Database Users

[Q] What affect the points of benchmark?

I'm using MiniCM RC1 rom with OC.But it doesn't have any different between OC/non-OC
I just want to ask what would affect the score in benchmark and could you tell about your marks?
ps:English isn't my mother language I wish you could understand what I'm talking about

samyeung97hk said:
I'm using MiniCM RC1 rom with OC.But it doesn't have any different between OC/non-OC
I just want to ask what would affect the score in benchmark and could you tell about your marks?
ps:English isn't my mother language I wish you could understand what I'm talking about
Click to expand...
Click to collapse
I have 869 as my lowest score and 1069 as my highest. I'm using Racht's Floyo v0.15. Also, If you overclock only ur Cpu scores increase. At the bottom you should see your CPU score.

samyeung97hk said:
I'm using MiniCM RC1 rom with OC.But it doesn't have any different between OC/non-OC
I just want to ask what would affect the score in benchmark and could you tell about your marks?
ps:English isn't my mother language I wish you could understand what I'm talking about
Click to expand...
Click to collapse
The current OC avail for the X8 is for the CPU only. However, you might see an increase in the performance of mem, 2D and 3D performance.
Check out the CPU scores on quadrant, since this will be most affected.
Factors that affect benchmarks: in quadrant, theres a hack which increases scores by fooling quadrant to think a test was completed but actaully not done. this is the default of minicm and most HTC ROMS. Memory and Cache settings also affect the benchmark scores.
Right now, my X8 on RC1 MiniCM with 748 OC has a constant 1200+ quadrant score with Stagefright true and 550 with Stagefright false. 45+ FPS in Neocore. 1020+ on Antutu System Benchmark.

Running services can slow down your phone and lower your score.

Accurate benchmarking.

Quadrant is very inaccurate the tests are not up to spec for testing +1Ghz devices accurately.
The reason i say this is due to the fact that the benchmark varies from test to test.
I suggest AnTutu Benchmark.
https://market.android.com/details?id=com.antutu.ABenchMark
It is much more accurate, as the the scores are closely similar time and time again.
The tests are a bit longer
But that is the good thing "accuracy"
It also gives info on each testing stage.
I got 2176 on WB's 007.1 TW rom, BB.67 with custom kernel.

OmegaRED^ said:
Quadrant is very inaccurate the tests are not up to spec for testing +1Ghz devices accurately.
The reason i say this is due to the fact that the benchmark varies from test to test.
I suggest AnTutu Benchmark.
https://market.android.com/details?id=com.antutu.ABenchMark
It is much more accurate, as the the scores are closely similar time and time again.
The tests are a bit longer
But that is the good thing "accuracy"
It also gives info on each testing stage.
I got 2176 on WB's 007.1 TW rom, BB.67 with custom kernel.
Click to expand...
Click to collapse
Good, will give it a try but it'd been better if you post this in themes & apps I guess
by the way what scores do you get with this rom in quadrant?

Chainfire released his new benchmarking tool yesterday, its called CF-Bench, search in the market

i've got 1973 on CM6 006 at 1.1Ghz

benchmarking
It does not matter what benchmarking utility you use. They all give you a general idea of the performance of your phone. I think the OP just wants to use the tool that gives him the score he wants to see.
In reality the score of any phone using any utility is always relative to the scores of other phones using the same utility.
The thing to focus on is not the actual figures you get back but how your device measures up to other phones using the same benchmark test.

kryptoner said:
Chainfire released his new benchmarking tool yesterday, its called CF-Bench, search in the market
Click to expand...
Click to collapse
2232 hmm right behind the nexus s
Cm6 wb bb67 custom kernel z
Sent from my X10i using XDA App

Try smartbench 2011. That's the best one
Sent from my x10 on gingerbread

All benchmarking apps generaly suck. Its about how you feel using your phone. Not the numbers.
Sent from my X10i using XDA App

pepy_24 said:
All benchmarking apps generaly suck. Its about how you feel using your phone. Not the numbers.
Sent from my X10i using XDA App
Click to expand...
Click to collapse
But it's a good comparison between phones after having a average score for each phone. But like you say, the numbers don't mean anything. Everything is personal.

zodiac100 said:
It does not matter what benchmarking utility you use. They all give you a general idea of the performance of your phone. I think the OP just wants to use the tool that gives him the score he wants to see.
In reality the score of any phone using any utility is always relative to the scores of other phones using the same utility.
The thing to focus on is not the actual figures you get back but how your device measures up to other phones using the same benchmark test.
Click to expand...
Click to collapse
True about benchmark apps but I don't think I think the OP just wants to use the tool that gives him the score he wants to see. What the OP is saying that with this app the difference between each time's score is not much hence why calling it accurate.

To look for a good benchmark is more important if you use a new phone with dual core CPU cos the old benchmarks aren´t optimised for them. I read that CF-Bench is optimised for dual cores and stresses the CPU and not so much the GPU of the phone.
But interesting that it seems that the overclocking have such a huge influence. With the first run (Wolfbreaks CM6 TW007) i had 2800 points with 1152MHz and in the second run with 998MHz 2400 points.
But true....it´s more a interesting gadget than to take it too seriously

maddes1402 said:
To look for a good benchmark is more important if you use a new phone with dual core CPU cos the old benchmarks aren´t optimised for them. I read that CF-Bench is optimised for dual cores and stresses the CPU and not so much the GPU of the phone.
But interesting that it seems that the overclocking have such a huge influence. With the first run (Wolfbreaks CM6 TW007) i had 2800 points with 1152MHz and in the second run with 998MHz 2400 points.
But true....it´s more a interesting gadget than to take it too seriously
Click to expand...
Click to collapse
It comes down to what you have happening in the background.
If your phone suddenly "for example" decides to fetch data off of 3G or suddenly refreshes the media midway thru the bench then it can effect your score. this is why it's best to do tests when the rom is clean and gone silent. "no background activity" with data disabled.
Benchmarking is used to place the current performance of a device in a certain category.
It's no different from benchmarking PC's.
A whole profession has been built around PC benchmarking with people competing for a first place.
If your score is higher than everyone else you win the competition.

i got 2232 on custom kernel at 1113 mhz processor
Sent from my X10 TripNMiUI-IRIS

CF-Bench
Smartbench 2011
Quadrant
Antutu Benchmark
what is the best now?

Wolfbreak said:
CF-Bench
Smartbench 2011
Quadrant
Antutu Benchmark
what is the best now?
Click to expand...
Click to collapse
CF Bench (full version) has the most stable results, and IMO its the best out there.
Sent from my Sony Ericsson Xperia™ phone

Linpack is perfect

kantk20111 said:
Linpack is perfect
Click to expand...
Click to collapse
Linpack only tests the cpu.
I haven't checked how CF bench works yet.
will report.

Wolfbreak said:
CF-Bench
Smartbench 2011
Quadrant
Antutu Benchmark
what is the best now?
Click to expand...
Click to collapse
CF bench 101%

Wolfbreak said:
CF-Bench
Smartbench 2011
Quadrant
Antutu Benchmark
what is the best now?
Click to expand...
Click to collapse
Chainfire one!
that thing is scale-able across cores (single/multi-cores) and fairly accurate

[Q] Quadrant low score

Doing the complete benchmark i only get 700, with FroyoBread 21, almost all apps closed, and no undervolt. Is this normal?

I got only 768, almost everything closed/turned off. I haven't tried quadrant since v.12, but i used to have about 1100. Don't know why.
edit: With 3G off I get 893, but still...

With data, localization, 3g off i get 800..

With wifi on ~980 with undervolt, smatass and multitouch (without OC) froyobread v021b

I now got 1038 with very good 2d and 3d results but sometimes my phone became slow and low ram and system fc all time only reboot can fix this.

With Froyobread 12 i got a quadrant score about 1100, but with 21b i got only 780~...
Any Ideas ??

Why do you guys care about quadrant scores.
What really matters it's ROM speed
but... if you want very high scores, click here

D4rKn3sSyS said:
Why do you guys care about quadrant scores.
What really matters it's ROM speed
but... if you want very high scores, click here
Click to expand...
Click to collapse
Awesome, 1594!!
Not real but funny xD
thanx

Well I've tried xGin 5.4 after doing a full wipe and i got about 800-900, so maybe the "problem" isn't in our ROMs, maybe there are just updated statistics of quadrant, because I haven't noticed any slowdown of my phone.

1150 with gingercruzt

Actually I found that if you are using Standard's Quadrant,the score will lower than Advacned's Quadrant.
For example,I'm using Floyo v1.0,the score are:
Standard:~850
Advanced:~1100

[Q] really slow a500/ quadrant score 656???

Hi,
as i'm using my iconia a500 i noticed some heavy Performance issues. i rooted and installed several Roms but my Quadrant score keeps being low. right now i am on lightspeed 1.0.1 with Richard's Kernel and my Quadrant score is only 656!!!
i'm really curious about this any suggestions?
thanks in advance
kiark

Just a guess, but the cpu probably is not running full during the test. Might have to use a app to force to 1000 during the test.

Running 2.4 kernel? If so, the default intersctive scaling governor is borked so either use setcpu to change to on demand or flash back to 2.3. Im personally running 1.6ghz as max speed with on demand.

thank u very much, setting the cpu up with setcpu worked!

CPU really does not have much effect on quadrant score. It is the combination of the Rom and the kernel it self.
Sent from my SGS-t959 using XDA Premium App

IM using the xoom 3.2 rom
My acer is over clocked to 1.5 and here are my benchmark scores.
Useing Quadrant I got 3421 (best of three runs) and frame rates around 45
Useing Linpack I got 45.438 mflpos on single thread in 1.85 seconds and
71.53 mflops on multi-thread in 2.36 seconds again best of three runs
Useing AnTuTu Benchmark I got a score of 6159 and frame rates right at around 60 fps
Useing Vellamo (its new and a benchmark tool for a tablets browser to see how ell the browser performs does things like java and flash stress tests and load speeds of web pages) I got a score of 1072

naturallight said:
CPU really does not have much effect on quadrant score. It is the combination of the Rom and the kernel it self.
Sent from my SGS-t959 using XDA Premium App
Click to expand...
Click to collapse
not sure on this. if it doesnt have much effect why do you see such a huge difference between 216 Mhz and 1.5 Ghz clock speeds in quadrant?

Underclocking CPU Gives Better Benchmark Scores?

Hi,
I don't know if these benchmark applications are valid, but I downloaded AnTuTu Benchmark and messed around with my CPU frequency. At first, I maxed out the core clock speed to 1.3 GHz. Next, I set the maximum clock speed to 475 MHz. When I set it to 475 MHz, I get a better CPU score.
Something does not seem right. Any thoughts?

...and now you know why folks don't trust benchmarks.

There is a thermal CPU control. By overclocking, you are overheating, and it's underclocking to prevent itself overheating...

CrazyPeter said:
There is a thermal CPU control. By overclocking, you are overheating, and it's underclocking to prevent itself overheating...
Click to expand...
Click to collapse
That was funny to read.
By overclocking, your in fact under clocking or so it seems.

Yes that can happen. If it gets too hot it will probably start throttling each core back which gives worse and worse scores.
Right now I have mine up at 1600Mhz and I get better scores than 1300Mhz but as well I'm running nearly the same voltages that 1300Mhz is.

Nice!

Interesting guys,
I guess I'll just leave it at default settings

Too bad I can't root my device .......

STXInnovation said:
Interesting guys,
I guess I'll just leave it at default settings
Click to expand...
Click to collapse
I made a bunch of (repeated) measurements with a CM10.1 nightly kernel (at various Fmax settings) and the results were quite squirrelly. When I get a little time, I'll make some more tests and post up the data.
BTW - if you were using TricksterMod are you sure you actually applied the changes? You didn't provide any numbers (your results) and it is not uncommon for Antutu to produce 10% different results in that benchmark from run to run (and much much worse at different Fmax values!!) - so is it possible you just mistook a random variation for a "change for the worse"?

bftb0 said:
I made a bunch of (repeated) measurements with a CM10.1 nightly kernel (at various Fmax settings) and the results were quite squirrelly. When I get a little time, I'll make some more tests and post up the data.
BTW - if you were using TricksterMod are you sure you actually applied the changes? You didn't provide any numbers (your results) and it is not uncommon for Antutu to produce 10% different results in that benchmark from run to run (and much much worse at different Fmax values!!) - so is it possible you just mistook a random variation for a "change for the worse"?
Click to expand...
Click to collapse
Yeah would have to agree here. Between voltage and clocks you can really affect any benchmark score. I knits I've been playing with it. Though a swing of 10% in AuTutu is pretty big. I've had maybe a few hundred points at most.
Sent from my Nexus 7 using xda app-developers app