[Q] Memory management in ICS ROMs - HD2 Android Q&A, Help & Troubleshooting and Genera

I have a question mainly for developpers.
Playing around with severals ICS CM9 ROMS, I noticed big differences in Antutu RAM benchmark score.
A first family of ROM reach 330 in RAM score (CM9 nightlies, ...)
A second family of ROM reach 530 in RAM score (CM9ight,iCyanogen).
This is a huge difference.
I've done some tests, and the difference don't come from kernel, from build.prop, from framework.
So the question is : from what does it come ?... some libs ?...
Thank you to developpers that read me ... any ideas ???

As a user and not developer, there is no problem with consuming more RAM.... as long as the Rom... is ruining smooth and fast...
Jelly bean uses even more RAM than ICS
Sent from my HTC HD2 using Tapatalk 2

Related

[Q] What is the best android rom for games ??

hi everybody , i want to make a question how i tiered to having the answer !!
i want to have a answer for my question !! What is the best android rom (nand or sd ) for games ( like pes 2011 , asphalt 5 , .... ) !!
thanks !!
Any non sense build. Hyper, ultimate, cyanogen, nexus, etc.
yeaah ! but i like sense build like desire z or rafdroid !! thanks for your answer !!
That's the thing. Sense builds take up a lot of ram, including desire hd and 7 builds. They leave you with 180 mb max. But that's only in the aspect of ram. In terms of processing/gpu, they should all perform the same.
ooh ! but with a2sd they will work with my partiton of ext3 !! i will chose my space !! thanks a lot !!
I'm talking about RAM not ROM. RAM is the memory for which apps/software is currently running on the device is placed. ROM the memory for anything installed. E.g. the android rom, bootloader, HSPL, apps, etc.
yeaah ! i agree with you !! but desire z rom such been powerfull !! his quadrant score up to 2800 !!
-.-"
well I guess beliefs and opinions are one thing... and that speed can be considered relative. But the fact is non sense build are immensely faster than sense builds and that quadrant scores, like in my siggy, mean nothing.
Proof? I'm using a captivate right now running cog 4.3, and it gets quadrant scores like 1500 and lower, but it runs buttery smooth, better than even the non sense builds from the hd2, in fact it will even run 720p youtube videos, whilst the last time I tried with my hd2, it wouldn't even load. But anyways you want sense so better you stick with it and you're under an opinion I believe you likely will stick to, due to it being well, your god forgiven opinion, so my talking means nothing.
hh ! i'm Noob ! i proov it ! and i'm sorry !! thanks for your comments and your ideas !! !!
I'm sorry if I came off as rude. And i'll admit it was rude of me to say "god forgiven". It's simply quadrant scores that ticked me off, not you. Hence why my siggy is dedicated to the foolishness of quadrant scores.
But anyways, apart from non sense builds, try a simple desire build. e.g frank's/prj. They're faster than desire hd/z builds, but marginally slower than non sense builds. They offer 210 mb ram at max.
!! thanks you so much !! i will test this roms !!

Back to Gingerbread

I've noticed a lot of talk about Jelly Bean and ICS being slow on the Nexus S. So, has anyone reverted back to Gingerbread? It's been so long, I don't even remember what GB is like.
Sent from my Nexus S using Tapatalk 2
Ulver said:
I've noticed a lot of talk about Jelly Bean and ICS being slow on the Nexus S. So, has anyone reverted back to Gingerbread? It's been so long, I don't even remember what GB is like.
Sent from my Nexus S using Tapatalk 2
Click to expand...
Click to collapse
Yep, I use Bugeless GPA19, stock kernel is OK, but if you want 30mb more of free RAM use Glados 1.12.
Nope. If you keep your running apps under control I believe Jellybean is just right on the Nexus S. =)
Ulver said:
I've noticed a lot of talk about Jelly Bean and ICS being slow on the Nexus S. So, has anyone reverted back to Gingerbread? It's been so long, I don't even remember what GB is like.
Sent from my Nexus S using Tapatalk 2
Click to expand...
Click to collapse
2 days ago i tried GB again, its not that much faster...
The only thing faster in gingerbread was installing apps and opening up search.
Jelly bean has vsync so you dont get screen tearing which was everywhere in GB, apps open up quicker in JB, the camera is much faster, the browser is much smoother and last but not least the stock launcher never crawls to a halt like it sometimes used to in GB and had to be restarted. I am not talking about launcher redrawing but in GB sometimes it would just start stuttering from one home screen to another and had to be restarted.
And I may not remember correctly this one but kinetic scrolling seems faster and smoother in JB than I had ever experienced before in ICS and GB.
The one area of disappointment for people comes from gaming. Well an OS with so many more features is obviously going to take more ram and games are going to get more and more demanding so it is not unreasonable that a new OS makes the phone a less capable gaming device.
Yeesh! I used to have GB. I cannot go back there after running JB. Just CANNOT!
It's like using iOS or BADA.
My NS runs JB smoothly. No flicker no lag. Using Jellybro Rom.
How can you say no, look at the quadrant results
GB (no oc) - 2500
JB (no oc) - 1250
RAM left (same number of running apps)
GB -160-180M
JB - 90-110M
I would never let a quadrant change my view of an os! Im using black bean with matr1x and dont have anything to complain! Gb is ugly and i believe it wasnt faster! As a matter of fact, my phone doesnt feel slow at all...
fieldside99 said:
How can you say no, look at the quadrant results
GB (no oc) - 2500
JB (no oc) - 1250
RAM left (same number of running apps)
GB -160-180M
JB - 90-110M
Click to expand...
Click to collapse
Quadrant is the worst benchmark to go by... A monkey high on PCP can alter i/o scores. And GB not using available RAM is actually a bad thing and will slow things down.
fieldside99 said:
How can you say no, look at the quadrant results
Click to expand...
Click to collapse
Please people, forget about listing synthetic benchmark results as a ultimate measure of the phone. Nobody cares if GB scores twice as much points in Yet Another Very Accurate Benchmark, when JB feels faster and more responsible in everyday usage.
I haven't even thought of downgrading. Why switch to an old out of date unsupported OS When you have access to the most current.
I think the prob might be in your choice of ROMS/kernels try AOKP or Slim with AIR BigMem and freeze or remove the system apps you don't use. Has been working great here. No complaints.
Sent from my phone using my thumbs and a keyboard
I could not go back to Gingerbread after JB, it would be like riding a stallion and then going back to a donkey. The JB interface is so much better than GB and using BlackBean 6 I have no lag at all and it's so smooth!
I agree jelly bean is slightly faster then gb as im users who use over 160+ apps and notice in each faster from 360 mb to 380 ram of 512 use. I have 140 mb left on jelly bean compare to 80 on ginger.
Another thing is when u have no apps download your phone have JB 240 mb free of 360-380 of ram the os show.
GB It was about 210 mb ram left free and 150-70 mb used up for apps preinstalled. 230 mb for os overall.
Like my bionic have 840 mb of ram of it 1,024 mb of ram
Single core
guyz jb and ics are optimised for multi cores.
GB is the best for single cores.
are you sure?
back to Gingerbread?
if you looking for stability and speed, JellyBean it is..
I know.. ICS is a bit sucks..especially 4.0.3
but JellyBean? it's awesome...really awesome...
forget about benchmark scores..it's just pain in a butt, LOL
as long as you run your apps smooth.. what's the benchmark score for?
unless you're the benchmark geek
nexussgeek said:
guyz jb and ics are optimised for multi cores.
GB is the best for single cores.
Click to expand...
Click to collapse
Best in what aspect?
I'm on JB now and sometimes considering going back to ICS, but I have zero desire to go back to anything else. ICS is the first version of Android that I feel I can use stock, with regards to performance, UI design and functionality. JB is a smoother ICS, with the caveat of faster battery drain.
Jb just need ram. 512 mo is the minimum i think.
envoyé de puis mon Nexus s, aiiiiight ! (Michel rules the world)
I tried going back to ICS once, and couldn't do it. I certainly couldn't go back to GB. Too many nice features have been added that I just wouldn't be able to live without ...
I have some Pros of GingerB
*GB battery life is much better than JB. Specially in standby.
*Animations are faster (feels much faster)
*Scrolling is faster
*More available memory at boot.
However JellyB Software is awesome, I love it on my GN and N7. However for the Nexus S I use GB, it isn't fancy but gets all the job done and excellent battery life.

[Q] Strange Benchmark Scores and kernel issue

Hey everyone
I ran antutu benchmark on my nexus 4 running PA v3.97 and i got a score of a little more than 16 thousand and then i install the semaphore kernel and got an embarrassing score of 14162. Whats happening? Arent these a little odd considering that a STOCK Nexus4 has a score of 20 thousand?
And another un-related thing, The carbon rom is not booting up with the matr1x kernel. the google sign comes up and no bootanimation and it stays there.
Really need your help, Thanks
If there is one thing I learnt here it would be that benchmarks are unreliable.
Make sure that kernel matches the ROMs Android version and build, it may not be even supported.
Sent from my Nexus 4 using Tapatalk 4
Stock doesn't score 20k
Check your CPU speed on your Antutu Detailed Score, I would say most likely it was caused by the CPU throttling if it's not 1514 (or something around that),
Custom ROM can be the culprit too, since I only get 13k-14k score with MIUI (from miuiandroid) which is pretty embarassing, that was MTK6579 level.
My stock 4.2.2 doesn't even come close to 20k score. Only 17,900 something.

[ANTUTU] 4.1.2 Vs 4.3 Vs 4.4 (DALVIK & ART)

I wanna share with you some benchmarks I did... with Antutu.
The winner is.... Samsung 4.3 ROM ! :angel:
Warning: the ART score with the 4.4 is not accurate; I had several gapps & gallery crashes while testing the GPU score... but look at the CPU and Dalvik score...
Comparison here:
COMPARISON IMAGE
nice info!
n0my said:
nice info!
Click to expand...
Click to collapse
It seems like GPU works better with the 4.3 samsung rom... but I don't think the score is cheated.... I noticed that Chrome, Facebook and everything involving pinch to zoom, pan, etc operations on some content, it's way smoother with the 4.3 sammy rom.
:victory:
I had different experience myself.
I have n7105 and was on sammy 4.1.2 that came with the phone.
No lags, the zoom in and out was lag free. very smooth.
Updated to CM 10.2 JB 4.3.1 and it gave severe lag problem on browser zoom in and scrolling.
Now on unofficial CM 11 Kitkat 4.4. Now its a lot better than the 4.3.1.
ART is kinda in an Alpha stage but... that said i't's performing wonderfully for me (except not compatibile apps). I think that actually antutu isn't accurate at benchmarking with ART but i've seen an enourmus improvement:good::good:
danyuhuh said:
ART is kinda in an Alpha stage but... that said i't's performing wonderfully for me (except not compatibile apps). I think that actually antutu isn't accurate at benchmarking with ART but i've seen an enourmus improvement:good::good:
Click to expand...
Click to collapse
Sure... I don't trust antutu, I posted the comparison because lot of people ask about benchmarks on it!
I think it's a good news that they are working on making Android more efficient, so let's wait :victory:
What is the 'normal' Antutu benchmark value for stock 4.1.2?
Mine is only 16,412 really far from 20,000+ as shown in the OP's 4.1.2 ROM benchmark.
I am wondering if my low benchmark is because of the eMMC that just got replaced. The thing is that eMMC brickbug check says the chip date is 06/1997 which sounds really old...
Not sure what value this has at all... Oh wait I am, extremely little to none. Sorry OP but you need a far robuster test to give any insight and draw any conclusions. Not having a go, just saying what I see.
Benchmarks are better for comparing phones, hardware differences and of cause kernel tweaks / over clocking / under-voting. Not particularly useful for comparing ROM with ROMs. Try battery rundown tests, signal quality, camera quality, and other common daily use activities if you want to compare ROMs with one another.
EmptyArea said:
Not sure what value this has at all... Oh wait I am, extremely little to none. Sorry OP but you need a far robuster test to give any insight and draw any conclusions. Not having a go, just saying what I see.
Benchmarks are better for comparing phones, hardware differences and of cause kernel tweaks / over clocking / under-voting. Not particularly useful for comparing ROM with ROMs. Try battery rundown tests, signal quality, camera quality, and other common daily use activities if you want to compare ROMs with one another.
Click to expand...
Click to collapse
I feel like you replied that thread just to make polemic.
Maybe you didn't read the thread title.
I zoom it for you:
[ANTUTU]
4.1.2 Vs 4.3 Vs 4.4 (DALVIK & ART)
NOW, do you understand? It's just an ANTUTU SCORES THREAD.
Just read over the forum board, in about every kitkat thread people ask for benchmarks on ART.
Well, I did some on antutu (it's not just a try, it's a medium value) and I shared with others.
Benchmarks are not unuseful. They are an information like others to understand how is evolving a platform.
dvn2008 said:
What is the 'normal' Antutu benchmark value for stock 4.1.2?
Mine is only 16,412 really far from 20,000+ as shown in the OP's 4.1.2 ROM benchmark.
I am wondering if my low benchmark is because of the eMMC that just got replaced. The thing is that eMMC brickbug check says the chip date is 06/1997 which sounds really old...
Click to expand...
Click to collapse
With the latest Antutu they changed the algoritms and now Note 2 scores much higher than before.
With the previous versions I scored about 17000. Try the latest release (with cold phone, to prevent thermal throttling).
Neak kernel make some difference, but not too much... just about 500 points or less, over the stock kernel.
I have 21800 with super stock rom lol. Odex without root or anything else....4.1.2
Sent from my GT-N7100 using xda app-developers app
i thought 4.4 have better ram management.
thanks for sharing
Try out latest Devil Kernel !
latest Devil Kernel has new Samsung GPU Drivers ! on Antutu I get a score around 23k !
Note_two said:
latest Devil Kernel has new Samsung GPU Drivers ! on Antutu I get a score around 23k !
Click to expand...
Click to collapse
On what rom?
it´s CM 11 unofficial by xda member "Thddude"
A good comparison

Poor Benchmarks on KitKat ROMs

I've been flashing all the latest and greatest KitKat ROMs for the LG LS980 Sprint variant and after benchmark testing them all using AnTuTu and Quadrant, I've noticed horrible benchmark scores. To put things in comparison the Stock ROM usually provides 37,000 AnTuTu and 21,000 Quadrant. On the KitKat ROMs, I get around 24,000 AnTuTu and 9,000 Quadrant. So the scores are about half what stock provides. I have tested this running the latest builds of KitKat DU, PA, Nameless, and CM11 to name a few. I use Furnace Kernel to Overclock to 2.5, but even without the Furnace kernel on the KitKat ROMs, I still get low scores so it doesn't seem to be the kernel so much as the AOSP base. Something must have changed in the latest builds because I don't recall having this issue of lower scores with slightly older builds of KitKat.
I just wanted to bring that to the attention of people here. If someone can confirm this that would be great too.
Also here is a link to a thread for an extensive benchmark test of CPU kernels and governors, and also a ROM to ROM feature comparison.
http://forum.xda-developers.com/showthread.php?t=2625656
Running SlimKat on my D802 I get about 33k in AnTuTu compared to 37-38k on Rayglobe Flex. BUT, just like the Nexus 5, benchmarks DO NOT matter. Even though I get a lower score on 4.4, everything from gaming to browsing runs smoother. Also, as for now, I do not recommend installing any custom kernels on 4.4 roms since they only bring instability and higher benchmark scores yet lower overall performance. I truly respect the developers who are doing their best to create stable kernels, but as for know, dorimanx's kernel is the only kernel worth installing and it's only available for stock roms.
Sent from my LG-D802 using Tapatalk
wwewwf96 said:
Running SlimKat on my D802 I get about 33k in AnTuTu compared to 37-38k on Rayglobe Flex. BUT, just like the Nexus 5, benchmarks DO NOT matter. Even though I get a lower score on 4.4, everything from gaming to browsing runs smoother. Also, as for now, I do not recommend installing any custom kernels on 4.4 roms since they only bring instability and higher benchmark scores yet lower overall performance. I truly respect the developers who are doing their best to create stable kernels, but as for know, dorimanx's kernel is the only kernel worth installing and it's only available for stock roms.
Sent from my LG-D802 using Tapatalk
Click to expand...
Click to collapse
If I got a 33k in AnTuTu that would be normal. But I'm talking about 25k in AnTuTu. That's not a small change that can be ignored. That means there is some new code being compiled that is not efficiently making use of the processor's capabilities. I'm hoping that someone will look back at their code and realize that a vital piece of code was dropped and is affecting the whole system.
If it was all about just smooth performance, then there is no reason to even have an Snapdraon 800 in the first place. We can all learn to be happy with the perfectly reasonable performance of hardware that's 2 generations old. But if we are being sold or given software and hardware that is supposed to be better or optimized then it better be and not just rely on consumers to fall under a blind veil of a placebo effect.
Furthermore KitKat ROMs are not coded to support HTML5 whereas the Stock ROM is, so websites will be more compatible with the browsers in the stock ROM.
copitlory said:
If I got a 33k in AnTuTu that would be normal. But I'm talking about 25k in AnTuTu. That's not a small change that can be ignored. That means there is some new code being compiled that is not efficiently making use of the processor's capabilities. I'm hoping that someone will look back at their code and realize that a vital piece of code was dropped and is affecting the whole system.
If it was all about just smooth performance, then there is no reason to even have an Snapdraon 800 in the first place. We can all learn to be happy with the perfectly reasonable performance of hardware that's 2 generations old. But if we are being sold or given software and hardware that is supposed to be better or optimized then it better be and not just rely on consumers to fall under a blind veil of a placebo effect.
Furthermore KitKat ROMs are not coded to support HTML5 whereas the Stock ROM is, so websites will be more compatible with the browsers in the stock ROM.
Click to expand...
Click to collapse
If there is indeed some new code that is affecting then CPU, how can you explain the fact that I never got under 33k in Antutu? (I'm talking about SlimKat and Mahdi, I don't know about others)
Also regarding the placebo effect you're mentioning...so basically you're telling me that the increased smoothness across the whole OS is just my imagination and that it is better to have better results in benchmarks and slower performance? Also, the stock ROM is probably one of the most stable ROMs I've used so far so you really can't talk about lies. Both the benchmark scores and day to day usage are more than acceptable. When you decide to install a custom rom, you automatically need to make sure that you're willing to accept some sort of bugs or errors. The developers are not working for you. You're not paying them anything. So automatically you have no right to complain, just to mention the bugs so that the user developing the rom will try to fix them.
Regarding the browser's compatibility, you may be right, yet I haven't encountered any issues for now. That doesn't mean that they don't exist but as long as it doesn't affect me as an user, then I'm OK.
I'm not trying to say that KitKat roms are perfect. Far from it. But in my case, they're more than usable as a daily driver. Others may not feel this way though.
Sent from my LG-D802 using Tapatalk
http://forum.xda-developers.com/showpost.php?p=48602475&postcount=48
/thread
Benchmarking and roms
Ihave been searchingand searching for a nice rom with good features and low bloat n junk...derpBarebones...Comes with STweaks...OC'd to 2.7 running at almost 37000 benchmark on antutu and now lag or anything...runs VERY smooth and also gets GREAT benchmark compared to the other roms out there...also running the 3.4.1 KK kernel i think IT'S CLOUDY;'S??

Categories

Resources