Performance: MoDaCo Lag Fix versus EXT4 hack - Galaxy S I9000 General

For a good few weeks, I ran my phone on JG1, and used the MoDaCo lag fix... I wasn't entirely happy with this "solution" maybe because I knew there was a quite strict limit with what could potentially be installed without undoing the lag fix, and then re-applying later - bit too much hassle for me...
However, there was no doubt about it, lag was well and truly gone for me. Very fast in the vast majority of scenarios.
Now I am on JM2, and applied EXT4 hack, and have used it for several days now.
When I had JG1 and the lag fix, I got a benchmark score of around 900 in Quadrant. With the JM2 and EXT4 hack, I now get about 1700...
Oddly though, the phone seems slower now than it did before, despite the much improved benchmark. Don't get me wrong, it's still better than stock, but there is some lag there, and not all of the animations / transitions are as smooth as they used to be.
So, JM2 + EXT4 lag fix looks better on paper, but in reality for me, the normal lag fix script seemed better performing, at least, in terms of the responsiveness of things opening etc.
Anyone else find this?
As a side note, I thought the battery life in JG5+ was meant to be better? Seems exactly the same, all things considered to me as JG1.

Yes, I experienced the same.
With the Modaco Fix the phone was flying! But I hit the 130MB limit pretty soon, so I had no other choice than applying and staying with the ext4 fix, which is indeed a tad slower, but still far beyond Samsungs original method.
And I agree to that, that the benchmark does not reflect the performance in real life usage.

Related

Ryan's lag fix vs. Voodoo

I've got a rooted 2.1 stock Vibrant with ryan's lag fix. My quadrant score went from 876 to 2250 which is great. My linpak for android only went from 8.101 to 8.225 mflops. I would really like my mflops to be in the 10-15 range though. It also seems to be running much quicker. Would I see a bigger boost with voodoo or should I just keep ryans lag fix going. Is voodoo just as easy to install to. Also I've been dying for the Frodo OTA. Would both of these lag fixes work for 2.2
Huh. Huhuhuh. He said Frodo. Huhuhuh. Huhuh.
LOL, it's been a funny day today. This is almost as good as the guy who told the noob to put on three lag fixes, so it will be three time's as fast!!! LOL.
Let's all hope Frodo can get the One Ring to Mount Doom in time. Then him and the rest of Samsung dev team might get Froyo done soon. Only then can the world rest and our linpak score possibly rise!
I saw that post to and it was pretty funny. I just want to know the differences between the to lag fixes, and which one is better to have.
How about try them both and you be the judge of it? Results vary on people's setups. If I were you, I would wait for a more stable version of VooDoo.. I have applied it but many people have reported having problems detecting their SD card (external and internal). To fix this, you'd have to restore your apps using Titanium Backup etc etc.
I would definitely say VooDoo, I have used RyanZA and VooDoo. But since you're a new user, I'd just wait for the more stable version.
Does Voodoo reformat the entire SD to ext4 or does it symlink like Ryan's?
If it changes the whole file system to ext4 then I would say it has to be faster.
thanks foe responses, I will wait until voovdoo goes stable an see the performance myself.
Droidicus said:
Huh. Huhuhuh. He said Frodo. Huhuhuh. Huhuh.
Click to expand...
Click to collapse
lolol
________________
you CANT get 10-15 mflops, if you read how the lagfixes work you would understand what you are measuring. i think with OC you might get mid 9's

Quadrant score for galaxy s running froyo

I got 1932
Sent from my GT-I9000 using XDA App
Without any lagfix?!
You should really give more detail, which lagfix if there was one or just stock froyo from Samsung?
leoon said:
Without any lagfix?!
Click to expand...
Click to collapse
No, I'm sure he's got OCLF. Without lagfix froyo isn't any quickier than eclair as of now. Actually, it's slower. I used to get 2200 points with JM9 + OCLF. Now with JP6 + OCLF i get 1850p.
aitzo said:
No, I'm sure he's got OCLF. Without lagfix froyo isn't any quickier than eclair as of now. Actually, it's slower. I used to get 2200 points with JM9 + OCLF. Now with JP6 + OCLF i get 1850p.
Click to expand...
Click to collapse
Yeah that's what I thought, though I don't know why people post like he does there is no point to it, in no way does it help benifit the community.
aitzo said:
No, I'm sure he's got OCLF. Without lagfix froyo isn't any quickier than eclair as of now. Actually, it's slower. I used to get 2200 points with JM9 + OCLF. Now with JP6 + OCLF i get 1850p.
Click to expand...
Click to collapse
Do you mean it's slower, or rather that it gets a lower Quadrant score? Those are two entirely different statements. I realize the OP mentions Quadrant scores specifically and I apologize for sort of imposing my opinion, but such benchmarks, especially with such little score difference and such major operating system differences, do not tell you a whole lot at all.
One separate test of Quadrant might bottleneck the entire score, even when all other parts of the system are much faster. Just look at how a simple 'lag fix' suggests the phone is twice as fast (quadrant goes from 1k to 2k) which of course, is not the case. Also look at how Cyanogen gets a 3K score with a simple hack, where the phone speed was barely altered.
So I am curious about real world performances:
Do you still get lags on Froyo?
Does the GPS take only a few seconds to get a fix?
Does it launch apps almost instantly stock or with some lag fix?
Does it feel ultra responsive, as should be the result of using a JIT compiler?
Does it render webpages faster than iPhone 4 like the Nexus One, like it should?
Does it still run actual applications, such as Quake 3, faster than any HTC despite their higher Linpack scores, and is there an increase in FPS since Eclair?
Would love to hear about this!
Do you still get lags on Froyo?
- I think it lags more than last eclair builds
Does the GPS take only a few seconds to get a fix?
- Same as JM9. Haven't played with gps in outdoors so can't be specific with fix times
Does it launch apps almost instantly stock or with some lag fix?
- Same as JM9, or slower. Lagfix makes it ok.
Does it feel ultra responsive, as should be the result of using a JIT compiler?
- No, even though others have said that JIT is on.
Does it render webpages faster than iPhone 4 like the Nexus One, like it should?
- Webpages are ultra laggy when there is flash content
Does it still run actual applications, such as Quake 3, faster than any HTC despite their higher Linpack scores, and is there an increase in FPS since Eclair?
- Don't know
nin2thevoid said:
So I am curious about real world performances:
1 Do you still get lags on Froyo?
2 Does the GPS take only a few seconds to get a fix?
3 Does it launch apps almost instantly stock or with some lag fix?
4 Does it feel ultra responsive, as should be the result of using a JIT compiler?
5 Does it render webpages faster than iPhone 4 like the Nexus One, like it should?
6 Does it still run actual applications, such as Quake 3, faster than any HTC despite their higher Linpack scores, and is there an increase in FPS since Eclair?
Would love to hear about this!
Click to expand...
Click to collapse
1: yes (when no lagfix applied)
2: yes <5s on warm start, <30s on cold start
3: yes with lagfix only
4:i don't think thats the result of the compiler, just the lagfix but yes, and the JIT does make things a little faster overall
5: no the browser is a bit slow i don't know why. Even Fennec alpha is much faster.
6: yes Q3 is extremely smooth, full details + lightmap, getting 56-60fps (=FPS cap on the SGS), it never slows down
Note that Q3 is using the NDK as in native app. I don't know if Linpack tests native code.
Do you still get lags on Froyo?
Not that i've noticed but I will be installing Voodoo anways.
Does the GPS take only a few seconds to get a fix?
GPS is A LOT better.
Does it launch apps almost instantly stock or with some lag fix?
Certain Apps launch almost instantly where others take a few seconds such as PikPok Games, Fruit Ninja, etc.
Does it feel ultra responsive, as should be the result of using a JIT compiler?
Seems fast to me but only as it's a new install will wait a few days, plus I install Voodoo's fix anyway for the screen fixes too.
Does it render webpages faster than iPhone 4 like the Nexus One, like it should?
No, we have flash it's more demanding but you can set it to be on demand then its just on par with them no major difference.
Does it still run actual applications, such as Quake 3, faster than any HTC despite their higher Linpack scores, and is there an increase in FPS since Eclair?
Not sure but all games i've played are smooth without a hiccup.
hope this helps
1000 in Quadrant and 14 in Linpack
New Froyo+OCLF = 1722 Quadrant
No problems. Gps, camera,market... all is ok.
How are you guys getting such a large score? I'm getting about 1000 and it shows the Samsung Galaxy S on the graph even less (I assumed 1000ish was better than the average score). Or am I using some newer version of the app?? :/
XXJPK with OCLF
I've tested 3 times with Quadrant. Scores: 1550, 1764, 1995.
Phone works much faster after OCLF.
I takes 2 seconds to open a book in Aldiko instead 10.
My galaxy s with froyo, one click root and one click lag fix scores 2029. If i could post a url of the picture i would, but i'm not allowed
ive tried jp6 and jpm and not impressed with either. still getting some lag, even when switching home screens with a live wallpaper.
If you use oclf with quadrant you want get a real result.
What i understand it's one database test that is not real.
It don't do it right. The result is that it seems faster then it is.
Sent from GT-I9000 Jpm OS kernel with z4mod
My quadrant bench mark is 2017.
I've got JPK, OCLF v2.2 and SETCPU installed.
If anybody is interested in a little comparison to the HTC Desire:
1550 on LeeDroid 2.2f (Froyo) with OC to 1153 MHz. Some got speeds upto 1700 with more overclocking, but that'll heat up the phone beyond healthy values, I guess. Do not want to put in a fan there .
So with some optimizations like lag fix and file system enhancement the Galaxy S is a really fast Smartphone indeed, let alone the 3D graphic performance. Cudos!

FINALLY a fast phone (HTC Magic 32B)

Hey there,
I went through a lot of trouble to get my HTC Magic (32B) working really well,
just wanna share this for people who have issues with the phone's performance as well.
I've tried many things, including a swap partition (on a fast SD-Card) as well as compcache.
Turns out, disabling both did the trick. It FINALLY went from unusable to fast and enjoyable!
My current rom:
Cyanogenmod 6.1.0 RC1
Settings->Cyanogenmod-->Performance:
Compcache DISABLED
JIT Enabled (I noticed an increase in CPU Power with the linpack benchmark,
but it still seems underpowered at 3.3 mflops)
Dithering Disabled
Keep Home App in Cache Enabled
Keep MMS-App in Cache Enabled
VM Heap 16m
Also I am using SetCPU to overclock to 576mhz at the moment(feels like the battery is draining faster).
Another thing I did was to reduce number of Homescreens in ADWLauncher down to 3, since thats 'nuff for me.
Hope this helps someone out there, as I was about to shatter my phone into a million pieces :O
On a sidenote: How many mflops do you get with your Magic?
I have the same findings There was a point when swap, overclock, apps2SD, and compcache all would make a major speed-up on your phone, but the CM6.1RC and the nightlies that shortly preceded it have given me the best performance in my magic's history with all of it turned off.
My phone is configured very similar to yours. I went up to vm24 to smooth out how heavier applications run. I left JIT on despite the theoretical drawbacks because I notice no performance hit and it makes a difference on fractal rendering apps (I'm indifferent to benchmarks).
The CM forums seem to be in a bit of turmoil atm because people are slowing down their phones by applying mods that genuinely sped it up in the past.
But for the first time I can be widget-heavy, always go straight back to home, use window animations, never have choppy music (even when using navigation), use alternate keyboards with no performance hit, and even run a good few live wallpapers. All with unprecedented battery life? Not sure what CM did but he/they sure did it right.
One thing - overclocking your CPU will better your benchtest results but I think you'll surprise yourself by leaving it stock, and even (gasp!) underclocking it. CPU clockrate is one of many possible bottlenecks and considering the 1.5 and 1.6 official roms were capped at 384 (or 352 I can't remember), I doubt it is the culprit of possible slowdowns. I run at 384 almost all of the time and have noticed no side effect except for extreme battery life and added stability (615 gives me reboots). The only differences I've seen overclocking make are in fractal rendering apps and benchtests but Pandas vs. Ninjas and Raging Thunder rock at 384.
glad to see I'm not alone with this
I can still see the phone running out of ram when alot is going on, but its ok.
I'll try your suggestion and underclock the phone.
Another small tweak I just found:
under mobile network settings there is an option "only use 2G", which is checked by default. turning this off improved my connection speed (obviously).
but again this comes at the cost of battery-life.

[Q] Overclocking w/ NAND Desire HD/Z builds

Has anyone else noticed that when using NAND Desire HD/Z sense builds that their phone seems to run more smoothly after the 1st day or two running at stock speeds than it does when running overclocked?
I was getting a bit of lag here and there, and as expected after having streamed music literally all night via wifi the phone was a real dog first thing in the morning until the memory cleared out.
I noticed after a couple of days my phone would spontaneously reboot - I was overclocking to 1.5ghz using MDJ's 10.3OC kernel, so I figured that it was the overclock causing the reboot - sure enough the phone runs smooth as glass now that I've throttled it back to 1ghz (interactive governor).
Is it just my imagination?
It is hard to say really, as you will have different apps, widgets and syncs set up to me, plus we are running different builds, thus a direct comparison is very hard.
All that matters is what works for YOU.
FYI I don't find any meaningful benefit by overclocking. So I don't bother. I haven't really noticed a speed difference either way. If however it made a positive difference, I would.
Just for the record, benchmarking and Quadrant scores are nothing more than a waste of time and are only useful for arguments starting with the words "Mine is better than yours because...".
Curiousity always gets me to run Quadrant, but having seen varied phones with roller-coaster scores, I know the scores are not really any kind of valid gauge for real world performance.
I'm going to run my phone at stock speeds for the rest of the week and see how it does. I've got a feeling it will be running a lot smoother.

Cm7- Vm Heap Size on Htc Wildfire- Performance optimization

Hello,
two days ago i changed the size of my Vm-heap to 32mb. I was curious after i read in some posts that this could affect the performance of resource-intensiv apps. As i am using navigon-navigation a lot i had to give it a try...
After ive changed it and rebootet the device, navigon works MUCH better then before, track calculation is much faster and the app is much more responsive and running fluent. The difference is highly noticeable. For other apps like my browser i experiecend the same.
I read that the downside of it is that it is possible that apps could be moved out of memory because some other apps now use more ram and so the jumping between apps could be more time intensive. But i have noticed no slow down at all.
(Wildfire is not a fast phone, but i think 384 mb ram is really good for that kind of phone, some other phones like galaxy ace have only 287 RAM with better cpu)
My experience is that a lot of apps running much better know, and for me its a absolute performance enhancement, I would even go so far that i would say it was the best performance-enhacement after overclocking the device. So if you are running resource-intensive apps (like navigation browsing) you should definitely give it a try.
Whats your experience with that ? Ever changed the vm heap size? Noticed differences??
Cheers.
im rocking 32 since i s-off my phone with alpharevx beta,and yes its better for intensive apps (imo) but the one thing i cant understand is way my phone i lagging when i get a call??some times the ringtone is playing and the screen is black,any tips thx
darkstep said:
im rocking 32 since i s-off my phone with alpharevx beta,and yes its better for intensive apps (imo) but the one thing i cant understand is way my phone i lagging when i get a call??some times the ringtone is playing and the screen is black,any tips thx
Click to expand...
Click to collapse
im running my phone with the ondemand governor with cpu-min:528 and cpu-max:691 and i have no lag when someone is calling.
Nhs666 said:
im running my phone with the ondemand governor with cpu-min:528 and cpu-max:691 and i have no lag when someone is calling.
Click to expand...
Click to collapse
It is indeed said that such lag is due to the 'extreme' underclocking of the processor.
I'm on CM7 RC1, use SMARTASS with 245 as a minimum, and haven't noticed any lag yet..
Sent from my HTC Wildfire using XDA App
ErwinP said:
It is indeed said that such lag is due to the 'extreme' underclocking of the processor.
I'm on CM7 RC1, use SMARTASS with 245 as a minimum, and haven't noticed any lag yet..
Click to expand...
Click to collapse
I didn't want to implicate that the ringing lag is about the low freq., but i noticed that with such low freq's the phone is not so responsive when pulling suddenly the notificationbar, get it out of standby and such things. Never noticed something like that??
I was testing about it too. i settled with 32mb heap size, which i've found to be the best practice. Anything lower would make apps like navigon or opera lag, anything higher, would give me force closes on apps for some reason.
I also tried to experiment with compcache, with the following settings "disabled, 18%, 26%" and i think 26% makes my wildfire a bit snappier, although it might be just a placebo effect, because there's no difference in quadrant/linpack benchmarks.
why placebo effect....the question is how good can a benchmark represent the practial application in daily usage. Anybody?
I tried vm heap bigger than 32mb too. But while running navigon it seems to me not much as a big improvement so i switched back.
The compcage thing will be the next thing im testing.
Nhs666 said:
why placebo effect....the question is how good can a benchmark represent the practial application in daily usage. Anybody?
I tried vm heap bigger than 32mb too. But while running navigon it seems to me not much as a big improvement so i switched back.
The compcage thing will be the next thing im testing.
Click to expand...
Click to collapse
I have to agree with you Whatever makes my phone "feeling" snappier and faster is good enough in my eyes!
You should really check Juwes Ram optimization thread in the Android development section for some serious performance gain. I flashed it yesterday in CM 7.1 Nightly 144 and the responsiveness boost is quite noticeable.
i agree the synthetic benchmarks are not that important its the daily use performance that mater to me
I have disabled compcache and enabled swap and my phone is snappier than 18%compcache may be Coz of compressing and decompression techniques that slow the process in compcache and scores 3.9 to 4.1 on linpack score with 576 MHz speed...
Sent from my HTC Wildfire
vijaykirann said:
I have disabled compcache and enabled swap and my phone is snappier than 18%compcache may be Coz of compressing and decompression techniques that slow the process in compcache and scores 3.9 to 4.1 on linpack score with 576 MHz speed...
Sent from my HTC Wildfire
Click to expand...
Click to collapse
Hmm i have a swap partition already made just in case and i might try it, but in theory a swap partition (even worse if it's a file) is much slower than the compressed cache due to the SD Speed limitations, unless we are discussing about class 10 cards.
anyone ?

Categories

Resources