let me say that this is a great Kernel for a two day old thread. I can't wait to see what it looks like in a few weeks! Any chance of integrating [U] EDIT:
voltage & frequency scaling (user set) like incredikernel?
The devices are designed to run at certain voltages & at certain values for the various clock speeds. This was designed by an electrical engineer & while many times things work a bit differently in the real world than they do on paper, the base theory & science behind the optimum values that balance performance & power consumption are valid and often best.
This being the case, there are many times when hardware & to a lesser extent some software applications, will absolutely slow down, under-perform or even crash when the cpu or gpu speed values are increased & especially when they are maxed out.
When it comes to benchmarking, you must inherently take it with a large grain of salt. Rendering scores in particular can be affected by numerous things, such as rendering all via gpu, number of active threads, temperatures, & a host of other issues. You also need to keep in mind the shear number of hardware combinations is really staggering. If you think that a true benchmark of all of these different devices & their infinite hardware combinations can be accurately represented by a program that is a few mb in size (& that in some instances has absolutely no way of utilizing certain proprietary binaries & even some drivers)....well then I have some Ocean front property to sell you in Arizona. Samsung uses the same radio chip as HTC in many devices. However, Samsung devices are able to utilize the 5Ghz Wireless-N band while HTC doesn't. Why is this? Could it be drivers or proprietary code in a module? The same can be said for numerous OEM's & every piece of hardware. How on earth could a small program take into consideration these differences between all of the LG, HTC, Sony, Samsung, Motorola, Meziu, devices in combination with all of the hardware choices & OEM's?
Some of these programs are close enough when it comes to the mathematical functions (Calcs,\Ints\etc,) and some of the bench mark utilities ted even actual come close to using the standard accepted toolchain benchmarking methods as set forth by the Android Project Team along with Linaro (Ref. https://wiki.linaro.org/Platform/And...inBenchmarking
), but when it comes to rendering & the various methods used by the various manufacturers (many of whom use proprietary media binaries and/or drivers & sometimes even hardware as is the case with the "Media Processor" in the newest HTC devices) proper & accurate hardware benchmarking is just not accurate.
To properly benchmark, you need to perform testing numerous times & under numerous conditions, using the median or average values as the results. A single high or low score in a no-load situation is no indication of how a device truly performs in the real world.
In practical terms, many times the truest measure of a devices performance is in how you perceive it performing along with whatever included functions you find valuable. If you get a high benchmark using antutu or another general BM app yet your device takes 15 seconds to respond while you are utilizing navigation concurrently with media features or if wake up time is slow enough that you miss answering calls when device is sleeping, does that high benchmark accurately represent the devices true capabilities? You may get a decent scale of read/write operations, & some sense of processing power, but that's about it. You'd get the same sense though by closing the cached contacts application & then opening it, especially if like me you have a couple hundred contacts with addresses & complete contact info, or even by opening a folder within the directory with a lot of content in it. Do you know what sdk the BM app you are using was built for? What is it's targeted sdk? Is it hardware accelerated? Can it access those functions? Is it utilizing the same process and driver to render video as the device is by default?
Consider this, even when M$ rates its Windows 7 computers with a performance index number, they overall score is represented by the lowest value within any scored function set. This is because they understand that perception is an important factor & that even though you may have the fastest ssd memory & 8GB of RAM with your i7 or i9 processor, if the results of the processing take too long to be rendered due to sub-standard graphics processing hardware, the device will be perceived as being slow.
to buy me a cheeseburger if I helped you! Or even better, find a charity or local family fighting childhood cancer & help out however you can!
A brief & incomplete timeline of my love affair with mobile technology: