good as a Samsung or OnePlus flagship.
I have over 90 tabs open in Firefox right now, and nothing about my 9 Pro feels slow, except the OEM Gallery app @ over 5600 photos and 700 videos, some of which are 8K 30fps (which the Galaxy 22 Ultra can't take over 24fps) and even then the OEM Gallery doesn't hang or glitch...
Overall, a good experience, clocking just over 11 hours SOT.
Everything is themed black, especially the keyboard, to save the eyes more than the battery. It helps for both.
I have specific apps optimized and others not optimized, and others on intelligent control.
You could paper compare all you want... But the EXPERIENCE is a good one if you just use it.
If I'm shooting 8K 30 video it's buttery smooth perfect with no need for more CPU power. If I'm editing that video, it's buttery smooth, fast processing and encoding, that doesn't leave me wishing it could be done faster.
I have a feeling the Pixel 6 won't be any better. We'll see.
coburn_c - Tuesday, July 6, 2021 - link
Oneplus updates this last year have been near criminally broken. It's clear they have lost their development team. I would not recommend buying them at this time.kpb321 - Tuesday, July 6, 2021 - link
I think the more interesting question is if OnePlus is limiting performance to improve battery life why is it still noticeably worse than a Samsung? Did they have to do something like this to even get close to the proper battery life? If so, what is the root cause for this.I could see something like this with a nice UI on top of it being a decent feature. A per app setting for max performance, balanced or max battery life that then tied into this wouldn't necessarily be bad. For some things I may not care much about their performance and wouldn't mind a little more battery life.
As a bandaid over their otherwise crappy battery life it seems poor.
BDProductions - Sunday, August 22, 2021 - link
Notably worse than which Samsung?The 21 or 22 Ultras are the only ones with 1440 wide screens that are comparable when figuring battery life, right?
I can't complain when my manually optimized 9 Pro gets a little over 11 hours SOT, and over 16 hours without any little top offs of the batteries.
That will be changing now that I just got a warp charger for my car, and that means my 34 minute commute to or from work is enough time to charge it from 5% to 100%.
Smaller charge increments create less battery heat, which prolongs the battery... So, I will be treating the 9 Pro just like my 7 Pro it replaced. 2 years with no noticeable loss of battery capacity.
With a 4500mAh battery it's silly to complain about battery life.
Battery life is more than enough to survive a commute and full-time work schedule without need for interim charging, but the batteries will last longer if you do.
I record and edit 4k and 8k video with my 9 Pro and use Telegram all day.. no feeling the phone isn't enough, and I never run out of battery.
Seems a non-issue to me.
Maybe you should try one for 2 or 3 weeks so it has time to settle in to your schedule and you have time to manually choose which apps to take off optimization and which ones should be left on.
For the record, Telegram is an interesting app... In general it is best left on optimize, but, if you use calling (audio or video) you have to change it to smart mode or calls will drop if another app is on top or if the screen blanks...
Switching modes for an app is easy:slide down from the top, tap the battery icon or % at the top right, hit "view detailed usage", tap the app you want to adjust, and then select Intelligent, Optimize, or Don't Optimize.
That can be done in under 4 seconds, so it's not difficult or really inconvenient.
Not sure if it's Telegram programmers or 1+ that breaks calls in manual Optimize... And it doesn't really matter for me since I make Telegram calls maybe 5 or 6x a month. That's less than 30 seconds a month switching Telegram to Intelligent Control and less than 30 seconds a month putting it back on manual Optimize mode...
YMMV, as they say. ;)
Not seeing crappy battery life at all.
Given OLED screens require more power to display lighter colors than darker ones, your theme selection will decide how much power you waste.
If you want a lighter theme you will pay the price. This is true of every OLED screen.
Think about it. Maybe try a 9 Pro.
They have become pretty cheap on Swappa, where zI bought mine new in it's sealed box.
apeironian - Tuesday, July 6, 2021 - link
In the article the author states that the "responsiveness" of a phone is subjective and implies that it cannot be measured. I think it can be measured, but we will need new tools. I envision a very-high-framerate camera being used in conjunction with a log of nominal touch events (or visually detected touch events, possibly from an additional camera angle). This would allow the latency between a touch event and the resulting display update to be precisely measured.Would this not be a good "bumper to bumper" measurement of device responsiveness, and possibly a good yardstick for total performance as well?
ikjadoon - Tuesday, July 6, 2021 - link
IIRC, XDA gets a similar result with their JankBench test suite. It measures frame times seemingly very accurately.It’s not perfect (it’s a specific suite developed by Google, IIRC), but it’s something that can be used even today.
Example results: https://www.xda-developers.com/asus-zenfone-8-revi...
apeironian - Tuesday, July 6, 2021 - link
JankBench does look useful, thanks for the link. It appears to primarily measure actual display frame render rate vs. target screen refresh rate. This will show stuttering, but will not show delay between, for example, a tap and the desired UI element rendering on screen. This is the kind of thing IMO that would be pretty tough to implement on-device without prohibitive processing overhead. Instead, external monitoring (either by video mirroring, or camera) would be better. Probably some fancypants image analysis trained on the app being tested would also be required.shadowjk - Wednesday, July 7, 2021 - link
When the first iPhone came out, and people were commenting about how smooth it is, a team at Nokia actually acquired a high-speed camera and quantized the responsiveness of iPhone Vs the Symbian phones of the day. Seems like that was as far as it ever went through.But yes, high-speed camera is seemingly a valid method to test responsiveness.
ikjadoon - Tuesday, July 6, 2021 - link
OnePlus has a problem with the truth. It’s an Oppo phone, through and through.Why couldn’t they just write “Full Performance” vs “Optimized” vs “Battery Saver”?
Because then they’d need to be consistent and internal consistency is extremely scary for a weakening brand like OnePlus.
//
I mean, even interfering with benchmarks just because they have a different name?!
The entire reason people discovered the iOS throttling was through the Geekbench peaks being bimodal on identical phones but different iOS versions. It was hidden, but any benchmark app showed the difference immediately.
In this case, if you’re not Geekbench, but some other ordinary app, OnePlus kill your performance without telling you.
ikjadoon - Tuesday, July 6, 2021 - link
Andrei et al, apparently application-specific CPU / GPU throttling are becoming a theme of the OnePlus 9 Pro,> The company has whitelisted a very small number of games, such as Fortnite, PUBG Mobile, Pokemon Go, etc., and these are the only titles allowed to run at a maximum of 90Hz. This leaves a vast majority of games unsupported and there are also no titles that run above 90Hz. Moreover, OnePlus will also drop down the refresh rate of the display from 90Hz to 60Hz in supported games when the player stops interacting, which causes a jarring drop in fluidity because, unlike static UI screens, games tend to have moving elements on-screen animating at the display's refresh rate.
https://www.gsmarena.com/oneplus_9_pro-review-2241...
What is the purpose of forcing this? Shouldn’t they just let us choose a refresh rate per game? Why lock them at all?
Maxpower27 - Wednesday, July 7, 2021 - link
Pokémon Go player here, can confirm the "jarring drop in fluidity" is real and constant. The game runs much, much better on an S21.s.yu - Wednesday, July 7, 2021 - link
Oneplus lost its credibility as an "enthusiast brand" when they dropped the headphone jack right after Apple. These revelations are actually just more of the same.eastcoast_pete - Tuesday, July 6, 2021 - link
Wow, interesting and a bit weird! Initially, I thought OnePlus had seen the light and decided to not use the high performance/high power draw X1 core much, but keeping applications from the A78 cores as well makes no sense whatsoever.In general, I still see the 870 or 865+ as the better, reasonably efficient SoCs for Android smartphones vs. the 888. That being said, that OnePlus would go out of its way to throttle the 888 in one of its flagship phones like this is simply bizarre.
eastcoast_pete - Tuesday, July 6, 2021 - link
And, without me having any evidence for it (!), could this be the inadvertent result of OnePlus trying to cheat on benchmarks, but making an enormous hash of it? It would be incredibly incompetent programming, but maybe not impossible?Wereweeb - Wednesday, July 7, 2021 - link
That's... not how writing code works.eastcoast_pete - Wednesday, July 7, 2021 - link
I know that; hence the "incredibly incompetent.." part in my statement. It would/does take willfulness to do what OnePlus ended up doing; either this, or they really, really pissed off whoever wrote this code. As it is, it's bizarre.s.yu - Wednesday, July 7, 2021 - link
I think it's pretty simple, that it's largely what they say it is, a battery saver, only it doesn't leave the balancing of power and efficiency to the user, and they weren't being transparent about it.Why it's needed by default and why the battery life is still subpar is another issue.
Wereweeb - Wednesday, July 7, 2021 - link
Thank you Andrei.And yes, as it's been said, I'd love to be able to pick and choose which apps to condemn to the A55's. I'm looking at you, uninstallable google apps.
drajitshnew - Monday, July 19, 2021 - link
Have you tried using ADB shell. It is quite easy to uninstall Google app at least on realme (oppo). The problem is that your phone will stop working properly if you unusual some appszodiacfml - Wednesday, July 7, 2021 - link
Oneplus should stop pretending a premium brand. This mechanism is so cheap and easys.yu - Wednesday, July 7, 2021 - link
...then so should Leica 😂Maxpower27 - Wednesday, July 7, 2021 - link
I noticed the weird performance of the 9 Pro almost immediately after I started using it. I get unexpected lag and stuttery scrolling throughout the UI, and scrolling in general is weirdly slow and non-kinetic. Also, there's often (but not always) a weird delay of at least a second between tapping a notification and the relevant app actually opening. The feel of using the phone is just not great, especially when going back and forth between the 9 Pro and the S21, which is why I put the 9 Pro up for sale a few days ago. Now that I've seen this article, it pretty much seals the deal for me. It's very unlikely that I'll ever buy another OnePlus phone.s.yu - Wednesday, July 7, 2021 - link
Good thing you sold it early, look out for Vivo and Oppo phones too, they're essentially the same company after all. I currently own a Vivo but I don't like the way they handle design issues.caribbeanblue - Wednesday, July 7, 2021 - link
Good to know the latest Snapdragon phones are as much as 5-6% faster in a browser than my 4 year old iPad Pro. (Speedometer 2.0 Score: 101)Ken_g6 - Wednesday, July 7, 2021 - link
If this is "battery optimization", then Volkswagen's diesels were just doing "pollution optimization".eastcoast_pete - Wednesday, July 7, 2021 - link
Well, they did increase their NOx output, so that was "optimized" (:hanssonrickard - Wednesday, July 7, 2021 - link
And people complained about Apple crippling performance when the battery was starting to get old.I would say this is far worse.
Spunjji - Wednesday, July 7, 2021 - link
Thanks for publishing this. It was a fascinating read. Anecdotally as a OnePlus 8 Pro 8/128 user I've never noticed anything odd, and it feels substantially faster in use than the 6 I switched from. I'd not know how to check for this sort of behaviour formally, but inertial scrolling seems fine.ZEXX3S - Wednesday, July 7, 2021 - link
It's happening on OnePlus 6t scoring 41.2 and 36.7 running the Chrome benchmark twice.arashi - Wednesday, July 7, 2021 - link
... That's just old silicon showing it's age you mouth breather.chrmica - Thursday, July 8, 2021 - link
I just got a oneplus 9 pro this week and can confirm having these types of performance issues especially on Google news and twitter using 120hrz scrolling very jittery and reported issues to oneplus they stated they know about issue and looking into it !!! My old oneplus 5 preforms better, feel cheatedchrmica - Thursday, July 8, 2021 - link
Just ran the speedometer test got 13.1 !!!! Oneplis 9 pro and 32 on oneplus 5 this is ridiculousfrbeckenbauer - Thursday, July 8, 2021 - link
All these chinese phones just exist as cheap hardware to install lineageos onJon Tseng - Thursday, July 8, 2021 - link
My drawer full of $100 unlocked redmis agrees with you! :-ptkSteveFOX - Thursday, July 8, 2021 - link
A lot of these are because of the bad 888 design + ARMs BS marketing on the X1 core. No wonder Huawei chose to stay on A77 for the Kirin 9000.Samsung's Exynos 2100 clearly handles performance and power management of the new cores better. Something really smelled when QC used exactly the same frequncies from 865 in 888.
eastcoast_pete - Thursday, July 8, 2021 - link
It now looks more and more like OnePlus throttles these "flagship" phones (with the 888) to reduce their phones' tendency to overheat even at moderate ambient temperatures (around 20 C). So, yes, that gelding of the 888 is on purpose, but the "battery saving" argument is apparently pretense. In plain English, they did a poor job on the thermal design of their phone, and are now "fixing" the problem by severely diminishing performance. This is outrageous; anyone buying an 888 equipped phone has every reason to expect the performance promised by that SoC, and now find themselves with a smartphone that performs 1-2 classes below that.Thanks to Andrei for making OnePlus owning up to their misdeed! Now, OnePlus should make those who bought their phone whole, either with a sizeable rebate or right to return for full refund. As for me, I am glad that I didn't buy one to begin with.
eastcoast_pete - Thursday, July 8, 2021 - link
Forgot to add: This also means that testing phones should, once again, test their thermals; that can be as simple as running the phone full tilt, measure temperatures (IR thermometer), and repeat benchmarks several times and look for throttling (something Andrei does in his tests). The OnePlus phones here apparently also overheated when shooting video for more than a minute, so that behavior is, apparently, also worth looking at.Jagar - Thursday, July 8, 2021 - link
I would be curious if they implemented these changes for older OnePlus phones as well. The upgrade to Android 11 on my OnePlus 7 Pro has been suspicious.chavv - Friday, July 9, 2021 - link
1 question about browser expirience:Is this "throttling" of browser visible from the end-user?
Is there subjective slowness in browser response?
Who cares for benchmark score if subjectively, when a real user uses the phone its responsive enough?
just a 2 cents. never had Oppo phone nor planning to have
sandeep_r_89 - Friday, July 9, 2021 - link
Uh, have you read the Chromium development discussion? They purposely limit threads to running on the LITTLE CPU cores to reduce power consumption........This isn't a OnePlus problem.....
catadeluxe - Friday, July 9, 2021 - link
Look at the following graph, other phones with the S888 do perform better than the OP 9 Prohttps://images.anandtech.com/graphs/graph16794/122...
lagittaja - Monday, July 12, 2021 - link
Quite fascinating, thank you Andrei.What's beyond me is why they didn't just make it a system wide setting. Like for example Sony and their "STAMINA" modes. Have it default to efficiency if they so wish but give users an option.
They could even have it be a first setup thing, default to efficiency but let an user choose between it and high perf/power save. Then you could also have different profiles and scheduling depending on location/time etc.
As it is, is just dumb in my opinion.
Gratin - Tuesday, July 13, 2021 - link
Hey Andrei,May I suggest we use the terms "Blocked List" and "Allowed List" instead of "Blacklist" and "Whitelist"?
We should remember we are in 21st century and we should be inclusive and diverse.
LiverpoolFC5903 - Wednesday, July 14, 2021 - link
WTF???This is exactly what they mean when they say "PC gone mad".
Danielsansan - Saturday, July 17, 2021 - link
Will the problem still be there if you install a custom ROM like lineage os ?RTamas91 - Saturday, October 2, 2021 - link
Yes, since this "feature" is implemented to the kernel. Custom ROMs still use the base kernel code which is provided by OnePlus.NOTE: Im currently using a Nord N10
The results are (speedometer 2)
Chrome: 15~
Opera touch: 45~
Sometimes Opera touch gets a score of 20 or lower (even 12) sometimes not, weird...
This is not because oh high temperatures, heck I even tried to put the phone into the fridge for 40-50 minutes and re-run the tests, same results...
Also tried with LineageOS, same results
Altho even if the speedometer 2 scores are low (around 15) for me, the browsing performance is so-so okay some occasional micro stutters.
TomMN - Sunday, October 17, 2021 - link
I was thinking of 1 of 4 phones to replace my OnePlus 7T. Pixel 6 Pro, iPhone 13 Pro, S22 Pro, and removed the OnePlus 9 Pro (too slow).