[ROM] joenilan Ideos S7 Froyo Stable - (Android 2.2.2)
- Thread starter joenilan
- Start date
hihi, i got something to share here 
if u all cant find software to format ur partition into EXT4, no fear. go to the market, download Link2SD. after that, run tat app and u will be able to choose what to format for ur partition which is created for the ext4. after format finish, just reboot
if u all cant find software to format ur partition into EXT4, no fear. go to the market, download Link2SD. after that, run tat app and u will be able to choose what to format for ur partition which is created for the ext4. after format finish, just reboot 
manfly9884
Member
- Jun 18, 2010
- 31
- 0
Noob question gentlemen. The SD that needs to be formatted to EXT4, is that the internal or external? Trying to use easeus and it does not have the EXT4 option.
Thanks in advance
i used ext2 and did a full system reset. clearly it's working because it made my internal storage to 6.5gb which was 160meg before. i suggest you do a much smaller partition.
like:
fat32 500meg primary
2gb EXT2 primary
4.5 EXT2 primary
that way Joe here can get Apps2SD to work on the 3rd partition (if 3rd partition exists, run) and use the 2nd partition for Data2emmc . maybe he'll make a custom kernel for this purpose!! please
like:
fat32 500meg primary
2gb EXT2 primary
4.5 EXT2 primary
that way Joe here can get Apps2SD to work on the 3rd partition
(if 3rd partition exists, run) and use the 2nd partition for Data2emmc . maybe he'll make a custom kernel for this purpose!! please
manfly9884
Member
- Jun 18, 2010
- 31
- 0
Thanks for the reply. Is this done on the internal or external sd? Thanks
internal sd - i chose a huge partition 6.5gb and it runs terribly slow. i was warned to use 1.5-2gb max. so i think it would be great if we can use both apps2sd and data2emmc at the same time. i read the docs on apps2sd and it is possible.
that way you can have 2gb or so to install huge apps like navigon then apps2sd to move them to another partition.
the explanation i was given is that the data2emmc basically copies the contents of /data (160gb normally) to the ext4 partition so you don't run into out of storage - however i have a feeling that the internal SD or the EXT4 is slower than the internal rom or the filesystem it uses. very cool - but i'm getting horrendous slowdown compared to the traditional setup with apps2sd - i have to try again with a smaller partition size.
gparted is the only free app that will format to ext4 - but i didn't do that . i just used ext2 and it apparently can format it to ext4 when you do a full system reset.
perhaps the amount of free ram is diminished when handling such a large partition with ext4 versus ext2 - ext4 is far more robust but everything "more robust" has a price.
ext2 is more like FAT32 where you can wreak a ton of damage to the filesystem by improperly unmounting(crashing,running out of power,etc) where ext4 is more like ntfs/journalled hfs from what i understand (google ext2 wiki and ext4 wiki to read up on the filesystems to understand each).
that way you can have 2gb or so to install huge apps like navigon then apps2sd to move them to another partition.
the explanation i was given is that the data2emmc basically copies the contents of /data (160gb normally) to the ext4 partition so you don't run into out of storage - however i have a feeling that the internal SD or the EXT4 is slower than the internal rom or the filesystem it uses. very cool - but i'm getting horrendous slowdown compared to the traditional setup with apps2sd - i have to try again with a smaller partition size.
gparted is the only free app that will format to ext4 - but i didn't do that . i just used ext2 and it apparently can format it to ext4 when you do a full system reset.
perhaps the amount of free ram is diminished when handling such a large partition with ext4 versus ext2 - ext4 is far more robust but everything "more robust" has a price.
ext2 is more like FAT32 where you can wreak a ton of damage to the filesystem by improperly unmounting(crashing,running out of power,etc) where ext4 is more like ntfs/journalled hfs from what i understand (google ext2 wiki and ext4 wiki to read up on the filesystems to understand each).
manfly9884
Member
- Jun 18, 2010
- 31
- 0
Thank you very much for your excellent explanation.
Sent from my IDEOS S7 using Tapatalk
Sent from my IDEOS S7 using Tapatalk
as a rule of thumb, bigger partition size will use bigger block size (except you manually define the block size).
bigger block size is not good for small file. most of the active files in data partition is small files whereas big files (APK) only written once. (which is slow only when you install apk file)
you can try make bigger partition and make smaller block size.
you can format using mkfs.ext4 or mk2efs (using terminal emulator), i think all the files required to formatting is already in the system.
ext4 a little bit slower compare to EXT2, but for some cases it faster then EXT2.
the most important thing is journaling, make it more stable then EXT2 in case unclean unmount.
EXT2 supposed to be no problem, even if unclean unmount, it will do e2fsck on boot before remount it (but required longer time to e2fsck compare to EXT4). but somehow e2fsck don't do the job well and still get corrupted data)
ext2 is not journalled. ext3 is the journal + ext2 = you can mount ext3 as ext2 but not journal enabled. ext4 is whole rewrite to deal with limitations (ie 32 or 65K files per directory).
can you force format to ext4 when factory reset 2nd partition to use smaller blocksize force? you can define during mkfs.ext4 to not use automatic blocksize you want to of course align on boundary for flash.
I would be curious if you could do simple dd benchmark.
also run these benchmarks on all roms so we can compare each rom against each other. (htc flyer text below)
AnTutu – I really have no knowledge on this benchmark, but I can tell you that it took the Flyer a hell of a long time to complete. It scored 2739 when all was said and done, with most of that obviously coming from high amount of RAM (1GB) and the increased clock speed on the CPU. 2D and 3D graphics performance scores weren’t terrible, but were nothing special. The only device not overclocked that I could put it up against was the Motorola XOOM, which apparently scored a 5752 at stock 1GHz speeds. The XOOM dominated CPU performance, staying in the 1400-1900 range down the board.
Linpack – Linpack scores were impressive, but no less was expected from a Snapdragon chipset clocked at 1.5GHz. It scored an average of 56 MFLOPS, though this isn’t as impressive as it once was when Linpack first became the premier benchmark for testing CPU efficiency. Competing (single-core) overclocked to similar speeds actually produce better scores.
Neocore – Although this device doesn’t have Qualcomm’s absolute latest GPU inside, the Adreno 205 was enough to muster up exactly 50 frames per second. Neocore now seems outdated, however, with low resolution and polygon counts being a cakewalk for most GPUs. Note: The Neocore benchmark was made by Adreno Graphics, a brand of Qualcomm’s.
Nenamark 2 – Nenamark is a lot more taxing and it shows in its benchmark results. At the end of it all, 10.9 frames per second was all the Flyer could average and that low framerate was visibly noticeable throughout the benchmark.
Quadrant – 1899-2100 is what I kept scoring with Quadrant, and compared to some stock phones with supposedly weaker chipsets, this isn’t too great of a score.
Smartbench 2011 – The Flyer scored a 1495 in productivity and a 2332 for games. They’re modest scores up against competition but more powerful devices easily has it beat.
can you force format to ext4 when factory reset 2nd partition to use smaller blocksize force? you can define during mkfs.ext4 to not use automatic blocksize
you want to of course align on boundary for flash.I would be curious if you could do simple dd benchmark.
also run these benchmarks on all roms so we can compare each rom against each other. (htc flyer text below)
AnTutu – I really have no knowledge on this benchmark, but I can tell you that it took the Flyer a hell of a long time to complete. It scored 2739 when all was said and done, with most of that obviously coming from high amount of RAM (1GB) and the increased clock speed on the CPU. 2D and 3D graphics performance scores weren’t terrible, but were nothing special. The only device not overclocked that I could put it up against was the Motorola XOOM, which apparently scored a 5752 at stock 1GHz speeds. The XOOM dominated CPU performance, staying in the 1400-1900 range down the board.
Linpack – Linpack scores were impressive, but no less was expected from a Snapdragon chipset clocked at 1.5GHz. It scored an average of 56 MFLOPS, though this isn’t as impressive as it once was when Linpack first became the premier benchmark for testing CPU efficiency. Competing (single-core) overclocked to similar speeds actually produce better scores.
Neocore – Although this device doesn’t have Qualcomm’s absolute latest GPU inside, the Adreno 205 was enough to muster up exactly 50 frames per second. Neocore now seems outdated, however, with low resolution and polygon counts being a cakewalk for most GPUs. Note: The Neocore benchmark was made by Adreno Graphics, a brand of Qualcomm’s.
Nenamark 2 – Nenamark is a lot more taxing and it shows in its benchmark results. At the end of it all, 10.9 frames per second was all the Flyer could average and that low framerate was visibly noticeable throughout the benchmark.
Quadrant – 1899-2100 is what I kept scoring with Quadrant, and compared to some stock phones with supposedly weaker chipsets, this isn’t too great of a score.
Smartbench 2011 – The Flyer scored a 1495 in productivity and a 2332 for games. They’re modest scores up against competition but more powerful devices easily has it beat.
joe - what if we name the partitions. say FAT32 volume name: RESERVED 500meg -> for rom updates/etc [#1,FAT32,PRIMARY]
ext2, 1.5GB [EXT2DATA2EMMC] -> install process will format this to EXT4 and rename to EXT4DATA2EMMC for data2emmc [#2,EXT2, PRIMARY]
ext2, 5GB [EXT2APPS2SD] -> install process will format this to EXT4 for apps2SD and rename to EXT4APPS2SD(or leave it as ext2, EXT2APPS2SD) [#3, EXT2,PRIMARY]
then change the apps2sd hacked script to check the volumename to move apps to. Thus regardless of the location (internal rom, internal SD, external sdhc) - we could use both DATA2EMMC and APPS2SD at the same time, and format them out correctly.
does that make sense? A simple mount probe with the ext4 kernel could locate the partition names and format them to ext4 and rename the volumes when done -> then the apps2sd could be modded slightly to work together with data2emmc (confirmed - author of apps2sd says they both work together).
That way we can have 2GB of internal storage (and for those without internal SD - they could put an SD card in). Also folks with an extra 32gb SDHC class 10 could use that for APPS2SD or they could use the remains of the internal SD card.
Also we could use this method of checking the volume name to store things like DALVIK cache and /cache etc. It may be that a class 10 microsdhc is faster than whats soldered on the mobo - it's possible?
What do you think - this is just an idea i had in the shower today.
I'd prefer EXT4 all day long over ext2 if the overhead is minimal.
So TL;DR: use volumenames to direct the install process to format and use partitions (with no regard to internal/external) to integrate both apps2sd and data2emmc.
I think the idea is solid. It would be cool to have a simple benchmark run to determine best speed - ie if external class 10 sdhc is faster than internal sd -> allow user to move swap/cache/dalvik cache over there. (with big warnings of course).
overclocking i bet we can do 1.2ghz
ext2, 1.5GB [EXT2DATA2EMMC] -> install process will format this to EXT4 and rename to EXT4DATA2EMMC for data2emmc [#2,EXT2, PRIMARY]
ext2, 5GB [EXT2APPS2SD] -> install process will format this to EXT4 for apps2SD and rename to EXT4APPS2SD(or leave it as ext2, EXT2APPS2SD) [#3, EXT2,PRIMARY]
then change the apps2sd hacked script to check the volumename to move apps to. Thus regardless of the location (internal rom, internal SD, external sdhc) - we could use both DATA2EMMC and APPS2SD at the same time, and format them out correctly.
does that make sense? A simple mount probe with the ext4 kernel could locate the partition names and format them to ext4 and rename the volumes when done -> then the apps2sd could be modded slightly to work together with data2emmc (confirmed - author of apps2sd says they both work together).
That way we can have 2GB of internal storage (and for those without internal SD - they could put an SD card in). Also folks with an extra 32gb SDHC class 10 could use that for APPS2SD or they could use the remains of the internal SD card.
Also we could use this method of checking the volume name to store things like DALVIK cache and /cache etc. It may be that a class 10 microsdhc is faster than whats soldered on the mobo - it's possible?
What do you think - this is just an idea i had in the shower today.
I'd prefer EXT4 all day long over ext2 if the overhead is minimal.
So TL;DR: use volumenames to direct the install process to format and use partitions (with no regard to internal/external) to integrate both apps2sd and data2emmc.
I think the idea is solid. It would be cool to have a simple benchmark run to determine best speed - ie if external class 10 sdhc is faster than internal sd -> allow user to move swap/cache/dalvik cache over there. (with big warnings of course).
overclocking
i bet we can do 1.2ghz
manfly9884
Member
- Jun 18, 2010
- 31
- 0
Feel silly right now. I was trying to flash to this rom but accidentally wiped sd. Not I have a black screen with two options: One to reboot and another to fastboot. any help is appreciated.
Similar threads
