[STOPPED] CyanogenMod Beta2 Omnia 2 (updated: 04.05.2012)

[Guest shoarmabakpao]

Not really impressive, because it's taken from Spica. There is not many things to change to make it working ;-)

Oh, I see.

Cynoganmod does give some neat extra functions. Just checked them all out on my little brother's LG GT540. There are some cool functions. Like to put the bluetooth, wifi, data on-data off, music player play music on/of in the notification bar. In example the samsung Galaxy S has the same handy feature.

But I would put Cynoganmod somewhere under in the list since we need more improvements in the phone and those bug fixes.

1 thing that you Dev's could take a look is some other Android rom's. Since the LG GT540 (77 euro phone!) feels just as snappy or even snappier than an galaxy S.

The ram on the LG GT540 is156mb. Screen resolution=320 x 480 pixels Mabey this LG GT540 has his own vram. But still our Omnia 2 has 256mb ram. It is just weird that, that LG GT540 has at a fresh boot almost 90mb ram and after like a day use ->task killer-> kill all tasks ->it has then 71mb ram left. Also multitasking is very very very quick on it.

the gpu on it is an Adreno AMD z430 known as the Adreno 200. I have searched alot about it now and I came across that our omnia 2 can only do 4million triangles/sec while the Adreno 200 can do a whopping 22 million triangles/sec!

Could this be true about our omnia 2?

Edited October 28, 2011 by shoarmabakpao

[Guest rom_omnia2]

We removed tweak.sh because it's values caused some problems. There is no difference, becasue unused apps go to swap instead of keeping them in memory.

There isn't any magical tool/script which will make more memory for O2.

I agree, by the way, i'm using some different settings. My lowmemorykiller have this configuration: 1536,2048,1536,4024,14336,18432

I've been using this for a long time, no SOD so far. The good thing is that with the third value low, home screen launcher does not close every time.

But this is good only FOR MY OWN USE. I understand that devs work is pretty much harder because you have to build a configuration for all kind of users, with a safe configuration, avoiding freeze.

.

I also compiled a zImage with 48mb for G3D. But i'm not sure if i did it right, that hexadecimals values was driving me crazy, hahaha, but i'll not bother you devs with that.

Anyway, Angry Birds Rio theres no white squares, around 58mb free using sdcard as swap.

[Guest shoarmabakpao]

I agree, by the way, i'm using some different settings. My lowmemorykiller have this configuration: 1536,2048,1536,4024,14336,18432

I've been using this for a long time, no SOD so far. The good thing is that with the third value low, home screen launcher does not close every time.

But this is good only FOR MY OWN USE. I understand that devs work is pretty much harder because you have to build a configuration for all kind of users, with a safe configuration, avoiding freeze.

.

I also compiled a zImage with 48mb for G3D. But i'm not sure if i did it right, that hexadecimals values was driving me crazy, hahaha, but i'll not bother you devs with that.

Anyway, Angry Birds Rio theres no white squares, around 58mb free using sdcard as swap.

As you compiled a Zimage with 48mb for G3D. Could you mabey also test it and see if it really works? Try out a very demanding game and see if it has white squares (NFS SHIFT mabey)

[Guest marisaleh]

You are assuming I didn't do my homework :P

I looked at some technical documents on the S3C6410 (ARM11) chip beforehand, as well as some other datasheets.

What you say is true, to an extent. The internal DRAM, NAND ROM and Flash memory do have parallel paths, whereas the external (SD) flash memory has a serial one. All memory access paths have a related memory controller together with associated overheads (part of the reason as to why the processor needs cache DRAM).

However, the limiting factor here, in both cases, is the actual speed of reading and writing flash memory at the memory chip level, since (if you look at the specs for current technologies) flash memory access is still relatively slow compared with accessing say DRAM. Transferring the data is a relatively small part of the total access time. The 2GB/4GB/16GB "internal" storage (My Storage in WinMo) is definitely flash memory (and 3 year old memory tech at that).

I'm not sure why you deemed my simple benchmark as too "high-level", since it uses pretty much all the same mechanisms that the operating system would use to, say, write to the swap file. The relative results aren't going to change much irrespective of the underlying hardware implementation. The first test to the SD card is still going to run faster than the second test to the "internal" storage, however much you talk it down :D

Of course, to be fair, the read test (which I didn't post) is much closer and there is the fact that internal storage probably uses (I do not know for sure) less juice.Given the rate at which the battery runs down, it would be useful to test internal versus external (using small and large capacity cards) to see if there is a dramatic effect on battery life.

Please let me explain what I mean by High and Low Level here;

A viable computer system will have;

a CPU, I/O I/F, Storage(RAM, ROM, Flash, Sd etc..)

The CPU communicates with these things using its

own language(specific instruction set, machine code).

Now I call this a Low Level environment.

Communication between Human and the CPU is done through an

interpreter(Assembler).There are different ways to communicate;

programming(C, C++ etc, scripts), using various Operating systems

(Unix, Windows, Android, Benchmark tests etc..).

Now I call this High Level environment.

If low level communication is Serial, then it is far slower than Parallel. For example to transfer an 8bit word serially, The CPU uses 8 Clock cycles, While it uses 1 Clock cycle to transfer the same word in parallel.

If you are confirming that the CPU communicates with My Storage

in exactly the same way as it will communicate with an Sdcard,

then it doesn't matter where you install Android.

You are then left with having to choose which Sdcard to use and how does its performance (access time, read, write speeds) compare

with that of My Storage.

Edited October 28, 2011 by marisaleh

[Guest harrypotter222]

yes and no: sdcard is /storage since internal memory is /mnt/sdcard but it's possible to switch the names you just need to make sure you modify the right lines so it matches what you want.

For me, I follow all the steps except

vold.vstab

instead of using /sdcard

dev_mount sdcard /sdcard auto /devices/platform/s3c-sdhci.1/mmc_host/mmc1

I use /mnt/sdcard

dev_mount sdcard /mnt/sdcard auto /devices/platform/s3c-sdhci.1/mmc_host/mmc1

since i notice all the references in init.rc are defined as /mnt/scard

after I made this change in vold.vstab and restart android, the android recognizes the sdcard parition in "mystorage" as mounted

if I leave it as /sdcard, the sdcard partition is STILL MOUNTED under the file manager, but android doesn't recognize it as the SDCARD under applications depending on it (such as MUSIC)

[Guest rom_omnia2]

As you compiled a Zimage with 48mb for G3D. Could you mabey also test it and see if it really works? Try out a very demanding game and see if it has white squares (NFS SHIFT mabey)

Tried a new version of NFS Shift but didn't work at Froyo, i've no time to look for the right version. But i tried Asphalt 5, probably installed the wrong apk because i'm stuck at main screen, theres a button "quick play" but i can't press it. Anyway i'm able to move the screen around the car, didn't race but theres no white squares at this main screen.

And Chainfire3D really works with 3D games. Before use it, move the screen around the car was "really, really slow". Now it's just "really slow". What a improvement, ham? I'm sure that Asphalt 4 would work like a charm, the same version that we got at WinMo.

Edited October 28, 2011 by rom_omnia2

[Guest F53]

Please let me explain what I mean by High and Low Level here;

A viable computer system will have;

a CPU, I/O I/F, Storage(RAM, ROM, Flash, Sd etc..)

The CPU communicates with these things using its

own language(specific instruction set, machine code).

Now I call this a Low Level environment.

Communication between Human and the CPU is done through an

interpreter(Assembler).There are different ways to communicate;

programming(C, C++ etc, scripts), using various Operating systems

(Unix, Windows, Android, Benchmark tests etc..).

Now I call this High Level environment.

Yes, yes, yes. I have been programming for over 35 years and I started in 8080 machine code, without even an assembler.

But you missed my point. How does this change the real world result of the benchmark? Answer, it doesn't. It takes the time it takes.

If low level communication is Serial, then it is far slower than Parallel. For example to transfer an 8bit word serially, The CPU uses 8 Clock cycles, While it uses 1 Clock cycle to transfer the same word in parallel.

Yes, agreed; parallel transfers are faster than serial (I think I agreed with this on my last post), but the magnitude of the time difference between the transfer of the data and the access (actually getting, or putting, the data word from the memory array) makes the transfer time insignificant. For example, if it takes 1 second to transfer some data serially, 0.1 seconds to transfer in parallel and 100 seconds to access, then it is 101.0 seconds versus 100.1 seconds. The difference is neglible.

If you are confirming that the CPU communicates with My Storage

in exactly the same way as it will communicate with an Sdcard,

then it doesn't matter where you install Android.

No I am not confirming that (and I did not do so in may last post), but see above as to why it does not lead to your conclusion.

You are then left with having to choose which Sdcard to use and how does its performance (access time, read, write speeds) compare

with that of My Storage.

And this is what my benchmark demonstrated; that the access time of higher speed SDHC flash memory cards is less than that of the internal flash memory. Less access time = faster access, since the transfer time is neglible in this situation.

Anyway, I feel we are going a little off-topic here, so if you want to continue the discussion please PM me ;)

Edited October 28, 2011 by F53

[Guest voyteckst]

Yes, yes, yes. I have been programming for over 35 years and I started in 8080 machine code, without even an assembler.

But you missed my point. How does this change the real world result of the benchmark? Answer, it doesn't. It takes the time it takes.

Yes, agreed; parallel transfers are faster than serial (I think I agreed with this on my last post), but the magnitude of the time difference between the transfer of the data and the access (actually getting, or putting, the data word from the memory array) makes the transfer time insignificant. For example, if it takes 1 second to transfer some data serially, 0.1 seconds to transfer in parallel and 100 seconds to access, then it is 101.0 seconds versus 100.1 seconds. The difference is neglible.

No I am not confirming that (and I did not do so in may last post), but see above as to why it does not lead to your conclusion.

And this is what my benchmark demonstrated; that the access time of higher speed SDHC flash memory cards is less than that of the internal flash memory. Less access time = faster access, since the transfer time is neglible in this situation.

Anyway, I feel we are going a little off-topic here, so if you want to continue the discussion please PM me

Maybe You want to help if You have so much experience? PM if so.

[Guest rom_omnia2]

was checking and found a file called init.d increasing the RAM. I took the test and it worked.Initially 60 MB free RAM. Could Voysteckst of a look.File was developed by the Juwe11 xda-developers, now they have developed anapplication called RAM Manager. For those who want to test just download the file init.d and paste in the folder /system /etc and reboot. then check settings, information on the phone and check the total amount of RAM free.

script page

http://forum.xda-dev...d.php?t=1111145

Download link

http://www.megaupload.com/?d=G8DIGKP8

We removed tweak.sh because it's values caused some problems. There is no difference, becasue unused apps go to swap instead of keeping them in memory.

There isn't any magical tool/script which will make more memory for O2.

I agree about memory but i've noticied faster widget picker and settings scroll, Launcher Pro is more responsive too. This script posted by "Antonio Junior" have different values from tweak.sh.

[Guest voyteckst]

I agree about memory but i've noticied faster widget picker and settings scroll, Launcher Pro is more responsive too. This script posted by "Antonio Junior" have different values from tweak.sh.

Post it please.

[Guest rom_omnia2]

Post it please.

The script download is in the first quote at my previous post, anyway here it is:

if [ -e /proc/sys/vm/swappiness ]; then

	echo "50" > /proc/sys/vm/swappiness

fi


if [ -e /proc/sys/vm/vfs_cache_pressure ]; then

	echo "10" > /proc/sys/vm/vfs_cache_pressure

fi


if [ -e /proc/sys/vm/dirty_expire_centisecs ]; then

	echo "500" > /proc/sys/vm/dirty_expire_centisecs

fi


if [ -e /proc/sys/vm/dirty_writeback_centisecs ]; then

	echo "1000" > /proc/sys/vm/dirty_writeback_centisecs

fi


if [ -e /proc/sys/vm/dirty_ratio ]; then

	echo "90" > /proc/sys/vm/dirty_ratio

fi


if [ -e /proc/sys/vm/dirty_background_ratio ]; then

	echo "5" > /proc/sys/vm/dirty_background_ratio

fi

Tweak.sh values:

if [ -e /proc/sys/vm/swappiness ]; then

	echo "10" > /proc/sys/vm/swappiness

fi


if [ -e /proc/sys/vm/vfs_cache_pressure ]; then

	echo "50" > /proc/sys/vm/vfs_cache_pressure

fi


if [ -e /proc/sys/vm/dirty_expire_centisecs ]; then

	echo "3000" > /proc/sys/vm/dirty_expire_centisecs

fi


if [ -e /proc/sys/vm/dirty_writeback_centisecs ]; then

	echo "500" > /proc/sys/vm/dirty_writeback_centisecs

fi


if [ -e /proc/sys/vm/dirty_ratio ]; then

	echo "22" > /proc/sys/vm/dirty_ratio

fi


if [ -e /proc/sys/vm/dirty_background_ratio ]; then

	echo "4" > /proc/sys/vm/dirty_background_ratio

fi

I took out the lowmemorykiller part, cuse it's about memory not speed.

[Guest larrezz]

I've been using Froyo beta 3 now for just a week, Mystorage install.

The software is running very fast in the beginning, but after a while it slows down and I got some SOD's en freezing screens.

Thereby is there a problem with messaging, because I don't receive a lot of SMS's. They are just lost.

I reinstalled the beta for three times now, but I keep having this errors.

Someone else who has the same errors?

[Guest shoarmabakpao]

Tried a new version of NFS Shift but didn't work at Froyo, i've no time to look for the right version. But i tried Asphalt 5, probably installed the wrong apk because i'm stuck at main screen, theres a button "quick play" but i can't press it. Anyway i'm able to move the screen around the car, didn't race but theres no white squares at this main screen.

And Chainfire3D really works with 3D games. Before use it, move the screen around the car was "really, really slow". Now it's just "really slow". What a improvement, ham? I'm sure that Asphalt 4 would work like a charm, the same version that we got at WinMo.

Asphalt 5 with stock Beta 3 does not have white squares. You could try Cut The Rope and see if you can see all the effects and the green monster.

[Guest voyteckst]

The script download is in the first quote at my previous post, anyway here it is:

....

I took out the lowmemorykiller part, cuse it's about memory not speed.

Thanks. According to compiling g3d with less memory there is not hex values... There is clearly stated 64MB in O2 header...

[Guest Boddish]

I've tried using SetCPU to limit cpu on screen off to reduce battery usage but no luck... still using 4% battery/hour at sleep... and it gets slow to wake up...

I'm searching for some app to reduce battery usage right now, since it deplets fast i'm using WM, changing to Android just for tests... hope it gets some patchs to fix it.

I was thinking it was too hot at 35°, but it is Normal for Androids... My friend Galaxy 5 gets 35° all the time too

Well, good work anyway, going to test other things tonight yet..

[Guest rom_omnia2]

Asphalt 5 with stock Beta 3 does not have white squares. You could try Cut The Rope and see if you can see all the effects and the green monster.

Oh, i guess you misunderstood. I did a downgrade, not a upgrade. Beta 3 have 64mb for G3D, i compiled a new kernel with only 48mb for G3D. And as i said, Angry Birds Rio have no white squares even with 48mb, so theoretically next release could bring just 48mb since it's enough for AngryBirds Rio (i'm using it as a base) what results in more free memory for the system.

Some people ask for more 3D improvements, others for more memory... trying to find a middle term.

Edited October 28, 2011 by rom_omnia2

[Guest rom_omnia2]

Thanks. According to compiling g3d with less memory there is not hex values... There is clearly stated 64MB in O2 header...

Yes, i saw this 64MB value. But in reserved_men.h there's also changes in some hex values, i thought it could be because of G3D changes (or even swap). But don't worry, it's just a test that i'm doing, i'll post the results soon if i notice any improvement. Thanks!

[Guest shoarmabakpao]

Oh, i guess you misunderstood. I did a downgrade, not a upgrade. Beta 3 have 64mb for G3D, i compiled a new kernel with only 48mb for G3D. And as i said, Angry Birds Rio have no white squares even with 48mb, so theoretically next release could bring just 48mb since it's enough for AngryBirds Rio (i'm using it as a base) what results in more free memory for the system.

Some people ask for more 3D improvements, others for more memory... trying to find a middle term.

Oooh. Now I get it. So this means that even 64mb for G3D isn't enough to run demanding games.

But running Cut The Rope isn't that demanding I think. It looks like a basic game. The game runs very smooth on Beta3 + chainfire3d driver but the green boss is always missing and at the startscreen some textures are missing aswell.

[Guest rom_omnia2]

Post it please.

About that script, i took this line out:

if [ -e / proc /sys/ vm / vfs_cache_pressure ]; then

echo "10" > /proc/ sys/vm /vfs_cache_pressure

fi

Video (local mp4 and youtube) get slow with this line in. Without is ok, still testing.

Edited October 29, 2011 by rom_omnia2

[Guest mezz_89]

Hi.. I installed Chainfire3D Pro 2.8, but then it shows "Could not acquire SuperUser access ! Is your device rooted? If you are rooted but this message persists, reboot your device ! "

Our Omnia 2 has already rooted? I've tried reboot but then the same thing happen.. I installed SuperUser.apk application but also the same.. Anyone can share the steps? Thanks.. :D

[Guest thanh_nha0706]

Hi.. I installed Chainfire3D Pro 2.8, but then it shows "Could not acquire SuperUser access ! Is your device rooted? If you are rooted but this message persists, reboot your device ! "

Our Omnia 2 has already rooted? I've tried reboot but then the same thing happen.. I installed SuperUser.apk application but also the same.. Anyone can share the steps? Thanks.. :D

Try reboot your phone and open it again...

[Guest voyteckst]

Yes, i saw this 64MB value. But in reserved_men.h there's also changes in some hex values, i thought it could be because of G3D changes (or even swap). But don't worry, it's just a test that i'm doing, i'll post the results soon if i notice any improvement. Thanks!

Check also kernel log for allocation errors.

[Guest zappazz]

Firstly, thanks for your reply (it is the third time i ask and it is the first answer).

I'm a beta1+beta2+beta3 user, and i'm following these thread since the beginning. I looked for the faq you are pointing me, but i didn't find anything about sync,

neither in the first page of this topic (theese are the faq: http://www.modaco.co...ost__p__1804227)

neither here http://www.modaco.co...e-omnia-2-info/ or here http://code.google.c...droid/wiki/FAQ.

I executed the timeupdate script suggested in the project page (http://o2droid.phj.hu/index_en.php), but it didn't solved my problem.

Can you please give me a specific link?

Thanks.

I think it's not a big deal (since it still beta..)you can set your time manually.. if the time is not "refresh".. read post no 2, "Note 2",

and about network time sync, I think dev team just omit it for a while..after the bigger issue solved, I'm sure dev team will working on it.. :)

be patient please..

Ok, time sync based on network is not a solved issue (you are right, it is not a big deal, this beta 3 is great and i'm no longer using WinMo).

My question was different:

i manually set "time sync"=disabled and "use 2g"=enabled, BUT each time i reboot "time sync" is enabled and "use 2g" is disabled.

I took a look into the db where these information are stored /data/data/com.android.providers.settings/databases/settings.db:

-before reboot, the field auto_time (system table) is set to 0 and the field preferred_network_mode (secure table) is set to 1

-after reboot, the field auto_time (system table) is set to 1 and the field preferred_network_mode (secure table) is set to 0

What is changing these 2 values each time i reboot?

I'm patient, but i'm curious :-)

Edited October 29, 2011 by zappazz

[Guest Kapica]

I'd like to suggest something..

Been using beta3 practically since it was released as daily driver... Phone calls, sms, mail, facebook messenger, viber, mp3 player.. Practically all i need.. and my omnia is doing it on android :)

Now, the ram reduction to boost graphics like angry birds and youtube is waste of potential... android on omnia will never be as good as a standard android phone, and i doesnt really need to be?...

Watching youtube, playing angry birds, is it really important? I think not..

If i may suggest, focus on returning that little ram back, so phone could be little faster and more stabile.. i think on beta2 all in all device worked a little faster?

Concerning my usage... 3g active all day, wifi when possible, i get through one day barely, which is fine... i cant expect more...

What i'd like developers to focus, is on fixing the lag on incoming call when phone if sleep, if thats possible... Then, sometimes phone has some running bugs, starts to heating, some process probably goes crazy in backround, i have to restart it...

And y, another small bug, sometimes when i lock the phone screen comes on, but touch unresponsive, have to unlock and lock again.. not a biggie but kinda annoying...

Y sometimes android reboots on its own, but no biggie works fine after it... All in all.. beta3 is like a cyanogenmod on our omnia, and that is amazing...

Gotta say, i plan to buy android phone, but i will find the time before it to donate at least 10$ voyteckst :)

Thats all from me for now, will continue to monitor, what u guys do. :)

p.s. how can u calibrate my touchscreen? is little off, sometimes annoying.... tnx

Edited October 29, 2011 by Kapica

[Guest jingang]

now beta3 is very good，thanks all of the devs，much thanks

:lol: :lol: :lol: :lol: :blink: :lol: :lol: :lol: :lol: