Is geekbench really that truthful? I got 1.5x the "supposed" average? speed of oneplus x and it doesn't make sense. Also in addition, does physical acceleration of CPU have an impact to the performance of it? (FTL acceleration and electron computing)
is it open source?
can you validate those results in some way?
if no then fuck off
>>56406166
I validated over and over again and still got the same results (2300 to 2500+) even with the old lollipop OOS ROM.
>>56406197
Running the same test again is not validation.
The only results worth looking at are FP performance, everything else is pointless data.
Currently the Exynos 8890 has the multi-core performance of a haswell i3 @ 2.4 GHz. However it's not know for how long such performance can be sutained in a task like video encoding and 4 ARM mongoose cores at 2.3 GHz have to be used (single thread FP performance is shit).
>>56406484
You again? Fuck off, the floating point benchmarks in geekbench are utterly, utterly useless. Go compare them to an ACTUAL floating point benchmark, like linpack.
>>56406502
>You again? Fuck off, the floating point benchmarks in geekbench are utterly, utterly useless. Go compare them to an ACTUAL floating point benchmark, like linpack.
Fuck off you lincuck. Nobody gives a shit about your gentoo meme benchmarks. The FP benchmark in geekbench is just fine and gives you a good idea at how shit or non shit your phone's CPU is.
>>56406569
>Fuck off you lincuck. Nobody gives a shit about your gentoo meme benchmarks
You have no idea what linpack is, do you?
>The FP benchmark in geekbench is just fine and gives you a good idea at how shit or non shit your phone's CPU is.
Yes, it's perfectly adequate for comparing ARM CPUs to other ARM CPUs. It is absolutely god awful for comparing across architectures and operating systems.
>>56406719
>Yes, it's perfectly adequate for comparing ARM CPUs to other ARM CPUs. It is absolutely god awful for comparing across architectures and operating systems.
Literally how? Both ARM and x86 processors do floating point math, geekbench scores on how well they do for the duration of the test, how is it bad for comparing across architectures?