Fascinate: Rooted, Lagfix and Quardrant

[stanrc]

I just find it curious that the exact same phone with the only difference being the apps installed would have such a variance.

 

[713King]

If you root how do you unroot in case anything happens.

 

[DHGroove]

Rooted and LagFix
Love this phone!!!! Rooted and applied Lagfix and Quadrant is 2k and above up to 2300 Bing has been bonged, maybe it's not a word but you know what I mean and Google Search rules this phone. DAME THIS PHONE IS FAST!!!. Own rooted D1 and tried DX and D2 and returned them both days later. Wife hates it because the phone is forever in my hands. Well, love her but all I can say is, "Honey it's not going back!"

I just got moist.:shocked: Definitely picking this sucka up this weekend.

 

[tobytl]

Just to clarify. The insane quadrant scores are coming from the lagfix. It creates a virtual file system, which in this case means most of the second to second io events never hit the underlying file system, apart from raw read/write. This will make the phone score higher, and indeed feel faster. But, Quadrant currently gives too much weight to IO scores for their tests, they weren't built for it. It's like measuring contrast ratio on a true OLED screen, the blacks are so black, the numbers become meaningless, or mathematically impossible in some cases (yay division by zero). The tests just weren't meant to measure that.

Take a look at the green section of the first two quadrant graphs, that's your IO score.

I was correct. The Galaxy S phones use the PowerVR SGX540 GPU while the Droid, Droid 2 and Droid X use the PowerVR SGX530. Even the iPhone 3gs has a more powerful phone than the Droid X with its PowerVR SGX535.

Basically, the Fascinate can do 90 million triangles a second while the Droid can only do 7 million triangles a second. That is of course a stock Droid with its CPU and GPU running at slower speeds that the Droid X, but I would say the Fascinate is still 3-5 times more powerful than the Droid X when it comes to GPU horsepower.

Sentiment is correct, facts are well... The iPhone 3GS had basically the same SoC as the droid 1. More or less untouched Cortex A8 CPU, PowerVR SGX 530 GPU. Apple didn't start using the SGX 535 until the A4 SoC (iPad/iPhone4).

Triangles/s are nearly meaningless. 1) the numbers aren't even being measured the same way. To bring them in line, it's more like ~80Mt/s for the 540, and ~14Mt/s for the 530. 2) there's more to overall performance than just triangles. I/O, shading, PP, etc. When all is said and done, real world performance is much closer (though still a blow out). The 540 is 1.5x-2x faster than the 530 in 3D. The chip itself has further potential, but you start hitting bandwidth walls fairly quickly. I'll post some DX quadrant screens shortly, they will absolutely pale in comparison to these, but it'll give a graphics performance baseline.

 

[luv2increase]

Just to clarify. The insane quadrant scores are coming from the lagfix. It creates a virtual file system, which in this case means most of the second to second io events never hit the underlying file system, apart from raw read/write. This will make the phone score higher, and indeed feel faster. But, Quadrant currently gives too much weight to IO scores for their tests, they weren't built for it. It's like measuring contrast ratio on a true OLED screen, the blacks are so black, the numbers become meaningless, or mathematically impossible in some cases (yay division by zero). The tests just weren't meant to measure that.

Take a look at the green section of the first two quadrant graphs, that's your IO score.

I was correct. The Galaxy S phones use the PowerVR SGX540 GPU while the Droid, Droid 2 and Droid X use the PowerVR SGX530. Even the iPhone 3gs has a more powerful phone than the Droid X with its PowerVR SGX535.

Basically, the Fascinate can do 90 million triangles a second while the Droid can only do 7 million triangles a second. That is of course a stock Droid with its CPU and GPU running at slower speeds that the Droid X, but I would say the Fascinate is still 3-5 times more powerful than the Droid X when it comes to GPU horsepower.

Sentiment is correct, facts are well... The iPhone 3GS had basically the same SoC as the droid 1. More or less untouched Cortex A8 CPU, PowerVR SGX 530 GPU. Apple didn't start using the SGX 535 until the A4 SoC (iPad/iPhone4).

Triangles/s are nearly meaningless. 1) the numbers aren't even being measured the same way. To bring them in line, it's more like ~80Mt/s for the 540, and ~14Mt/s for the 530. 2) there's more to overall performance than just triangles. I/O, shading, PP, etc. When all is said and done, real world performance is much closer (though still a blow out). The 540 is 1.5x-2x faster than the 530 in 3D. The chip itself has further potential, but you start hitting bandwidth walls fairly quickly. I'll post some DX quadrant screens shortly, they will absolutely pale in comparison to these, but it'll give a graphics performance baseline.

Thank you for that post. I wonder if anything like the lagfix could be implemented on the Droid X?

Sent from my DROIDX using Tapatalk

 

[tobytl]

Thank you for that post. I wonder if anything like the lagfix could be implemented on the Droid X?

Sent from my DROIDX using Tapatalk

Lol, great minds... I've been tinkering with just that actually since last night (though I did have to crash, been up for 2 days...). I honestly don't think it would benefit that much, as we on the DX are already using Ext3. The only useful change is the removal of the journal really. I'm throwing some stuff around, tweaking the FS parameters and such, but it's been a long time since I've delved into this... Need to test the raw speed of the NAND, may have some other ideas for a virtual FS, we'll see.

 

[destroyerbmx]

Stock w/ just root - 898

After lagfix - 2071

Cant wait until a clean 2.2 rom is installed with no bloatware

 

[tobytl]

Thank you for that post. I wonder if anything like the lagfix could be implemented on the Droid X?

Sent from my DROIDX using Tapatalk

Lol, great minds... I've been tinkering with just that actually since last night (though I did have to crash, been up for 2 days...). I honestly don't think it would benefit that much, as we on the DX are already using Ext3. The only useful change is the removal of the journal really. I'm throwing some stuff around, tweaking the FS parameters and such, but it's been a long time since I've delved into this... Need to test the raw speed of the NAND, may have some other ideas for a virtual FS, we'll see.

First of all, as I was originally saying, this test is broken when it comes to I/O. Cyanogen proved it by mounting quadrants data dir to tmpfs, I'm proving it while still using a physical FS on the NAND (The partition is 2GB with 800MB being used, there's no way I mistakenly put the whole thing in ram... though that would be awesome...).

Completely stock:
View attachment 15506

Well... Can't use ext2 on stock DX kernel, won't mount partitions as ext2 either through loopback or from physical media. Soooo, I tried with ext3 through a loopback (basically the same as the Galaxy S lagfix with ext3 instead). Surprisingly, there was improvement, though not much:
View attachment 15505
Apparently the loopback interface itself does some write caching... But still stuck with ext3, and that damn journal...

I then went back and tweaked the fs parameters, and put the journal on a ramdisk (rebooting is tricky...), and got some even more surprising results...
View attachment 15507
Just as a note, apparently, the loopback is caching so much (or at least long enough) for this benchmark to effectively only be writing/reading ram with the tweaked fs. I get only slightly higher scores (~300) if I mount quadrant's data dir to tmpfs. More evidence the test is broken, without even outright cheating with tmpfs...

I also patched in a working version of stagefright (yes, quadrant tests it, and it plays h.264 just fine without sync issues... even records video and plays it back). It has many other problems, but so does opencore... Lagfix/Tweaked Lagfix:
View attachment 15509View attachment 15510

Does it "feel" any faster? Actually yea, yea it does. Most of the standard android "hiccups" are gone, unlock is super fast, and there's no real lag on the home screen. Is it worth the trouble for the DX? No, not really. The DX doesn't suffer from nearly the same level of lag that the Galaxy S phones do, and this hack is a _lot_ more work on a DX. (and I bricked mine 3 times getting the ramdisk to initialize a journal properly on boot... that was a bit of a *****)

 

[dricacho]

I just rooted and voodoo lagfixed and did 3 runs with Quadrant, an hour in between. I also closed all apps (except for the 2 auto-starting apps) before doing the test and this is what I got...

 

[khaf]

I just rooted and voodoo lagfixed and did 3 runs with Quadrant, an hour in between. I also closed all apps (except for the 2 auto-starting apps) before doing the test and this is what I got...

The scores posted by DJORDAN were from a lagfix that just inflated your scores. The ones from dricacho are legit min are a constant above 1700 check it out I made a post about it in here its project voodoo. You just have to try it out, you will see when you install the kernal. for anyone who doesnt pay attentyion to scores but looks at real world performance check it out.

http://www.droidforums.net/forum/sa...opment/87858-project-voodoo-finally-here.html

Hee is a lil explanation about it taken from the project voodoo site

On a side note between me my wife and friends I have access to most of the heavy hitter android phones and none of them out perrform my fascinate not the DX DI D2 Evo4G or Nexus1. Wait till the color fix comes out too.

Lagfix, Galaxy S Lagfix that work
Galaxy S hardware is different from previous generation’s Android phones.
To extend /data partition, the place where all your apps are installed and personal data are stored, Samsung use 2GB from the internal SD card.

There’s nothing wrong with this approach, but they use the proprietary RFS filesystem they build as a legacy FAT, adding journalisation and POSIX permission. Both are required to run safely Android.
Unfortunately, this filesystem tries too hard and don’t provide a consistent experience with many applications installed or a few days uptime.

And here comes Voodoo lagfix to the rescue

By replacing the faulty RFS filesystem with the Linux-standard Ext4 filesystem, Voodoo lagfix simply restores I/O performance and global user experience of the Galaxy S to what it should have been to begin with.
And the performance level is awesome