I barely understand what this means, however I have discovered that by going through tutorials, that one could do things like improve sd card performance, improve battery life, and many other things. Of course this interests me, but as I said I am new to android and I barely understand what this really means. I have hacked a lot of gaming consoles in the past and jailbroken many iphones, but that is about it. Definitely very new to all of this.
My device is the UMIDIGI A3X. Would it be possible to flash stock Android 10 onto this device (once I learn how to do this safely, still have to learn this) and then simply follow tutorials on kernel tweaks for the desired changes I want, without really truly understanding this kind of programming on a real level? In another words just going through the steps for the desired performance changes I seek.
Or am I way out of my depth here?
IMO SD-card performance basically has nothing to do with Android OS itself - but of course you can change relevant Android's cache size setting what probably doesn't result in any measureable results. Also you've to distinguish SD-card's IO read / write speed what depends on both the type of data bus the Android device supports and the SD-card's speed-class ( 2 - 10 ).
And battery life doesn't depend on Android OS: it solely depends on how the device - read: its CPU/GPU - is used/stressed and - most important - whether battery is always correctly charged ( never less than 30%, never more than 80% ).
Related
Lets all benchmark our XDA (MDA etc) . . . .
Then lets see how we can make them quicker!
Options:
- Registry Hacks?
- Overclocking Software?
- Etc . . . .
Surely somebody here has to be thinking along the same lines as me?
After all if you use something a lot, you want it better....
Can we clock this baby to 300Mhz + . . . .
MADmanFREE
:twisted:
I'm fairly certain anything over around 225-230 would exceed the refresh time for the memory and cause a hard reset. If it is even possible to manipulate the clock in the XDA by software.
Can someone explain
There is a large community devoted to overclocking. I have to ask why.
Is it really worth compromising the stability (and possibly integrity) of the hardware just for a few extra percent of speed?
These chips were designed to run at a certain speed (with a small margin to maintain stability). These chip designers know what they're on about, why do so many people think they know better?
I have to admit, i used to overclock my 486's, and it caused me so much trouble. Also a large high street retailer I worked for at the time used to overclock PC's and sell them as a higher spec - the amount of support calls we used to get!!!
By all means, if you want to try it, do so, but I have no problem with the speed of my phone.
Interesting so far.
Will have to get a poll going.
Just to see how much installing the extra software slows these things down...
MmF
JS's ipaqclock works, but i do not recommend oveclocking....
Our device isn't considered to be powerful anymore. It's became an complete underdog. Thanks to Android's flexibility, you can force to make your phone pretty much fly with several tweaks.
1. Delete as much bloatware as possible.
This is regarded as an easiest, noobiest option to improve your phone's performance. Many OEM apps stays hibernated in the RAM, thus limiting free RAM available to the user and sometimes even stressing NAND too much. In the end, it's the best to keep your phone as clean and deleting useless apps on the fly.
2. Don't install too much apps into NAND memory
Benchmarks shown, that our phone's NAND chip is considered to be rather slow. Since many apps access small amounts of random data pretty much constantly, it's better to have less apps installed on your phone.
3. Avoid using app killers
This is a placebo effect, since Android's memory management is inferior to what app killers are doing. Many apps stays in such state, where they are ready to be launched almost instantly. App killers, however, pretty much screw all the mandatory functions and stresses the phone even more than before. Launcher redraws are rather common occasion when app killer is being used.
4. Use custom kernel
Custom kernels often offer more features and are more optimized to make the use of available hardware. OEMs never seem to mess around with kernels much, since they want to have their product as stable as possible. Devs, however, mess around with kernels and extract almost double the real-time performance.
5. Never fill up your storage completely
The more data is available on the storage, the harder is getting to find it. Since data is laid randomly, it searches for the information location. When there are too much data, it gets harder to find the data needed. Often slower cards, like Class 2 or Class 4, are considered to be the better choice, since those cards are much faster at writing and reading marginal data randomly.
6. Select the I/O scheduler, CPU governor wisely
These things manipulate with the main hardware. The better optimized the governors are, the better the phone will run and won't drain the battery as much. Though keep in mind, that many governors have their own drawbacks.
sioplus is one of the better I/O schedulers. It allows access to random data pretty quickly, which ensures smooth and snappy performance in the system.
ondemand is the most common and is the stapple and the base of many custom governors available today. It's method is pretty simple - whenever phone registers a touch input, it automatically raises the CPU speed to the max. In retrospect, it should give great performance, but it usually suffers from poor response.
7. Play around with Dalvik VM settings
My optimized settings (feel free to use them):
dalvik.vm.heapstartsize=6m (size when first launched)
dalvik.vm.heapgrowthlimit=64m (limit of standard app)
dalvik.vm.heapsize=192m (heap for large app)
These settings pretty much controls our multitasking. Each phone has it's own specified settings, so it could run better.
Lowering these settings could majorly improve performance, but it could slow down around, when there too much heavy apps running in the background.
Raising these settings could improve multitasking, since less CPU power is required to extract certain data to the RAM. Scrolling a heavy webpage, for instance - raising these settings could improve scrolling smoothness and loading times, since there isn't a need to clean the heapsize as frequently as it was before.
More suggestions are coming later. If you found this article useful, please leave THANKS!
Good day.
Hi, I'm looking for a very low resource way of measuring power draw over a time interval. I'm looking to measure the power draw as various pictures and videos are displayed on the screen to get an idea of how much power the various components of the phone draw. I've done some looking and I've found this: http://developer.android.com/reference/android/os/BatteryManager.html. Particularly, I noticed Charge_Counter, Current_Average, and Energy_Counter. My issue is that while I have a bit of a programming background (more like a hobby), I have no idea where to start with this. I have some background in Python, Labview, linux scripting, and very little C and .NET, so I understand some of the basic concepts of programming, but anything beyond that is something I would have to pick up. I have virtually no experience in Android programming (other than a few Cyanogenmod compiles from source - as anyone who has been around long enough has ).
I wouldn't need a gui for this, so something like a script would work great. It can (maybe should) be able to be executed via ADB, and needs to be universal - working across modern devices. The app/script I'm envisioning would perhaps poll the battery for a power level, wait a specified time - 2 min, 5 min, 10 min, 1 hour, something like that - then poll it a second time, get the delta, and divide by time to get an average power. Other than those two polls, I would want the app/script to have no interactions with the processor. The greater the accuracy, the better, and if no root is possible, that would be my choice. I would prefer to keep the phones as stock as possible.
My questions are:
- Is something absolutely universal (or mostly universal among newer devices - say 1 yr old or less) possible (no root)?
- Since it needs to be universal, would Java be a good choice of programming language? Or perhaps this can be accessed via the command line (script)?
- How accurate would a method like this be? How do the phones know the instantaneous current, power level, etc? Is it basically a shunt resistor, measuring the V across?
- How much time would it take (generally) for a noob with no Android experience to get a working demo? Days, weeks, months?
Thank you for your time.
I would also mention, this doesn't have to be "polished" for an audience...just simply looking for a down and dirty method that works - mostly for myself.
Hey guys
I'm recompiling a kernel for my Galaxy S5. I'm disabling some parts of kernel code, trying different toolchains and compile options
But, how do I know if any of if is being effective in getting more performance out from my phone?
I know there are many benchmark apps out there, but I've tested ~10 and they all give inconsistent results. I mean, they all look pretty and stylish, but who can trust in a 1-minute test, with varying results even when not changing anything on the phone?
Don't tell me benchmarks are BS, because you have decent benchmarks for Databases (I work with some of them), Desktop PCs (even Windows had a built-in one), memory, disks, TVs, etc... I'm used to test and compare "things" forever, but I'm used to solid and consistent results. Sadly in Android world people give too much attention to pretty graphics or hearty words like "Your phone is fantastic" (this is the type of "conclusion" some benchmarks apps out there gives... LOL)
So, how to test a Kernel's performance in Android? Do I have to learn Java and create my own...? I hate Java!
Hey there, android enthusiasts!
Hopefully this post will gather many experienced users and solve these problems and doubts once and for all.
I've searched pretty much everywhere on google and the usual material showed up, copied and pasted around the web.
So I'm posting here, hoping that developers and hackers come and share their precious knowledge, producing new material that will put light in this often misunderstood field.
1) What does vfs_cache_pressure do? How does it interact with RAM and Swappiness?
2) What happens when I have Z-Ram enabled and Swappiness set at 0?
3) Are the init.d tweaks by franco's developers team still good for Nougat and Oreo? I understand these guys know what they do and put a good amount of effort in finding the best settings, however maybe they were wrong, or they intended to engineer them for a different kind of usage (I'm thinking about the perennial debate between Performance, Balanced, and Multitasking).
The VM tweak sets Swappiness at 0. Is this stupid if I have z-ram on because that z-ram will never be used?
4) That also tweaks vfs_cache_pressure at 100. Why not using lower values of vfs_cache_pressure like 50, or even 30 or 10? These all raise perceived performance and overall fluidity, at the expense of real performance. Even though this seems counterintuitive, one only needs to try these settings to be convinced of their value in overall fluidity and smoothness (to know more about that, see Tales from Responsivenessland by Rudd-o, and also Imoseyon's Sysctl tweaking for faster, longer lasting Android)
5) What about dirty_ratio and dirty_background_ratio? Why values of 20 and 5 respectively? What happens if we set them at, say, 90 and 55? How do they interact with swappiness and vfs_cache_pressure? I remember reading somewhere that setting them high would delay writes and save battery, but at what price?
6) How do dirty_background_ratio, dirty_ratio, swappiness and vfs_cache_pressure all interact with low memory killer values?
7) Not related: What does the Arch Power option do, exactly? Why is it good/bad for x or y?
8) What is RQ Affinity and how should I set it and why?
No one?
I've been looking for answers to some of these same questions. I know Google is my friend but when i search for what feels like forever and only find more questions, it gets frustrating. So like the OP said.....NO ONE?
Shameless bumb, we need answers
Nolfer said:
Shameless bumb, we need answers
Click to expand...
Click to collapse
Sorry for the necro.. but in May 2019 I wrote this and this on Telegra.ph in this subject, hope it still helps and you are still interested.