Whats a good app to use to benchmark the performance of a given ROM? I want to start keeping track of benchmark scores between all the ROMs that I flash. I've found 3 that I've seen used around here.
1) BenchmarkPI
2) Benchmark by softweg
3) Linpack
There is no ONE good one. Just keep track of all three scores and compare, because they measure different things.
There is one you forgot, the benchmark in setCPU.
Sent from my GT-I9000 using XDA App
They're pretty reliable I guess, they just can't be translated to real life performance
http://briefmobile.com/cyanogen-demonstrates-quadrants-flaws
Not so, with GB and lag fix kernel and modified prop I had over 2300, but rest of the UX was bad even games were jumpy, now i have less than 1000 with Quadrant and butter smooth UX (user experrience) and better and more stable framerate in games than ever
Let's say on PC you have great score with PC Mark and 3D Mark but you can't play 3D Tetris...
Quadrant Advanced is better in terms that it gives you separate score chart for different parts of test.
Its a benchmark that does not reflect the usability or response of the SGS .
Highest score on Eclair then with each upgrade score gets lower and phone gets faster more responsive and less laggy .
jje
Quadrant results are useful if the ONLY thing you want to run on your phone is quadrant, because that's what it tests. It's similar to how on PC's, some games prefer different video cards. It also comes down to space/speed of algorithms..
The best thing to do, is to uninstall Quadrant, and test the ROM's yourself
my quadrant results are in the 4k while my wife's S2 is in the same range (slightly higher) both without any overclocking
I was happy with 4k but now I wonder why my wife's s2 can achieve similar numbers with no overclocking.
My GN overclocked at 1.7 with performance gov hardly go over 5k
I think I read some quadrant are in the 7k
what's yours ?
guessing here but since the cpu/chipset are the same and the note is a little slower in benchmarks, I'm gonna say its the bigger screen resolution that it has to power, more pixels to push.
Bingo.
I got around 4200. Fully stock.
Sent from my GT-N7000 using xda premium
I got 5800 quadrant score overclocked at 1.7
and scored 5005 at default 1.4 Ghz of frequency
I am happy with it...
Thanks guys for the input I feel better now
ArcticCat said:
Thanks guys for the input I feel better now
Click to expand...
Click to collapse
well you just made me feel worse, im getting 3900
Quadrant score fluctuates a lot, so I don't know if it is reliable. I had about 5k and the next benchmark I had over 6k, so.... ofcourse, i had it set to 1.7ghz.
Usually around 4000.
1st 3800
2nd 4000
3rd 4200
4th 4600
5th 4700
6th 4600
Not sure which result is accurate even knowing that quadrant means nothing.
Quadrant is such a badly flawed benchmark, it can't even be used to compare the real world performance between two ROMs, let alone two different devices!
Regards,
Dave
Sent from my GT-N7000 using Tapatalk
foxmeister said:
Quadrant is such a badly flawed benchmark, it can't even be used to compare the real world performance between two ROMs, let alone two different devices!
Regards,
Dave
Sent from my GT-N7000 using Tapatalk
Click to expand...
Click to collapse
Your proof?
WE hear it all the time but nobody ever explains it.
1.4ghz Non-Overclocked, but massively tweaked and I get:
1st Score: 5900
2nd Score: 6200
3rd Score: 6200
4th Score: 6000
5th Score: 6300
xAnimal5 said:
1.4ghz Non-Overclocked, but massively tweaked and I get:
1st Score: 5900
2nd Score: 6200
3rd Score: 6200
4th Score: 6000
5th Score: 6300
Click to expand...
Click to collapse
wow that's impressive!
What kind of tweaks are you taking about?
ArcticCat said:
wow that's impressive!
What kind of tweaks are you taking about?
Click to expand...
Click to collapse
Deleted pretty much everything you can to create a stable system
Adjusted the voltages based on my use
Added some build.prop tweaks
Tweaked the system UI for less RAM usage
No background widgets/No background activity
Double checked an triple checked all apps to make sure no mail process is in use
RAM Optmization tweaks
A bunch of other tweaks that I can't recall off the top of my head... haha
My Score ...
this is my score with quadrant !!
seepage said:
Your proof?
WE hear it all the time but nobody ever explains it.
Click to expand...
Click to collapse
My personal proof is my own personal experiences, but you only need to Google terms like "quadrant benchmark flawed" to find lots of examples.
I've owned a great number of Android devices, and seen widely varying Quadrant scores, yet not one of them has ever indicated the comparative real world experience between the devices.
For example, my Galaxy Tab 7" would only score around 800 (stock) whereas my Note gets over 3600. Is the Note 4.5 times faster in real world usage - not even close! It is faster, but I would subjectively place the performance at around 1.4x to 1.5x.
In a similar vein, when I changed my Galaxy Tab 7" to use EXT4 instead of RFS, but Quadrant score shot up to nearly 2000, but was it over twice as fast as stock? No, once again not even close! However, it still felt faster in real world usage than my HTC Sensation which gets around 2500 on Quadrant.
Quadrant basically performs a lot of individual benchmarks on a device and then amalgamates the result into a single number. Those individual benchmarks may well be of use, but the ultimate number is derives clearly "weights" each benchmark inconsistently/incorrectly to be of any real use when comparing real world performance.
Ultimately, what do these numbers mean if they can't give you an idea of comparative performance?
I find it quite amusing the number of people on this site who chase high Quadrant scores. It's like comparing an M-series BMW with a regular model which has a M-series bodykit and then saying how they perform based purely on looks.
Regards,
Dave
Phandroid (or AndroidCentral) had an article about the crapyness of Quadrant some time ago.
And actually, a lot of mobile review sites refuses to run Quadrant anymore.
Sent from my GT-N7000 using xda premium
Hi, i went down quite a lot of pages and did not find any antutu score tests on galaxy s.
I was just wondering whats the best ones in what rom's and what kernels. I tried to google i9000 antutu best scores but seems there was not really that many hits and best scores i found was like 4k what sound quite a lot but not sure how legit those are when there was no pics.
I mean my friend got a galaxy note and it get a bit over 7k point and then reading more the damn old sucker galaxy s2 best with pic proof is like 8k.
Was just wondering how my fine old faithful galaxy s can do. I got a decent gingerbread rom and i get a bit over 3k without OC. Can't remember how much more i got with ICS OC'd to 1,4ghz. it was over 3k too tough but not remembering how much above.
Would be nice if you can post your roms that score over 3k and what roms + kernels you are using and how much OC you done. I mean the i9000 is not the fastest phone by far at this point but would be intressed to see if people got how high antutu scores.
Nowadays benchmark doesnt mean anything. There's many ways to cheat, and even if those score were honest it doesn't matter much. What does matter is real life performance
Oh antutu score can be fixed too these days? I mean i now the quadrant score has been always a bit question mark and i heard lots of cheating in it. Well i was just wondering the honest score. I'm trying to find a decent fast rom and the better score in benchmarks the less processor power it uses on stuff and that means a bit better battery life too.. well not always.
I'm just trying to get all out of my sgs before sgs3 comes out.. i must say i just love t his phone.. when i had a fast rom where it worked stable at 1.4ghz it was as fast as iphone 4s in same internet pages on stock browser.
But if people could post the LEGIT non cheated scores it would be better than the cheating ones lol
an0nym0us_ said:
Nowadays benchmark doesnt mean anything. There's many ways to cheat, and even if those score were honest it doesn't matter much. What does matter is real life performance
Click to expand...
Click to collapse
+1. SGS with Froyo and EXT2 lagfix got a much higher score in Quadrant, but with ICS the experience is much better.
So the real question is what ROMs have given the best real life performance for you?
For me, it's 3827( on Antutu ) ;D with TiramiSlim RC2.0
I looked a bit onto numbers posted on One benchmark post and one thing bothers me, specifically with CF bench:
Native scores are not really that bigger - around 10% higher than let's say S3 (26728 vs 23966). However, java scores are significantly better and are even nearly on par with native scores (and around 3.5x higher than current devices - 20119 vs 6489 for S3).
If I understand it properly, native benchmark is done natively with compiled C application, while java benchmark is running in Dalvik.
Considering that java scores are really ones that got so improved, is there a chance that HTC One contains some huge Dalvik optimizations?
matejdro said:
I looked a bit onto numbers posted on One benchmark post and one thing bothers me, specifically with CF bench:
Native scores are not really that bigger - around 10% higher than let's say S3 (26728 vs 23966). However, java scores are significantly better and are even nearly on par with native scores (and around 3.5x higher than current devices - 20119 vs 6489 for S3).
If I understand it properly, native benchmark is done natively with compiled C application, while java benchmark is running in Dalvik.
Considering that java scores are really ones that got so improved, is there a chance that HTC One contains some huge Dalvik optimizations?
Click to expand...
Click to collapse
I think with Android 4.4 you should have even better performance... thanks to the ART engine.. maybe with following updates you'll score even higher