My specific issue involves SetCPU but I'm wondering about the larger scheme of how things work.
My phone gets hot in the car dock when running GPS, Navigation, and Bluetooth (I don't turn it on but I think my dock does it automatically). The general consensus is to underclock slightly when charging because charging generates heat, and I have a SetCPU profile to do that. Then comes the issue of heat generated from GPS and Bluetooth, and possibly processor load when running apps while docked, so I have two profiles for that: one to underclock a few steps down from the charging CPU speed when the temperature of the phone is at 40* C, another to clock down one more step when the temp hits 42* C, which I feel is at the point the phone gets a little too warm for use.
I've noticed that my phone doesn't necessarily stay cooler and in fact, sometimes it gets hotter than before I started using SetCPU. Over the past few days I've seen temps of near 45* C (113* F), toasty enough that I have to let my phone cool down before I can use it again.
So, my question is if underclocking too much is bad. For example, if I'm underclocking too much, does that strain the processor while it is trying to run apps and hardware? Is it better to let the system handle processor speed? I see this in driving a manual/standard transmission car - if I attack a hill in a high gear, I can eventually reach the top but with increased load on my engine. Maybe the processor speed works opposite - too low and it strains too much?
I used to be concerned about clock speed vs. battery life but now that I'm docking in my car more and noticing the heat, I'm more concerned about the heat. I'm currently using redstar's kernel but get the same results with that or the CM kernel that comes with the nightlies.
If it helps to troubleshoot, or if you want to critique me, here's my profile list.
Normal: 245 min, 998 max (default settings)
Priority 100 - temp > 42*C: 460 max, 245 min
P 90 - temp > 40*C: 499 max, 245 min
P 80 - screen off: 384 max, 245 min
P 70 - charging AC: 576 max, 245 min
(SetCPU related question, sorry if this is too far off forum topic) Are there better profile settings I could be using?
Also, I don't quite understand undervolting in terms of kernels. What effect does that have, if any?
The Broadcom® BCM47755 location hub is a single-chip device that combines location awareness capabilities with the typical functions of a sensor hub. The combination provides synergistic benefits that cannot be achieved with multiple ICs, such as low power consumption, higher accuracy, reduced footprint and a smaller BOM.
The BCM47755 supports two frequencies (L1+L5), and as a result, achieves lane-level accuracy outdoors and much higher resistance to multipath and reflected signals in urban scenarios, as well as higher interference and jamming immunity.
Furthermore, the BCM47755 incorporates numerous technologies that enable ultralow power consumption in both the location function and the sensor hub function. The device features a low-power RF path, a Big/Little CPU configuration composed of an ARM-based 32-bit Cortex-M4F (CM4), an ARM-based Cortex-M0 (CM0), and is built in a 28 nm process.
https://www.broadcom.com/products/wireless/gnss-gps-socs/bcm47755
I comment on real life performance. I cycle 5 miles to and from work and use Strava to track my ride.
The time between opening Strava and getting a lock on my position is insanely quick. Pretty much instant.
Also, on other phones GPS tracking would to sap my battery. With these activities and typical usage throughout the day I have only gone through 30% of the battery. Took it off charge 15 hours ago.
Here is a test I did on my S21 Ultra (Exynos), that started a bit spontaneously with me wondering about the impact on battery of different refresh rates and resolutions, together with use cases (touching the screen or just looking at it). So I started playing around and landed in some kind of improvised test.
The phone is an S21 Ultra Exynos model, as already mentioned. The model number is SM-G998B/DS. The FW was G998BXXU2AUBB.
The test was performed at 20% screen brightness and 30% battery left on the phone. The phone was charged between tests to hover around 30% battery, +/- 2%, i.e. it was between 28% and 32% during the tests.
There were two types of tests; "Screen Interaction Test" that intended to find the current consumption while repeatedly touching the screen and therefore keeping the CPU ready, and an "Idle Current Consumption Test" where the screen was just turned on for 3 minutes without interacting with the phone. In both tests, the phone was running the app Ampere in the foreground, thus displaying it.
The data logging was done manually by me watching the app Ampere and noting down its shown current consumption every second into a Google Sheets document. Ampere collects 50 data points, discards the 10 lowest and 10 highest, then takes the average of the remaining 30 points before presenting the current consumption. These values were gathered for 5 minutes in the interaction test, and 3 minutes in the idle test, in order to acquire a robust sample set that was resilient against temporary peaks in current consumption. Since I don't have control over the phone's background processes, it was possible that a background task would cause a sudden peak in current consumption. Averaging these values makes the measurement more resilient against such peaks.
The phone was put in Airplane Mode in order to eliminate Wifi, 4G, Bluetooth and all those things from the measurement.
Test ResultsInteracting With Screen in 120 Hz Adaptive Refresh Rate and Different ResolutionsThe screen was tapped 3-4 times per second in order to keep it active. The phone was displaying the app Ampere. Test duration was 5 minutes and data was logged about every second from Ampere, and then averaged over the entire test duration.
HD 120 Hz Adaptive Refresh Rate: 204 mA
FHD 120 Hz Adaptive Refresh Rate: 184 mA
WQHD 120 Hz Adaptive Refresh Rate: 206 mA
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Interacting With Screen in 60 Hz Adaptive Refresh Rate and Different ResolutionsThe screen was tapped 3-4 times per second in order to keep it active. The phone was displaying the app Ampere. Test duration was 5 minutes and data was logged about every second from Ampere, and then averaged over the entire test duration.
HD 60 Hz Standard Refresh Rate: 175 mA
FHD 60 Hz Standard Refresh Rate: 161 mA
WQHD 60 Hz Standard Refresh Rate: 176 mA
Idling With Screen On in 60 Hz Standard Refresh Rate and Different ResolutionsThe screen was turned on it displayed the app Ampere. The screen was not interacted with at all for the whole 3 minutes this test took. The test was only done for 60 Hz because there is no point in doing it for 120 Hz since the screen drops down to 60 Hz when idling in the app Ampere. The values are the average value of the entire test duration (3 minutes).
HD 60 Hz Standard Refresh Rate, Screen Idling: 144 mA
FHD 60 Hz Standard Refresh Rate, Screen Idling: 145 mA
WQHD 60 Hz Standard Refresh Rate, Screen Idling: 147 mA
Bonus Test: 48 Hz and 96 Hz Display ModeThe two hidden display modes were also tested but there were no advantages in using these modes regarding current draw. Plus, these modes actually tint the display slightly greenish in my opinion and in my phone. YMMV.
TLDR/ConclusionThe most efficient resolution is FHD when the phone is actively used. However, while not interacting with the screen, the resolution has almost no impact on battery consumption. Note: the resolution probably affects a lot when gaming is involved, which I have not tested. Between 120 Hz Adaptive and 60 Hz Standard, the current draw is about 15% to 17% higher for 120 Hz Adaptive Mode, depending on resolution. In other words, running the phone at 120 Hz does not exactly eat up the battery, which is very good to see!
A quick word about the current draw numbers: the current draw will be lower when the battery is charged more, since the battery voltage will be higher in that case. The lower the voltage, the greater current draw to keep power constant. At full charge, these numbers would be 88% of the ones I had at 30% battery charge. This should not change the relative numbers, i.e. FHD 120 Hz Adaptive Mode should still draw about 15% more current than FHD 60 Hz Standard Mode, I think.
What combo of settings do you recommend
The differences are quite small. Interesting.
10 or 15 ma extra is next to nothing.
da1e8 said:
What combo of settings do you recommend
Click to expand...
Click to collapse
I personally use 120 Hz Adaptive in FHD resolution. The display smoothness is too good to give up for a little bit of extra battery saving. With that said, I do use Power Saving Mode if I have to save battery during the day and that puts the phone in 60 Hz Standard refresh rate. I used to also switch to HD resolution in that case, but I will not do that any longer since FHD draws less current for some reason.
So my advice would be to stick to FHD resolution (I don't notice any difference between FHD and WQHD anyway) and 120 Hz Adaptive.
Curious to ask, is this an Exynos model or a Snapdragon model? Firmware version number?
zjhao said:
Curious to ask, is this an Exynos model or a Snapdragon model? Firmware version number?
Click to expand...
Click to collapse
Hey, good question! My bad, I should have written this down from the start, since it's quite relevant. I updated the original post.
The phone is an S21 Ultra Exynos model. The model number is SM-G998B/DS. The FW was G998BXXU2AUBB.
blackhawk said:
The differences are quite small. Interesting.
10 or 15 ma extra is next to nothing.
Click to expand...
Click to collapse
Isn't this for a test duration of 5mins. Please correct me if I'm wrong: if I take this calculation ahead..and assume that the test was carried out for a long period of time at 100% battery charge, then at 120Hz and FHD resolution - we should get a bit over 27hrs (5000mah/ 184ma) of SOT and WQHD resolution+120hz, it would be a bit over 24 hrs. In all, ceteris paribus, the difference is between 3 and 3.5 hours in all!
meh!
amirage said:
Isn't this for a test duration of 5mins. Please correct me if I'm wrong: if I take this calculation ahead..and assume that the test was carried out for a long period of time at 100% battery charge, then at 120Hz and FHD resolution - we should get a bit over 27hrs (5000mah/ 184ma) of SOT and WQHD resolution+120hz, it would be a bit over 24 hrs. In all, ceteris paribus, the difference is between 3 and 3.5 hours in all!
meh!
Click to expand...
Click to collapse
With a idle ma draw of 150-300 ma probably being typical, an extra 15-18 ma max for the screen resolution/refresh rate is chump change.
The lame Android scoped storage is likely using much more... and giving you nothing but trouble.
Scoped storage, it tastes like Apple
very helpful to me
blackhawk said:
With a idle ma draw of 150-300 ma probably being typical, an extra 15-18 ma max for the screen resolution/refresh rate is chump change.
The lame Android scoped storage is likely using much more... and giving you nothing but trouble.
Scoped storage, it tastes like Apple
Click to expand...
Click to collapse
My friend, you just went Christopher Nolan on me...didnt understand a word about the storage thingies.
amirage said:
My friend, you just went Christopher Nolan on me...didnt understand a word about the storage thingies.
Click to expand...
Click to collapse
Look it up and weep... it's one of the biggest reasons I'm still running on Pie, as are many others. Q's adaption rate has been dismal.
No coincidence.
PirateYarr said:
Here is a test I did on my S21 Ultra (Exynos), that started a bit spontaneously with me wondering about the impact on battery of different refresh rates and resolutions, together with use cases (touching the screen or just looking at it). So I started playing around and landed in some kind of improvised test.
The phone is an S21 Ultra Exynos model, as already mentioned. The model number is SM-G998B/DS. The FW was G998BXXU2AUBB.
The test was performed at 20% screen brightness and 30% battery left on the phone. The phone was charged between tests to hover around 30% battery, +/- 2%, i.e. it was between 28% and 32% during the tests.
There were two types of tests; "Screen Interaction Test" that intended to find the current consumption while repeatedly touching the screen and therefore keeping the CPU ready, and an "Idle Current Consumption Test" where the screen was just turned on for 3 minutes without interacting with the phone. In both tests, the phone was running the app Ampere in the foreground, thus displaying it.
The data logging was done manually by me watching the app Ampere and noting down its shown current consumption every second into a Google Sheets document. Ampere collects 50 data points, discards the 10 lowest and 10 highest, then takes the average of the remaining 30 points before presenting the current consumption. These values were gathered for 5 minutes in the interaction test, and 3 minutes in the idle test, in order to acquire a robust sample set that was resilient against temporary peaks in current consumption. Since I don't have control over the phone's background processes, it was possible that a background task would cause a sudden peak in current consumption. Averaging these values makes the measurement more resilient against such peaks.
The phone was put in Airplane Mode in order to eliminate Wifi, 4G, Bluetooth and all those things from the measurement.
Test ResultsInteracting With Screen in 120 Hz Adaptive Refresh Rate and Different ResolutionsThe screen was tapped 3-4 times per second in order to keep it active. The phone was displaying the app Ampere. Test duration was 5 minutes and data was logged about every second from Ampere, and then averaged over the entire test duration.
HD 120 Hz Adaptive Refresh Rate: 204 mA
FHD 120 Hz Adaptive Refresh Rate: 184 mA
WQHD 120 Hz Adaptive Refresh Rate: 206 mA
View attachment 5255207
Interacting With Screen in 60 Hz Adaptive Refresh Rate and Different ResolutionsThe screen was tapped 3-4 times per second in order to keep it active. The phone was displaying the app Ampere. Test duration was 5 minutes and data was logged about every second from Ampere, and then averaged over the entire test duration.
HD 60 Hz Standard Refresh Rate: 175 mA
FHD 60 Hz Standard Refresh Rate: 161 mA
WQHD 60 Hz Standard Refresh Rate: 176 mA
View attachment 5255211
Idling With Screen On in 60 Hz Standard Refresh Rate and Different ResolutionsThe screen was turned on it displayed the app Ampere. The screen was not interacted with at all for the whole 3 minutes this test took. The test was only done for 60 Hz because there is no point in doing it for 120 Hz since the screen drops down to 60 Hz when idling in the app Ampere. The values are the average value of the entire test duration (3 minutes).
HD 60 Hz Standard Refresh Rate, Screen Idling: 144 mA
FHD 60 Hz Standard Refresh Rate, Screen Idling: 145 mA
WQHD 60 Hz Standard Refresh Rate, Screen Idling: 147 mA
View attachment 5255213
Bonus Test: 48 Hz and 96 Hz Display ModeThe two hidden display modes were also tested but there were no advantages in using these modes regarding current draw. Plus, these modes actually tint the display slightly greenish in my opinion and in my phone. YMMV.
TLDR/ConclusionThe most efficient resolution is FHD when the phone is actively used. However, while not interacting with the screen, the resolution has almost no impact on battery consumption. Note: the resolution probably affects a lot when gaming is involved, which I have not tested. Between 120 Hz Adaptive and 60 Hz Standard, the current draw is about 15% to 17% higher for 120 Hz Adaptive Mode, depending on resolution. In other words, running the phone at 120 Hz does not exactly eat up the battery, which is very good to see!
A quick word about the current draw numbers: the current draw will be lower when the battery is charged more, since the battery voltage will be higher in that case. The lower the voltage, the greater current draw to keep power constant. At full charge, these numbers would be 88% of the ones I had at 30% battery charge. This should not change the relative numbers, i.e. FHD 120 Hz Adaptive Mode should still draw about 15% more current than FHD 60 Hz Standard Mode, I think.
Click to expand...
Click to collapse
Beautiful test. Well done and thanks.
I don't even own this phone, but this is a very useful test for other also! Kudos
Experiment well done bro! This community needs more people like you.
It's possible to overclock some monitors, and I have seen that it is possible to overclock some phone screens, so would it be possible to overclock the 4a 5g screen? Where would you even begin to do it? If it is possible to get it near 90hz would there be any long term damage?
deadlynoodle said:
It's possible to overclock some monitors, and I have seen that it is possible to overclock some phone screens, so would it be possible to overclock the 4a 5g screen? Where would you even begin to do it? If it is possible to get it near 90hz would there be any long term damage?
Click to expand...
Click to collapse
i was wondering the samething
Not an expert or anything but IIRC the Pixel 5 has a different screen type altogether which allows it to get to 90Hz compared to the P4a 5G
No idea.
Most phones use a MIPI-DSI-2 interface the display. MIPI DSI is a high speed low-voltage differential signaling (LVDS) connection serial interface. It allows the SoC to have a high bandwidth simplex (one way) communication to the display and write to registers within the display. The maximum refresh rate of a mobile display will be limited by one or more of the following:
* maximum throughput of the SoC
* maximum throughput of the display
* number of lanes used (1, 2, or 4).
* greatest common supported "mode" between SoC and display, and hardware connection.
You can easily calculate how much throughput you need to run a specific framerate and resolution, multiply the resolution, framerate ,and color depth (usually 8 bits per color channel, 3 color channels. 2560*1080*90*3 = about 0.75GB/s (gigabytes per second). as opposed to 500Mbps.
You will need to increase the number of lanes or the clock rate. If this phone happens to have all 4 pairs hooked up but is running only 2 of the pairs, and both the screen and SoC have a common supported resolution and framerate that is higher than what you got, then sure! But I really doubt this is the case. Running the clock a little bit faster might be possible at the cost of power consumption but it will more than likely result in bit errors, corruption of the display contents, crashing/hanging, who knows.