12.19 3d package with better performance for testing

[ryanmv213]

Originally Posted by ndno

OK, here's my working all-in-one package version using andboot.

I finally got one of my "free" file servers to work . I'm using dropbox. Here's the link: http://dl.dropbox.com/u/12488850/andboot.7z The package is a 7zip format; if you don't have the utility, you can grab it here: 7-Zip It supports all other formats as well: rar, zip, tar, etc.

NOTE: This is off tiad8's 12/19 updated files (3d driver, gralloc, zImage, modules, etc); I took the FRX03 baseline and updated it with his files and inserted the 3d driver, gralloc, bootanimation, build.prop into the system.ext2 so that you don't have to worry about making sure where those files are and whether froyo.user.conf needs to mount them.

Are you not having the same results with the newer kernels?

[tiger2wander]

Yeah, its normally to move all files and folders under your "andboot" directory root of your sdcard. I can confirm that HaRET will boot up with that structure because of what's I see "init" script inside rootfs.img is:

1. Check for "rel_path" HaRET's parameter is set, then it will set $CARD_PATH (represent as root dir) variable to that path.
2. If the $CARD_PATH is not set then set it to "andboot"
3. If "/sdcard/$CARD_PATH" is not exist or not a directory then set "$card" back to sdcard's root directory: /sdcard

By the way, the important thing is: you need to be sure that your sdcard don't have "andboot" folder in the root, so HaRET will boot normally!

About performance of "flat" structure, I think android will faster in most it's IO actions because of short path can make shorter time to lookup and complete a IO actions.

Best regard!
Uoc Nguyen

[kcwebmonkey]

Originally Posted by tiger2wander

About performance of "flat" structure, I think android will faster in most it's IO actions because of short path can make shorter time to lookup and complete a IO actions.

As a developer I have never heard of I/O time being shorter/longer based on whether a file is in a subfolder or not... it takes just as long to access the file regardless of where it is in your folder structure.

[arrrghhh]

Originally Posted by kcwebmonkey

As a developer I have never heard of I/O time being shorter/longer based on whether a file is in a subfolder or not... it takes just as long to access the file regardless of where it is in your folder structure.

I would agree completely. I'm fairly confident this folder path mess is just placebo effect.

[manekineko]

Originally Posted by arrrghhh

I would agree completely. I'm fairly confident this folder path mess is just placebo effect.

I agree it sure sounds weird as hell. However, Tiad8 said he saw the difference in benchmarks.

[demonlordoftheround]

Wirelessly posted (Mozilla/5.0 (Linux; U; Android 2.2; en-us; MSM Build/MASTER) AppleWebKit/533.1 (KHTML, like Gecko) Version/4.0 Mobile Safari/533.1)

Ran the benchmark and scored 563. Not as good as tiad8, but the performance difference is noticable. I have a slow card most likely.

[InvincibleLiving]

Originally Posted by tiad8

this is what i have on my root sd card now.

1-android
2-android apps
3-bootanimation
4-build.prop
5-cache
6-conf
7-data
8-media
9-tmp
8-tapatalk
9-openfeint
10-program file (wm stuff should be here)
11-titanium backup
12-dcim
13-download
14-shootme
15-tmp
16-data.img
17-gralloc 11.29.10
18-libGLES_qcom 12.18.10
19-haret
20-initrd 8.18.10
21-libGLESv2_adreno200 12.15.10
22-lib3dtools_adreno200 12.15.10
23-modules-2.6.27.46-01232-gb7033df.tar 12.18.10
24-roofts 11.08.10
25-startup
26-system.ext2 11.19.10
27-ts-calibration 7.9.10
29-zimage 12.18.10

new installation (fresh install).
1-android apps
2-conf
3-media
4-haret
5-roofts 11.08.10
6-startup
7-system.ext2 11.19.10
8-ts-calibration 7.9.10
9-zimage 12.18.10
10-initrd 8.18.10
11-gralloc 11.29.10
12-libGLES_qcom 12.18.10
13-program file (wm stuff should be here)
14-modules-2.6.27.46-01232-gb7033df.tar 12.18.10

THANKS

Thanks! Answered my question PERFECTLY!

Do you have any plans to update to any of the newer 123X kernels (1239 just came out last night) with all of your tweaks intact?

[tiad8]

Originally Posted by invincibleliving

thanks! Answered my question perfectly!

Do you have any plans to update to any of the newer 123x kernels (1239 just came out last night) with all of your tweaks intact?

i will check this out today. Thanks

[ndno]

Originally Posted by ryanmv213

Are you not having the same results with the newer kernels?

I have tried the default one that came with FRX03 but wasn't get good results. Besides that, I'm only using the zImage and modules posted by tiad8.

For those who tried my all-in-one package, does it work? How's the performance? Mine is running as smoothly as tiad8's video. If you notice, I don't have any of the drivers (*.so) or other extra files in the folder because they have been updated in system.ext2.

[tiad8]

hi guys, important

I did more testing with anboot and w/o
im still getting low score. with andboot with OC at 646800

test 1 with anboot. Neocore 18.1 quadrant 579 with driver 3d 11.26 plus gralloc

test 2 W/O anboot. Neocore 19.2 quadrant 610 with 3d driver 11.26 plus gralloc

test 3 with anboot. Neocore 17.5 quadrant 510 stock drivers (no drivers mounted)

test 4 W/O anboot. Neocore 18.5 quadrant 547 stock drivers (no drivers
mounted)

guys, stay with what works for you. if people likes to use anboot fine, if people don't want to use anboot that fine too.

remember, this is just a few testing.

ON A SIDE NOTE--------------

i was playing with sleep mode setting. its very interesting what i found. i have do a few more testing.

THANKS