Related
Just a quick question for you guys: I recently purchased an Infuse from the marketplace here, and so far I've gotten an absolute TON of lag using the device. I've flashed Infused 1.5 and Bionix with the Infusion kernel, and so far the lag has gotten better, but it's not totally cleared up. I'm definitely not new to Android, but I haven't seen anything like this before. I know that Quadrant isn't a reliable benchmark, but I'm scoring about 1900, OCed to 1.5Ghz.
Just to give an example: I'm using desktop visualizer to add some custom icons to my screen. When I go to select an icon image, it's fine, but it freezes for about 90 seconds when a list of my apps comes up.
Anyone experience this before? And if so, how did you solve it? If this was widespread, I'd imagine there would be complaints all over here.
Which app are you using to set your overclock settings, how many clock speeds do you have enabled, are you under or over volting, and *which governor* are you running?
I have seen some similar problems here when I was using conservative governor with poor choices of clock speed. Right now Infusion 1.5.2 kernel has some issues with not liking the ondemand governor, so I reduced my number of available clock speeds so it would jump to the top more quickly while using the conservative governor.
Also, did you use either the return to stock, or the ultimate unbrick flash before you flashed your ROM? If it was a used Infuse, I would do one of the above before I start looking anywhere else, then reflash your rom/kernel of choice.
I'm using SetCPU with the OnDemand gov (400/1500), no undervolting. I have a profile set to underclock it to 300/700 on conservative when I'm charging, but that's about it.
I think I'm going to use GTG's unbrick/return to stock and root the thing myself, I figured I'd ask before I did anything drastic. My UI is a ***** to set up every time, and I hate wiping because of it.
Just a quick note, I've had some serious lag using the oc kernel... Like real world usage got worse after oc'ing
Sent from my SAMSUNG-SGH-I997 using XDA App
Although it is counter intuitive to the way governors should work I have seen less lag on Infusion 1.5.2 when I use conservative. The ondemand governor seems to be having some issues with the most recent update. If you flash back to 1.0 Infusion you get a max overclock of 1.4, but it is much more stable IMHO. Also FFC is presently broken in 1.5.2 and will hard freeze the phone (battery pull).
I'm debating flashing back to community kernel, but I am going to wait and see how long it takes my battery to drain because I am using some undervolting. You might try using conservative governor and only enable 100, 400, (800 or 1000), (1200 or 1300) and 1500. Honestly the infuse is my first smartphone, but I've tinkered with desktop hardware the last 15 years. It literally amazes me everytime I pull the battery out of this phone and feel how light it is.
I also used the desktop vis app and yes it lagged when chosing apps however it seemed to get better after three or four times but that seems to be the only place it lags for me im using the oc-uv kernal 1.0 clocked to 1.4 seems bulletproof compared to later versions im getting great speed and batt life with this kernal but i had the same lag in the same event as you
Sent from my SAMSUNG-SGH-I997 using XDA Premium App
I'm going to try bringing it back to stock and re-rooting it myself, if only because I trust myself with adb//scripts/rooting, and I'm hesitant to trust someone I don't know. Lag has gotten a lot better since switching to conservative, OCing to 1.6 and undervolting a tad, but it still lags when opening apps or scrolling through long lists. Desktop Vis is of course, still a nightmare.
Oh well, I'll report my findings tomorrow I guess. Hopefully I'll be scoring higher than the 1700 I just pulled in quadrant.
entropism said:
I'm going to try bringing it back to stock and re-rooting it myself, if only because I trust myself with adb//scripts/rooting, and I'm hesitant to trust someone I don't know. Lag has gotten a lot better since switching to conservative, OCing to 1.6 and undervolting a tad, but it still lags when opening apps or scrolling through long lists. Desktop Vis is of course, still a nightmare.
Oh well, I'll report my findings tomorrow I guess. Hopefully I'll be scoring higher than the 1700 I just pulled in quadrant.
Click to expand...
Click to collapse
Sent from my SAMSUNG-SGH-I997 using XDA App
Do you use spareparts?
usually yes, but I dont believe I installed it yet on the infuse. Why?
lag on stock also
I'm 100% stock & get lag also. Definitely in the situation you describe, when the app list comes up for adding shortcuts to any homescreen. Takes forever for that list to come up. Secondly, simply from running for a while, i have to kill all the apps using the built in task manager in order to get it to run smoothly again. It's like it never closes an app. I'm new to android, but guessing this is a problem inherit to android. Can't see how samsung or at&t could have screwed this soo badly to cause it to never close an app. Hoping memory management is better in Gingerbread.
Blario, you're new to Android, so you need to read up on how memory management works with Android. It's not SUPPOSED to kill apps, and that's not what's lagging your phone. In your case, I'd think it was more of a case of WHICH apps you were running.
yeah, the longer I have the phone the more it lags it seems
probably going to push me towards the Evo3D
If you're on stock and after a week or two your phone gets slow as hell (mine did), install a Voodoo lagfix kernel. All the current ROMs include it.
superweapons said:
If you're on stock and after a week or two your phone gets slow as hell (mine did), install a Voodoo lagfix kernel. All the current ROMs include it.
Click to expand...
Click to collapse
I did
I still have random slow downs when it gets bogged down with stuff
I'd give a left nut to have this phone with dual core
entropism said:
Blario, you're new to Android, so you need to read up on how memory management works with Android. It's not SUPPOSED to kill apps, and that's not what's lagging your phone. In your case, I'd think it was more of a case of WHICH apps you were running.
Click to expand...
Click to collapse
I don't do much. Twitter, "Web" (some mozilla stock ish), SMS, phone, Android Market app (browsing), and Maps. I've had it lag up using only those. Far less often, I use Yelp... that's about it.
Not supposed to kill apps? When I kill them it fixes the problem. Maybe it *should* be killing apps?? If its supposed to flush their state to disk... it seems to be doing a poor job at it.
I would be suspicious of your microSD card (if you have one). Do you know what class it is? Have you tried removing it and performing the same functions in which you see lag?
hydrogenman said:
I did
I still have random slow downs when it gets bogged down with stuff
I'd give a left nut to have this phone with dual core
Click to expand...
Click to collapse
I'd more readily assume the 512MB RAM is the issue. Looking forward to SGSII
SolusCado said:
I would be suspicious of your microSD card (if you have one). Do you know what class it is? Have you tried removing it and performing the same functions in which you see lag?
Click to expand...
Click to collapse
It's the stock card that came with the phone. Any way to check? The apps are installed to it... so removing it I wouldn't have the apps available anymore.
A little update: I returned back to stock and re-rooted. I installed 3E, then CWM, then threw on Infused 1.5 and Infusion, OCed to 1.6Ghz and slightly undervolted (about 25-100mv along the range) using SetCPU.
As of right now, it still lags on occasion, but not as bad as it did. Quadrant is giving me a whopping 1300 average score, with a low of 800 and a high of 1600. This is just pathetic.
any good sensation guides for getting setcpu to work?
thanks
Setcpu should be self explanatory since we now have root. You set the frequency you want the device to operate at. That being said if you are looking to OC then you need a kernel capable of overclocking
but what option do i select when i first start it as the sensation is not listed
I too would like to know.
Sent from my HTC Sensation 4G using XDA Premium App
i think that there soon will be a update. so that we can select our sensation directly. but till than you have to select the tegra2 device.
thank you very much
decided to use this custom freq list that goes upto 1.19ghz so just under the sensations max threshold
put into root of sd and select the custom option
Kurisu87 said:
decided to use this custom freq list that goes upto 1.19ghz so just under the sensations max threshold
put into root of sd and select the custom option
Click to expand...
Click to collapse
Thanks but this is stock, overclocking is making it do more than it should.
Thanks for the file tho, i'll modify that and see what i can get out of it.
Cheers for the building block
Went on with increments of 38400 however, even though the slider will show 1.5 the actual speed is still restricted to 1.18ghz
Bit weird, i've got no experience of creating these set cpu scripts so i really don't know what's the matter with it, i think the next increment needs to be a different number, as before 38400 it was something like 19000 roughly.
dladz said:
Thanks but this is stock, overclocking is making it do more than it should.
Thanks for the file tho, i'll modify that and see what i can get out of it.
Cheers for the building block
Went on with increments of 38400 however, even though the slider will show 1.5 the actual speed is still restricted to 1.18ghz
Bit weird, i've got no experience of creating these set cpu scripts so i really don't know what's the matter with it, i think the next increment needs to be a different number, as before 38400 it was something like 19000 roughly.
Click to expand...
Click to collapse
the speed is restricted by the kernel so you would need a Kernel that supports over clocking.. these should be everywhere shortly now most have s-off, I just used the stock for now to manage power i,e when screen is off, use 1/4 of the cpu (saves battery)
I've been playing with why a lot of people are getting slowdowns with PBJ and I came up with a potential solution. I posted this in the Eclipse forums and it is working for there, so I'm posting it here to a wider audience.
I realized that the difference for me may have been the fact that I wasn't running default LMK minfree values. I don't run V6 because that causes other issues for me, I do up the minfrees a bit because I find that helps. When I removed my minfrees and let it use the stock values, I noticed an immediate difference...I noticed exactly what everyone was talking about.
I use Autokiller Memory Optimizer from the Market to change LMK values. It only changes the minfrees and doesn't do the other stuff from V6 that seems to bog me down. Try the "Strict" setting. They seem to be working well for people. I'm not sure if imnuts has reduced the minfree values on PBJ from stock, but this definitely seems to be working for people.
Interesting I tried this, but I can't download the app. It says it's not compatable with my device
Sent from my SCH-I510 using XDA
I changed the low memory killer settings to those contained in the stock FP1 kernel when I updated it.
Interesting. I'm definitely seeing a major performance improvement by increasing those values a bit. Leaving at default results in things being essentially unusable after a few hours. Bump them up a bit and I'm running indefinitely...I rebooted this morning to install a theme and I'd been running for 5 days without a reboot to that point.
Sent from my SCH-I510 using Tapatalk 2
OK, no wonder og 0130 PBJ is working flawless for me.
Sent from my SCH-I510 using xda premium
shrike1978 said:
Interesting. I'm definitely seeing a major performance improvement by increasing those values a bit. Leaving at default results in things being essentially unusable after a few hours. Bump them up a bit and I'm running indefinitely...I rebooted this morning to install a theme and I'd been running for 5 days without a reboot to that point.
Sent from my SCH-I510 using Tapatalk 2
Click to expand...
Click to collapse
Send me/Post the values you're using for it and I'll change them in the init.rc so everyone can have it without installing an app.
shrike1978 said:
I've been playing with why a lot of people are getting slowdowns with PBJ and I came up with a potential solution. I posted this in the Eclipse forums and it is working for there, so I'm posting it here to a wider audience.
I realized that the difference for me may have been the fact that I wasn't running default LMK minfree values. I don't run V6 because that causes other issues for me, I do up the minfrees a bit because I find that helps. When I removed my minfrees and let it use the stock values, I noticed an immediate difference...I noticed exactly what everyone was talking about.
I use Autokiller Memory Optimizer from the Market to change LMK values. It only changes the minfrees and doesn't do the other stuff from V6 that seems to bog me down. Try the "Strict" setting. They seem to be working well for people. I'm not sure if imnuts has reduced the minfree values on PBJ from stock, but this definitely seems to be working for people.
Click to expand...
Click to collapse
I was hearing of this issue with the key board and it has been happening all day to me on eclipse 2.0 with the 5/4 kernel
Sent from my SCH-I510 using XDA
imnuts said:
Send me/Post the values you're using for it and I'll change them in the init.rc so everyone can have it without installing an app.
Click to expand...
Click to collapse
I've been running 3072, 4608, 6144, 15360, 17920, 20480 in pages (12, 18, 24, 60, 70, 80 in MB). Those last three number are likely a bit higher than they need to be, but I've been loath to mess with what's been working. I'm trying 10240, 12800, 15360 (40, 50, 60) for the last three right now, and I'll report back tomorrow on how it's doing.
shrike1978 said:
I've been running 3072, 4608, 6144, 15360, 17920, 20480 in pages (12, 18, 24, 60, 70, 80 in MB). Those last three number are likely a bit higher than they need to be, but I've been loath to mess with what's been working. I'm trying 10240, 12800, 15360 (40, 50, 60) for the last three right now, and I'll report back tomorrow on how it's doing.
Click to expand...
Click to collapse
The first 3 are the same as they are in stock. I've made the change so that the last 3 split the difference as 40, 50, 60 MB is just over what it is stock, so I set them to 50, 60, 70 MB. I'll rebuild later, likely tomorrow or Wednesday then.
PBJ repack
Did the kernal get repacked with the new values?
I would assume it is not a good idea to use Autokiller Memory Optimizer with Ram Manager Pro installed as they would conflict? I currently have Ram Manager installed and do experience lag occasionally.
I would just like to point out that I had terrible lag experiences with the latest PBJ on stock deodexed FP1. I think I even posted about it here because no matter how much I tweaked the memory/CPU settings, I would still be waiting 15-20 seconds for just about anything to happen.
Now I should also mention that I flashed PBJ without a full wipe.
The other day, I decided to back up all my apps and whatnot and wipe the system, flash stock FP1 and then flash PBJ (still RFS by the way).
Now, my phone runs like it's on speed.
So for those who are having issues with PBJ and upgraded without wiping, maybe give it a shot and see how it works out for you.
Good luck.
Lol.....
Edit : maybe I should be more clear ...
Wiping had nothing to do with the kernel ...
SENT FROM THE DOG POUND
jeremyumo said:
Did the kernal get repacked with the new values?
Click to expand...
Click to collapse
Yes
Droid Charge powered by Tweaked 2.1 TSM ext4 PBJ w/ Transparent ICS
cujo6801 said:
Lol.....
Edit : maybe I should be more clear ...
Wiping had nothing to do with the kernel ...
SENT FROM THE DOG POUND
Click to expand...
Click to collapse
I didn't say it did... point was to illustrate how clearing out the data/cache worked for me. I didn't have this kind of lag on the stock kernel, but I wanted PBJ specifically for Voodoo sound. After I swapped kernels, I was surprised to notice that my phone was basically rendered useless. I mean we're talking literally minutes to complete any operation (even lock/unlock). After a wipe and a reflash of the kernel, all is good.
No need to laugh. I'm not a developer or much of a computer person at all really but I know when something works and I am well versed in the concept of the process of elimination.
That is all.
Just to let everyone know...I changed the minfrees to 12, 18, 24, 46, 62, 78 and my phone has been smooth and virtually lag free
Droid Charge powered by Tweaked 2.1 TSM ext4 PBJ w/ Transparent ICS
hey guys. was wondering what is the best cache size for the internal memory? I believe by default it is 2048. I get an option ranging up to 6144. should I go for the max? Is the improvement significant? and does it affect battery life? Thanks guys!
azreen91 said:
hey guys. was wondering what is the best cache size for the internal memory? I believe by default it is 2048. I get an option ranging up to 6144. should I go for the max? Is the improvement significant? and does it affect battery life? Thanks guys!
Click to expand...
Click to collapse
Why don't you test it yourself amd tell us?
https://play.google.com/store/apps/details?id=ales.veluscek.sdtools
You can go higher than specified on the stock setup of 2048, and you might get an increase in certain scenarios, but keep in mind that the read ahead is mostly effective in its nature of fetching that chunk of data for a purpose. Not just to get filled up with wasted data you aint gonna need anyway. It can be fine for speeding up eg. Browsing your pictures with Quickpic, but you probably waste a good chunk of cachespace that could be more usefull for data you actually need in a fast sequencial read scenario.. Most of what advices for the readahead Ive read from the old dogs on xda points at a 512kb set, as a balance between effectiveness and wasted cache, atleast on phones, but it could be a tad different on our tabs as they are already at 2048 probably for a reason. So digg in and let us know how you fare
Omnius
Sent from my A510 using xda app-developers app
I was wondering why people keep on asking to try it by myself, It doesn't hurt to ask before trying right?
and yes I have tried increasing it and I don't see any improvement with the read speed(since write speed is not going to be affected as been told in a forum topic)
I believe omnius001 was right, it could have been set to 2048 for a reason or two, so increasing won't do better, let's just stick to 2048 then thanks for replying anyway!
azreen91 said:
I was wondering why people keep on asking to try it by myself, It doesn't hurt to ask before trying right?
Click to expand...
Click to collapse
Well, as we say, no guts-no glory It really doesn't hurt anything.
The problem is that all stock Acer tabs suffer from a very poor IO, mostly when it comes to DB writes and other IO functions. Running a simple Quadrant benchmark at standard setting will affirm this. Just look at the IO results. The poorest of any device out there, and poorest by a lot!
Not sure whether this is because of the "noop" scheduler Acer insists on using. Generally, this is an ok scheduler, however BFQ handles things better in my opinion.
Tweaks to improve IO performance have been around for awhile. Generally, as in the Iconia Tab series, it involves some changes in ramdisk, and not the kernel itself. Unless Devs add the BFQ scheduler, which requires a kernel rebuild.
I have a modified kernel (701), unfortunately, never got a chance to test it, before the CWM recovery borked. If I compare to the 501, it should increase IO performance by 400%, more inline with other devices.
Why Acer in 2 years, hasn't improved IO performance when they easily can, well, it escapes logic.
MD
It's been a while since I managed this stuff myself so I'm foggy on how to troubleshoot, but something along the lines of I don't have an /etc/fstab or it's empty so swapon is failing.
previously used FauxClock to manage this, unsure why it's not working like it used to.
thor1k said:
It's been a while since I managed this stuff myself so I'm foggy on how to troubleshoot, but something along the lines of I don't have an /etc/fstab or it's empty so swapon is failing.
previously used FauxClock to manage this, unsure why it's not working like it used to.
Click to expand...
Click to collapse
well, to answer the question in the title.. because its pointless. zram/zswap is good for our old devices(like g1 or old school droid), with low ram. as thats what you are using the zram/zswap for, to increase ram. the nexus 5 has 2gb ram, i bet you will never run low on ram with a nexus 5. and if you think its a benefit to the nexus 5, then you are living in a placebo.
what happened to yours, i have no idea.
simms22 said:
well, to answer the question in the title.. because its pointless. zram/zswap is good for our old devices(like g1 or old school droid), with low ram. as thats what you are using the zram/zswap for, to increase ram. the nexus 5 has 2gb ram, i bet you will never run low on ram with a nexus 5. and if you think its a benefit to the nexus 5, then you are living in a placebo.
what happened to yours, i have no idea.
Click to expand...
Click to collapse
can you help me get it back?
I feel like a FauxClock update broke it.
What it was actually doing was a bit more complicated, virtualizing the diskspace (compressed, too) as a buffer layer on top of the flash. This improved random read and especially random write performance which helped with multitasking a LOT as resume from suspend in the app lifecycle was resuming from RAM, not from flash. This meant less hitching/FPS drops when bringing up android recents, as well as during animation/draws when switching to the other app
This is the only reason I want it. Simply as swap space, yes not necessary
thor1k said:
can you help me get it back?
I feel like a FauxClock update broke it.
What it was actually doing was a bit more complicated, virtualizing the diskspace (compressed, too) as a buffer layer on top of the flash. This improved random read and especially random write performance which helped with multitasking a LOT as resume from suspend in the app lifecycle was resuming from RAM, not from flash. This meant less hitching/FPS drops when bringing up android recents, as well as during animation/draws when switching to the other app
This is the only reason I want it. Simply as swap space, yes not necessary
Click to expand...
Click to collapse
that sounds what disabling fsync would do, help with random read and write. have you tried just disabling fsync? anyways, its your kernel that will have the zram feature, reflash it. if you have flashed/updated your rom, the kernel should be reflashed.
simms22 said:
that sounds what disabling fsync would do, help with random read and write. have you tried just disabling fsync? anyways, its your kernel that will have the zram feature, reflash it. if you have flashed/updated your rom, the kernel should be reflashed.
Click to expand...
Click to collapse
no fsync is totes different
current guess is no init.d script for the zram because restored fauxclock from titanium and enable/disable only modifies the script, relying on the script existing in the first place, and it doesn't because I didn't manually install it myself. Uninstalling and reinstalling fauxclock doesn't work though, it's like certain settings persist somewhere. It's always been a bit goofy like that, and I don't feel like wiping and installing just for the ZRAM at this moment. Currently waiting on someone to post their zram init.d script, or if anyone here sees this and wouldn't mind uploading it, much obliged.
another question: how can I low-level format the phone? IE not just wiping the partition, but actually checking for bad blocks and cutting them out of the flash table. I have another phone that I think needs that, but all the CWM and TWRP recoveries seem to just do wiping not complete formats.
init tasks are somewhere besides /etc/init.d now it looks like
if anyone else needs it, flashing completely stock image straight from google and then rooting got me what I needed. Unsure why the "stock rooted" zips people provided weren't good enough. oh well