Which benchmarking software?

OK, so I'm a little (very) bored with TouchWiz and I fancy a change so I'm going to go back to the old rooting game but this time I'm curious; do the kernals actually increase the speed of the device, or is it just a smoother GUI?

Anyway, the closest I can get to quantifying this is going to be benchmarking, before and after. Any suggestions on what Benchmarking software I should use?

Quadrant/Antutu/Linpack are my current thoughts.

Add a comment

JanSt / MOD  Oct. 4, 2012 at 19:53

No idea :D
Where's matt when we need him?

jaybear88  Oct. 4, 2012 at 20:04

Haven't noticed him commenting on articles as much as he used to.... Last few I've read, I noticed that he hasn't found the current market too exciting, do you think he's slowly going to abandon us?

Pondlife  Oct. 4, 2012 at 21:39

Reading a comparison on razr m and i atm and they use quadrant and GLBenchmark which they say are the defacto standards..
Never found benchmarks that useful personally

jaybear88  Oct. 4, 2012 at 22:06

They are far from reliable/accurate, but it's about the closest we can get to actually measuring the performance differences unfortunately.

Cheers for the GLBenchmarking one there Pondlife, I think that might be EgyptOffscreen within that.

Started the job, rooted the phone and got distracted prior to kernal flash so it won't be done until (at least) after this weekends fishing :)

JanSt / MOD  Oct. 4, 2012 at 22:30

Tell us all about it once you're done :)

(I assume matt has real life affairs to tace care of?)

matt101101 / MOD  Oct. 4, 2012 at 22:33

Benchmarks are all utter crap for...well...everything they claim to do. I've had my S2 hit over 6000 in Quadrant before, I'm pretty sure that even with the CPU and GPU OC'd to within an inch of their phone a year and a half old phone can't actually outperform all the brand new quad-core devices (including its very own replacement; the S3).

Can the Note run the Siyah Kernel? That's what I use on my S2, it seems to be more stable and have better battery life than the stock CM10 kernel.

jaybear88  Oct. 5, 2012 at 00:38

I agree, not really my area but I'd assume that it has a load to do with background apps and their thread priorities etc. While I'm tempted to say that comparing the device to itself only (ignoring other devices) it might be a *little* more useful but again, it depends what I have installed pre and post flash.

However, if I want to look at literal difference ie how my device performs now vs later, forgetting about the particulars of hardware vs software increases. I feel that it may be of a little interest? If you disagree, please let me know as I'm really just bluffing my way through at this point and wouldn't mind a few more opinions!

matt101101 / MOD  Oct. 5, 2012 at 01:03

The thing is, comparing one device to another is just as pointless as comparing one device to itself. Developers manipulate the IO mark to massively up the overall quadrant score.
Benchmarks are pointless, just go with whatever feels best for you when you're using it (though remember to include battery life in your calculations).

CTPAHHIK  Oct. 5, 2012 at 09:21

BenchmarkPi is CPU integer.
Linpack is CPU floating point.
Both are for hardware performance, and I would not expect number to change unless you are overclocking.

You should stick with benchmark suite as they are better analyzing software performance. As you already mentioned Quadrant/Antutu should be good. Geekbench is crap.


You don't need an account to comment. Just enter your email address. We'll keep it private.