Related
Hi I've read some where in this forum that the galaxy S has some I/O problems which are leading to the frequent lags that the phone experiences
Frankly I do not know what I/O is but my question is whether I/O problem is a hardware or software problem. If it is software then fair and well, I'll wait for samsung to ooptimize the software
I have noticed that the benchmark software (Quadrant) runs relatively smoothly all the tests except the I/O test at which it stops for a while before moving to the next test. I dont know if this relates to the I/O problem.
Thanks for your answers
RADLOUNI said:
Hi I've read some where in this forum that the galaxy S has some I/O problems which are leading to the frequent lags that the phone experiences
Frankly I do not know what I/O is but my question is whether I/O problem is a hardware or software problem. If it is software then fair and well, I'll wait for samsung to ooptimize the software
I have noticed that the benchmark software (Quadrant) runs relatively smoothly all the tests except the I/O test at which it stops for a while before moving to the next test. I dont know if this relates to the I/O problem.
Thanks for your answers
Click to expand...
Click to collapse
Theres still no solid proof that its a software issue.
****
EarlZ said:
Theres still no solid proof that its a software issue.
Click to expand...
Click to collapse
****...............then you're saying its hardware ??????
He is not saying its hardware, just that there is no solid proof that its software.
However, based on the amount of working fixes, and reports of great improvements using Froyo I would put money on it being software related.
In fact...I am!
I am ordering a SGS in 3 days when my contract is up for renewal.
Actually, I think everyone's overlooking another obvious possible source of lag: clock-scaling for power conservation. If a phone slows down to 200MHz when it thinks it's inactive, and won't speed up until it sees evidence of activity lasting for 400ms, well... that's 400ms of lag you wouldn't get if the phone were running full-bore 1GHz all the time.
There's even an easy way to test the theory (on a rooted phone, at least) -- take two otherwise-identical phones, fully-charged, root one (while keeping the same rom), then install SetCPU and lock it into 'performance' mode so the phone can't slow down.
If the one locked at 100% CPU speed doesn't lag, and the one that's allowed to slow down to prolong the battery life does... well... there's the answer.
I mention this because I just experienced the night-and-day difference between the CDMA Hero's default power/speed (528MHz max, going down to 250MHz or less when "inactive") and with it locked to 712MHz in performance mode. Pretty much all of my lag problems vanished instantly when I locked it to performance mode. I have a hunch right now that perceived lagginess is almost entirely due to cpu scaling (particularly the time it takes to scale back up, and the criteria used for doing it).
Makes Sense
bitbang3r said:
Actually, I think everyone's overlooking another obvious possible source of lag: clock-scaling for power conservation. If a phone slows down to 200MHz when it thinks it's inactive, and won't speed up until it sees evidence of activity lasting for 400ms, well... that's 400ms of lag you wouldn't get if the phone were running full-bore 1GHz all the time.
There's even an easy way to test the theory (on a rooted phone, at least) -- take two otherwise-identical phones, fully-charged, root one (while keeping the same rom), then install SetCPU and lock it into 'performance' mode so the phone can't slow down.
If the one locked at 100% CPU speed doesn't lag, and the one that's allowed to slow down to prolong the battery life does... well... there's the answer.
I mention this because I just experienced the night-and-day difference between the CDMA Hero's default power/speed (528MHz max, going down to 250MHz or less when "inactive") and with it locked to 712MHz in performance mode. Pretty much all of my lag problems vanished instantly when I locked it to performance mode. I have a hunch right now that perceived lagginess is almost entirely due to cpu scaling (particularly the time it takes to scale back up, and the criteria used for doing it).
Click to expand...
Click to collapse
Very interesting theory, and it makes sense to be frank.
Are there any software out there that would enable me to lock the CPU speed to 1 GHz. I am willing to try this
RADLOUNI said:
Very interesting theory, and it makes sense to be frank.
Are there any software out there that would enable me to lock the CPU speed to 1 GHz. I am willing to try this
Click to expand...
Click to collapse
The guy just told you, root phone and install "Setcpu". That's the only way.
Looking at the benchmarks and the various fixes implemented I tend to lean towards the opinion that it may be hardware related.
hxxp://twitter.com/koush/status/20321413798
I'm not familiar enough with the internals of the phone. If there is faster flash memory located on the phone, then a repartition may be enough to fix the device. If not, then I'm afraid we may be stuck with some lag.
Would anyone be so kind as to explain what I/O is and why the setup in the SGS causes lagging while other android phones with similar specs don't seem to suffer from the same problems?
Thanks in advance
RADLOUNI said:
Very interesting theory, and it makes sense to be frank.
Are there any software out there that would enable me to lock the CPU speed to 1 GHz. I am willing to try this
Click to expand...
Click to collapse
not very interesting, as it's closer to the truth than you think.
think about it, Android OS is really Linux, the SGS is a miniature PC with phone capability.
everything else aside for the phone part, works just like a PC running Linux.
even on a Windows PC the Duo Core or Quad Core machines that has the Power Saving option enable behaves the same, when they are on iddle mode they run at 50% CPU power or less, and it takes them a fraction of a sec to speed back up, people that don't like that tiny lag, they always set the PC on performance mode, or always on, or simply not install the power saving software that comes with the PC.
we can do the same on the SGS phones, the only downside is that your battery will be out of juice faster than you think.
not to mention the Screen is the most power hungry part in the phone, just like most other phones with large LCD displays
did the install and...
Hi
I just did the install of setcpu and i will monitor the device for sometime before i give some feedback. My initiall impression is that the performance got better.
i set the software to performance mode and kept the limits between 100Mhz and 1000Mhz
i will also try to set the min limit to 800Mhz
Actually, that reminds me... the other thing I've seen cited a lot for causing lag is the way Android manages memory by terminating apps instead of using a swapfile. This can cause lag, because it simply takes time to call onPause()/onStop() and wait for it to finish, compared to unceremoniously just suspending the app and dumping a few megs to the microSD card.
Apparently, manufacturers don't use swapfile because most/all Android phones ship with class 2 microSD flash, in which case it would hurt performance more than it helped.
With that in mind, I'd say the following two things should be tried:
1) Buy a class 6 (or better) microSD card, format a swap partition, and use a rom on a rooted phone that supports it. For the record, swap with class 2 would be detrimental; swap with class 4 would be of minimal benefit; swap with class 6 is a big improvement; class 8 basically doesn't exist, and class 10 in real-world use -- with small, scattered files and random read-writes -- is only a little bit faster than class 6, because at that point the time it takes to deal with protocol matters becomes huge relative to the time it takes to actually DO the flash write (the SD card SPI and 4-bit protocols are *really* ugly, and overwhelmingly optimized for sequential reading and writing of bulk data. The moment you start doing random-access rewrites, their performance -- regardless of class -- goes to hell. That's part of the reason why pro gear still tends to use CompactFlash... it still has to deal with flash a page at a time, but it can access arbitrary tiny chunks of data scattered all over the place a lot faster and with a lot less ceremony than (micro)SD).
2) Install SetCPU and lock the CPU to max speed in "performance" mode.
SetCPU alone seemed to make the biggest difference with regard to keyboard input lag. My guess is that Android (or HTC's modifications for the Hero, Evo, etc... and quite possibly Samsung's too) slow the phone WAY down whenever an input area is displayed, on the theory that "most" apps at that point are just displaying the picture of a keyboard and waiting for the user to mash the screen with his finger. Without SetCPU, Graffiti is almost unusable and makes weird errors. With SetCPU locked to performance mode, Graffiti is almost flawless. It's literally a night-and-day difference.
Kpkpkpkp said:
Would anyone be so kind as to explain what I/O is and why the setup in the SGS causes lagging while other android phones with similar specs don't seem to suffer from the same problems?
Thanks in advance
Click to expand...
Click to collapse
Input/Output -data communication to/from the phone and other devices/networks
It's like when you are writing information to the system that comes from downloads, so whether you are syncing files, downloading from the marketplace or uploading...you are doing I/O....
"In computing, input/output, or I/O, refers to the communication between an information processing system (such as a computer), and the outside world possibly a human, or another information processing system. Inputs are the signals or data received by the system, and outputs are the signals or data sent from it. "
Wiki
bitbang3r said:
Actually, that reminds me... the other thing I've seen cited a lot for causing lag is the way Android manages memory by terminating apps instead of using a swapfile. This can cause lag, because it simply takes time to call onPause()/onStop() and wait for it to finish, compared to unceremoniously just suspending the app and dumping a few megs to the microSD card.
Apparently, manufacturers don't use swapfile because most/all Android phones ship with class 2 microSD flash, in which case it would hurt performance more than it helped.
With that in mind, I'd say the following two things should be tried:
1) Buy a class 6 (or better) microSD card, format a swap partition, and use a rom on a rooted phone that supports it. For the record, swap with class 2 would be detrimental; swap with class 4 would be of minimal benefit; swap with class 6 is a big improvement; class 8 basically doesn't exist, and class 10 in real-world use -- with small, scattered files and random read-writes -- is only a little bit faster than class 6, because at that point the time it takes to deal with protocol matters becomes huge relative to the time it takes to actually DO the flash write (the SD card SPI and 4-bit protocols are *really* ugly, and overwhelmingly optimized for sequential reading and writing of bulk data. The moment you start doing random-access rewrites, their performance -- regardless of class -- goes to hell. That's part of the reason why pro gear still tends to use CompactFlash... it still has to deal with flash a page at a time, but it can access arbitrary tiny chunks of data scattered all over the place a lot faster and with a lot less ceremony than (micro)SD).
Click to expand...
Click to collapse
To ask a question and summarize, if you were given a choice of any class card to put in your phone you'd chose a class 6 because of the performance benefit here? Or would you maybe go with a higher class because it'd get greater benefits in other areas? Thanks for the help, again, don't consider money as a factor for the main issue, just trying to learn a bit.
result
bitbang3r said:
Actually, that reminds me... the other thing I've seen cited a lot for causing lag is the way Android manages memory by terminating apps instead of using a swapfile. This can cause lag, because it simply takes time to call onPause()/onStop() and wait for it to finish, compared to unceremoniously just suspending the app and dumping a few megs to the microSD card.
Apparently, manufacturers don't use swapfile because most/all Android phones ship with class 2 microSD flash, in which case it would hurt performance more than it helped.
With that in mind, I'd say the following two things should be tried:
1) Buy a class 6 (or better) microSD card, format a swap partition, and use a rom on a rooted phone that supports it. For the record, swap with class 2 would be detrimental; swap with class 4 would be of minimal benefit; swap with class 6 is a big improvement; class 8 basically doesn't exist, and class 10 in real-world use -- with small, scattered files and random read-writes -- is only a little bit faster than class 6, because at that point the time it takes to deal with protocol matters becomes huge relative to the time it takes to actually DO the flash write (the SD card SPI and 4-bit protocols are *really* ugly, and overwhelmingly optimized for sequential reading and writing of bulk data. The moment you start doing random-access rewrites, their performance -- regardless of class -- goes to hell. That's part of the reason why pro gear still tends to use CompactFlash... it still has to deal with flash a page at a time, but it can access arbitrary tiny chunks of data scattered all over the place a lot faster and with a lot less ceremony than (micro)SD).
2) Install SetCPU and lock the CPU to max speed in "performance" mode.
SetCPU alone seemed to make the biggest difference with regard to keyboard input lag. My guess is that Android (or HTC's modifications for the Hero, Evo, etc... and quite possibly Samsung's too) slow the phone WAY down whenever an input area is displayed, on the theory that "most" apps at that point are just displaying the picture of a keyboard and waiting for the user to mash the screen with his finger. Without SetCPU, Graffiti is almost unusable and makes weird errors. With SetCPU locked to performance mode, Graffiti is almost flawless. It's literally a night-and-day difference.
Click to expand...
Click to collapse
HI I tried setCPU at performance mode , and i have to say that it improved the perofrmance A BIT. But i would not say that much has improved.
I guess that the class6 SD card option has more bearing on this issue than CPU speed scaling
RADLOUNI said:
HI I tried setCPU at performance mode , and i have to say that it improved the perofrmance A BIT. But i would not say that much has improved.
I guess that the class6 SD card option has more bearing on this issue than CPU speed scaling
Click to expand...
Click to collapse
Does the SD card lag fix option REQUIRE a class 6 card? That limits the size a bit, doesn't it? What is the biggest class 6 card available?
borchgrevink said:
Does the SD card lag fix option REQUIRE a class 6 card? That limits the size a bit, doesn't it? What is the biggest class 6 card available?
Click to expand...
Click to collapse
16GB
no, it does not need to be Class 6, it depends on the SD card build quality, some Class 2 performs as good as a Class 6
but it's a luck of the draw, if you have a known good Class 2 or Class 4 microSD card, then use it, no need to buy a new one
i suggest you to test the speed of the SD card before you do the mimocan thing
use this app, it's pretty accurate
http://forum.xda-developers.com/showthread.php?t=739083
The CPU theory doesn't really explain why symlinking the /dbdata/data folder to /data/data eliminates lag.
hxxp://android.modaco.com/content/samsung-galaxy-s-s-modaco-com/312298/got-the-stalling-problem-rooted-try-this/
It also seems that a 32gb class2 SanDisk card is OK.
http://android.modaco.com/content/s...rt-microsd-cards-that-work-with-mimocans-fix/
borchgrevink said:
It also seems that a 32gb class2 SanDisk card is OK.
http://android.modaco.com/content/s...rt-microsd-cards-that-work-with-mimocans-fix/
Click to expand...
Click to collapse
kinda pricey at the moment, aprox $135 for real non fake ones
Take a look at this: http://www. laptopmag.com/review/cell-phones/samsung-epic-4g.aspx?mode=benchmarks&cids=2214,2290,pa
Go to that page and add the other Galaxy S variants to the comparison charts and check the FileSystem benchmarks. As you will see, The Epic 4G beats the other phones 5+ times over, and even destroys the EVO 4G's score by a little over 4 times. This is great news for those of us concerned about the lower ROM capacity!
They are missing a lot on those benchmarks. I can't find the nexus one at all, and the linpack benchmarks all seem to be testing android 2.1 phones. I think what people are afraid of with this phone is that the JIT compiler works amazingly for Snapdragon processors but not for TI OMAP processors or the Samsung Hummingbird cpu. People are claiming extremely high scores in Linpack with afaik is a cpu benchmark utility. Now All of the Cortex A8 1 ghz processors should have the same performance give or take 5-10% variance. However Snapdragon cpus seem to get 30-50 on linpack with 2.2 on the nexus one and evo 4g, but the droid 2 only gets 13-14, and what few beta or alpha builds of 2,2 have leaked for galaxy s phones they seem to get similar scores. This is what worries people. However I think we need to wait and see why the droid 2 is getting such low scores on 2.2 and see if the official smasung 2.2 build gets similar low scores.
Also that site says that the droid x/2 has the same gpu performance as the epic 4g and on top of that the evo 4g scores less than 10% below them. Last I checked the power vr 530 in the x/2 is half as powerful or less than the 540 in the galaxy s phones and the amd gpu in the snapdragon processors is supposed to be even weaker than the 530. Given this knowledge I have to say those benchmarks are fairly unprofessional.unscientific and useless as they are clearly biased.
There's no denying the inaccuracies in the majority of benchmark tests, but the only one I want you to see is the FileSystem benchmarks. Not the cpu or gpu ones. FileSystem measures how fast the storage/memory on the phone is writable and readable. The Galaxy S phones have an issue where out the box all the apps are stored on an internal class 2 SD card, instead of the faster internal NAND memory causing severe lag and even stalls when transitioning between screens. That has been fixed through a hack though, which requires reformatting the file system and 1.5 GB of internal memory which the Epic doesn't have, making a fix impossible for the time being if indeed there is a lag/stall issue on the Epic 4G.
The benchmark for FileSystem, if accuratee, is a sigh of relief, cause it seems like there will be no lagging or stalling between transitions out the box with the Epic 4G. Let's hope This is the case, otherwise it will be a huge deal breaker. The Vibrant can take up to 10 seconds to get back to the home screen from an app when the home icon is pressed. That would really kill the experience if it's present in the Epic 4G and unable to be fixed due to low ROM.
I don't put much stock in benchmarks, but I believe he was trying to point out just 1 benchmark in particular that shows that the Epic doesn't suffer from the flaw in the Captivate\Vibrant that causes the slowness. It's interesting that the benchmarks didn't include Quadrant... which the Captivate\Vibrant somehow destroy all over phones in without 2.2 post lag fix. I'd be interested to see how the Epic performs out of the box there... but I can wait.
My bad I was just tired and didn't realize it was the filesystem speeds that were being pointed out. Though I still question this sites benchmarks and would love to know what software they used for each one.
Does anyone find it odd that the droid 2 and droid x have different filesystem benchmark speeds? I would have assumed they had the same storage and file system. However the droid 2 on 2.2 is showing speeds considerably slower than the droid x in benchmarks for that. It just makes me wonder if it is 2.1 and 2.2 just being wonky, or do these benchmarks possibly have little to no meaning.
Nice find. This test has to be running from the internal memory since it scores the same as the droid x with 8 gig internal storage. The epic has less then 500 mb of internal storage so the sd card has to come into play after the internal storage is full. The issue with that is the sd card runs 2 to 3 times slower then the internal memory. My class 6 sd card with ext2 partition scores 1000 and my captivates internal storage with ext2 partition scores 2600. This may never be a issue if you only install limited amount of apps to internal storage and music, video, games and pics to sd card. This is the only thing holding me back from the epic right now since I have a lot of games that I want on the internal storage. I will have to test and see if it even makes a real world difference. If it doesnt make a real world difference then the epic will be my new phone for sure if the gps issue is really fixed.
My sources tell me, the quadrant score was 889 on the epic 4g
noobnl said:
My sources tell me, the quadrant score was 889 on the epic 4g
Click to expand...
Click to collapse
Hmm thats weird. That means the ****ty partitioning that causes the lag is not gone. The captivate gets 900+ with no lag fix. It gets over 2500 with ext2 internal partition lag fix applied. It doesn't really matter for the captivate since there is internal storage ext2 lag fix but that wont work on the epic since it does not have enough internal storage. They will have to use the slower sd card lag fix. I really dont think they changed the way samsung partitions the storage just for the epic but we will see soon enough.
or maybe they are adding RFS R/W buffer in their 2.6.32 kernel for FroYo
So this means if there is a "lag" or whatever, for those unfamilar with all the tech stuff..I'll be stuck with a broken phone? That sucks..I'm really psyched but I don't want to be stuck with the epic if either the vibrant or captivate would be better because they have enough memory to fix the lag issue..or would just simply switching roms fix it?
tpma4life said:
So this means if there is a "lag" or whatever, for those unfamilar with all the tech stuff..I'll be stuck with a broken phone? That sucks..I'm really psyched but I don't want to be stuck with the epic if either the vibrant or captivate would be better because they have enough memory to fix the lag issue..or would just simply switching roms fix it?
Click to expand...
Click to collapse
The whole point of this thread is to say that the Epic does not suffer from this problem. Go watch the myriad of videos showing the Epic which proves this.
hydralisk said:
The whole point of this thread is to say that the Epic does not suffer from this problem. Go watch the myriad of videos showing the Epic which proves this.
Click to expand...
Click to collapse
Exactly! Its the not the filesystem at all thats causing the lag, its the fact that the other galaxy s phones are using slow sd cards as their internal memory and the Epic is using much faster NAND memory. File system is irrelevant here....
http://briefmobile.com/samsung-epic-4g-benchmarked
"Conclusions
The Samsung Epic 4G is the fastest Galaxy S phone to hit the market. With Samsung’s latest optimizations, this phone flies. We’d venture to say it is faster than the Google Nexus One, Motorola Droid, Motorola Droid X, or any other competing Android 2.1-sporting device. With Android 2.2 Froyo coming soon this year, this phone will undoubtedly rock the mobile world.
While other Galaxy S phones have forced users to place custom launchers (namely LauncherPro) over TouchWiz, we can say without a doubt that the Epic 4G’s Touchwiz UI runs smooth as butter. So, if you like what Samsung’s done to the user experience, you’ll be able to keep the Touchwiz interface without losing any speed advantage.
This phone won’t need a custom Ext2, Ext3, or Ext4 lag fix come launch time, it is snappy and ready to go right out of the box. Overall, Samsung gets a 10/10 for getting it right with this one. Well done."
The Epic 4g was scoring in the 900's out of the box (picture shows 960) on quadrant. I don't think we'll have lag problems.
From experience reading this board, if i was in the tablet making business, I would make one with like 4GB RAM, an unlock[ed | able] bootloader and crap battery life, just to see how many people here at XDA would buy the darn thing and try to make a ROM with battery battery life and OC.
The most verbose people here about the "Eww, only 1GB" types have probably moved on to primes and are demanding TF700T's or whatever is next.
In most cases, we're likely just better off screwing with Dalvik and tuning our TF101's ROMs. It's still a VM and Java involved after all, not assembly code running natively to start with.
Sent from my Transformer Prime TF201 using Tapatalk; and my TF101 is half a foot away!
I've proposed many times that asus should come out with a TF model called Asus Transformer Whiner's Edition TFwhine. In it, there will be 4gig of ram, 100gig of sdcard space, 30hr of bat life, paper thin, water proof, fingerprint proof, sticky as hell (for the butter fingers), and foldable full size PC keyboard. But I'm sure the whiners will find some other thing to whine about.
To me, 1gig of ram is enough. I've never had any issue with running out of ram. I'm serious. This has never come up for me. I mean, what the hell are people running that requires more than a gig of ram?
More RAM would be nice, but I'm more excited about Windows 8 and the new 'ultrabooks' that we'll see in the near future.
letsgophillyingeneral said:
More RAM would be nice, but I'm more excited about Windows 8 and the new 'ultrabooks' that we'll see in the near future.
Click to expand...
Click to collapse
Can you give us a few examples of when you couldn't do something because it only has 1gig of ram?
goodintentions said:
To me, 1gig of ram is enough. I've never had any issue with running out of ram. I'm serious. This has never come up for me. I mean, what the hell are people running that requires more than a gig of ram?
Click to expand...
Click to collapse
I have 2 other tablets and 2 phones with only 512mb of ram, even that seems to be enough for Android. IMO, anything over 1gb would be overkill.
goodintentions said:
Can you give us a few examples of when you couldn't do something because it only has 1gig of ram?
Click to expand...
Click to collapse
Sorry, I thought I was still on the Charge forums. The Charge only has half a gig.
just lou said:
I have 2 other tablets and 2 phones with only 512mb of ram, even that seems to be enough for Android. IMO, anything over 1gb would be overkill.
Click to expand...
Click to collapse
+1
Seriously!! What the hell are people doing that 1GB isn't enough. We used to run (some people still do) Microsoft Windows computers with less than a gig of RAM. It's ridiculous to even sit there and think, that you need more RAM on a phone or tablet.
The browser kinda eats up lots of RAM (if I have around 3-4 pages open, as usual, it eats around 400MB!!), and basically, just the GPU eats off 120MB of all the thing.
FYI, a fast teardown:
- 1024MB RAM
- From that, 728MB is available mostly (some kernels push it to 768MB tops)
- Booting the system, with all the Asus crap loads approx 220-240MB into RAM, including all the things (Launcher, settings, applications autostarting, widgets). And that's a pretty good ratio.
- Basically you now got less than 500MB RAM. If you have tons of widgets, that can reduce free RAM to 400MB. And we're at the same point as smartphones - except that our devices have a lot bigger display, with a lot bigger resolution (2.666666 times more pixels), what means, we need bigger graphics too, what increases application RAM occupation.
Conclusion: With the current setup, we can run less applications on a 1GB RAM tablet than on a 512MB RAM phone (assuming it's a mid-end device with HDPI WVGA display, like, a Galaxy S). That's why we need more RAM, if you increase it to 2GB, you get approx 1.6GB free RAM, 1.3 after system start, and with many apps open, you'll get around 200MB free.
Guess what? That means you're running to much lol. I've got about 450MB if one trusts the settings app; Google and Exchange services are eating at least 20 megs combined! Never mind crap that I don't actually use burning some too. Devices have limits, and 1GB seems to hold plenty until content creation enters the mix, and doing that on Android has more pressing limits in software and users than RAM available. Besides, 2GB tablets will more likely result in fatter apps to fill it than more space to run most apps. Just take a look at Minecraft PC.
My old development system had 1GB of RAM running FreeBSD, if you try to combine GCC, pidgin x 6 logins, Firefox x 20+ tabs, and streaming music with Flash on that laptop, it would all but hit a HCF. I could either compile or have a smooth web, or buy a more powerful laptop. I didn't need one because I could still fit everythig well enough inti its constraints, I wasn't about to start re-encodig lord of the rings whileni waited either.
goodintentions said:
I've proposed many times that asus should come out with a TF model called Asus Transformer Whiner's Edition TFwhine. In it, there will be 4gig of ram, 100gig of sdcard space, 30hr of bat life, paper thin, water proof, fingerprint proof, sticky as hell (for the butter fingers), and foldable full size PC keyboard. But I'm sure the whiners will find some other thing to whine about.
Click to expand...
Click to collapse
Nah, instead of a physical keyboard built in, it ought to hover on it's own and type by mental uplink, or project one in 3D if we can't do that one yet. That way ASUS can still make money off the dock .
Sent from my Transformer Prime TF201 using Tapatalk
Actually, wasn't there supposed to be development from an MIT project, where it was a device that projected keyboards and "touch screens" on whatever surface you aim it at? It was like one of those projector things you always see in exhibitions and fancy places, where the projector interacts with the things blocking the projections, but more cellphone-like and computer-like.
Also, who could argue with more RAM? I think the thing that holds that idea is back is tablet space. Just wait for the next generation RAM/memory structures. STT-RAM, it will be a universal RAM-Memory! Non-volatile RAM ftw!
goodintentions said:
Can you give us a few examples of when you couldn't do something because it only has 1gig of ram?
Click to expand...
Click to collapse
Ubuntu TF101 = Not enough RAM lol
Android = More than enough RAM for now atleast
Sent from my tf Enigmatic V2 beta 1.65Ghz Panda.test cust kernel settings
And ultrabooks are irrelevant to this discussion. You can take your ultrabook and Windows 8 but that's at the point where it's not even remotely comparable to a tablet, much less an Android tablet like the TF101 or TF201 due to the hardware costs. The licensing cost of Windows 8 alone is going to increase the price significantly.
1GB of RAM is more than enough, this is meant to be an enjoyable, lightweight consumer tablet. Not a supercomputing cluster on the go. If you want more RAM you're on the wrong device. Go get an ultrabook or a laptop because you're attempting to cram your needs into 1GB of memory, not designed for your needs.
My 2 cents it is that the memory requirement is a function of Apps. Right now most tablet's are running apps with limited functionality. But as the processors get better, more and more traditional desktop and notebook apps will migrate to tablets. This will be especially true for people that have tablets, like the Transformer, that blur the line between a tablet and a notebook. One can imagine something like Photoshop CS5 running on a tablet and easily requiring 4 GB or more of RAM.
jerrykur said:
My 2 cents it is that the memory requirement is a function of Apps. Right now most tablet's are running apps with limited functionality. But as the processors get better, more and more traditional desktop and notebook apps will migrate to tablets. This will be especially true for people that have tablets, like the Transformer, that blur the line between a tablet and a notebook. One can imagine something like Photoshop CS5 running on a tablet and easily requiring 4 GB or more of RAM.
Click to expand...
Click to collapse
Oh, absolutely. I am sure that in due time we will have tablets with like 20GB of ram. But for now, with the software we have now and the foreseeable future, anything more than 1GB of ram is just stupid.
But that's a completely misguided notion of where apps and tablets are headed. Memory is not a function of apps, at least not linearly or continuously. I see memory hitting a ceiling in terms of heat, space and cost. I see apps not being developed with serious work in mind. You cited photoshop as an example and I think that's just bizarre. This is a tablet. There's this misguided notion that all mobile devices have to converge to the desktop in terms of performance and usage and I think it's bizarre and insane. When you cram a desktop into a tablet what is that? Why even bother when you can go for a laptop that would probably cost half in terms of price and multiples in terms of comfort? Why would any company bother doing this? Why would Google even push Android in this direction?
It's a tablet. You have fun, maybe do typing and word processing at the maximum and in special use cases. But it's meant to be lightweight, enjoyable and made for ENTERTAINMENT. User input at a minimum, simple touch movements, big on audio and visuals. Look at the screen to read a book, watch a video, listen to music. Pull your finger around or tilt the tablet. Not sit there and furiously work on a Photoshop project. That's what a mouse, a full sized keyboard and a desktop computer were made for.
When you talk about a future where RAM increases to meet the needs of apps, Photoshop is on the Android market and CPU power similarly increases to meet the needs. It's no longer a tablet, it's a laptop without a keyboard. It will be priced ugly and its design and specs will match that ugliness.
Maybe in the future we will have hardware so slim and so energy efficient and an input system so perfect all computers will be touch tablets, wafer thin, bendable and capable of running as fast as the U.S. govs fastest superclusters and everyone can download a genome sequencer app that lets you sequence the human genome on every tablet redundantly for ****s and giggles and within two seconds but I highly doubt this future will arrive within our lifetimes. Any attempt to head there by cramming a desktop ideal into a tablet form factor is misguided IMO.
Tubular said:
But that's a completely misguided notion of where apps and tablets are headed. Memory is not a function of apps, at least not linearly or continuously. I see memory hitting a ceiling in terms of heat, space and cost. I see apps not being developed with serious work in mind. You cited photoshop as an example and I think that's just bizarre. This is a tablet. There's this misguided notion that all mobile devices have to converge to the desktop in terms of performance and usage and I think it's bizarre and insane. When you cram a desktop into a tablet what is that? Why even bother when you can go for a laptop that would probably cost half in terms of price and multiples in terms of comfort? Why would any company bother doing this? Why would Google even push Android in this direction?
It's a tablet. You have fun, maybe do typing and word processing at the maximum and in special use cases. But it's meant to be lightweight, enjoyable and made for ENTERTAINMENT. User input at a minimum, simple touch movements, big on audio and visuals. Look at the screen to read a book, watch a video, listen to music. Pull your finger around or tilt the tablet. Not sit there and furiously work on a Photoshop project. That's what a mouse, a full sized keyboard and a desktop computer were made for.
When you talk about a future where RAM increases to meet the needs of apps, Photoshop is on the Android market and CPU power similarly increases to meet the needs. It's no longer a tablet, it's a laptop without a keyboard. It will be priced ugly and its design and specs will match that ugliness.
Maybe in the future we will have hardware so slim and so energy efficient and an input system so perfect all computers will be touch tablets, wafer thin, bendable and capable of running as fast as the U.S. govs fastest superclusters and everyone can download a genome sequencer app that lets you sequence the human genome on every tablet redundantly for ****s and giggles and within two seconds but I highly doubt this future will arrive within our lifetimes. Any attempt to head there by cramming a desktop ideal into a tablet form factor is misguided IMO.
Click to expand...
Click to collapse
You should google graphene. That's where we're heading. Graphene is so strong and so conductive that electronic parts as thin as 1 atom thick could be made. We're talking ultra thin electronic devices in the near future.
I don't think 1GB RAM is enough for Honeycomb, I regularly find my free RAM dropping to double figures <70MB and the system requiring a reboot.
Even using a task killer over time will not free up the same amount of RAM as was available after the previous reboot.
Honeycomb memory management needs a massive amount of work before it can be left to control RAM on its own, I think 1.5-2GB RAM would be a much better solution
In my opinion, more RAM on Android devices is just to keep more applications in memory for multitasking.
I couldn't imagine that a 256mb game, or even a 2mb app, would benefit from a device having more ram than the size of the app.
Sent from my DROIDX using XDA App
goodintentions said:
You should google graphene. That's where we're heading. Graphene is so strong and so conductive that electronic parts as thin as 1 atom thick could be made. We're talking ultra thin electronic devices in the near future.
Click to expand...
Click to collapse
http://arstechnica.com/science/news...hell-holds-the-secret-to-bendable-screens.ars
Here's another interesting one.
infazzdar said:
In my opinion, more RAM on Android devices is just to keep more applications in memory for multitasking.
I couldn't imagine that a 256mb game, or even a 2mb app, would benefit from a device having more ram than the size of the app.
Sent from my DROIDX using XDA App
Click to expand...
Click to collapse
The size of an app has nothing to do with how much RAM it consumes.
A prime example is firefox on a PC... around 17MB app, can use GB's of RAM due to memory leaks, and even without leaks, would still use a couple hundred MB
Viruses are another example, designed to use up as much system resources as possible while only being KB's in size
anybody else getting results like this? What can I do to improve on this. The tablet is not rooted.
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
Third one down is my result.
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
I didn't benchmark but my tf300 feels faster than my hox, a fair bit actually....
I'm not at all disappointed (and it will only get better)
Edit: I've only done the build.prop modifications in the thread mentioned below.
I have never been one to hover over benchmark scores. But for not running optimized custom roms, the only really thing you can try is this http://forum.xda-developers.com/showthread.php?t=1641219
dilfatx said:
anybody else getting results like this? What can I do to improve on this. The tablet is not rooted.
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
Click to expand...
Click to collapse
You. No.
Here are some benches after rooting and clocking to 1.5.
I also watched a video from Richie's closet on YouTube the Asus infinity tf700 vs ipad 3. So I tried 2 of the benchmark tests they used.
First one mine beat the ipad 3 by 15 to 20,000 but the tf700 got like 40,000 more.
Second one beat both ipad 3 by triple the score and beat the tf700 by 200.
So I am happy with the tf300 performance.
http://m.youtube.com/index?desktop_uri=/&gl=US#/watch?v=4jdbtiNnZzE
Quadrant is a bad benchmark, can be influenced easily an score is heavily dependand on IO performance.
Hence the relatively bad scores for the transformers.... internal memory performance is very bad!
I tried overclocking with antutu cpu but I wasn't getting any performance gains when I benchmarked.
I respecfully suggest the quadrant scores do correlate. Over the past year I have had the following tabs and here is the rank in actual performance, using MAME Reloaded, FPse and N64 to test ram speed and cpu power (GPU as well for FPSe and N64). The HTC View is an apparent exception to the ranking from a single and dual core standard.
Thrive
View (Thrive benches better for some benchmarks, but actual View usage seems smoother- 1ghz vs 1.5ghz).
A510/TF300 Basically a tie (within 100 points).
Prime
Excite
Excite benchmarks about 13% faster than the A510/TF300 and the FPS of the emulators (on average) correlate. That said, ignorance is bliss, so if you do not actually sit with them and nitpick, they are all functionally the same for performance- IMO.
One thing I did notice is in spite of the View being a single core S4, it's 1.5ghz speed and memory channel appears better when downloading apps at the same time than the Tegra 2 and 3. All three SoCs have single channel memory, so none will win multi task battles. BTW, the memory channel (edit) a constraint/bottleneck most users experience when trying to download from the market or other sources while trying to run certain apps that are also using it.
Added: The 28nm S4 dual that is in the newer phones actually blows our Tegra 3 away for performance. In part due to the dual channel memory. I would prefer that SoC in the TF300 since also even better battery life and runs cooler (both due to the 28nm).
Sadly, it will be a while before we see many tablets with it since demand for then new phones from HTC and especially Samsung are killing supply until first quarter next year.
rushless said:
I respecfully suggest the quadrant scores do correlate. Over the past year I have had the following tabs and here is the rank in actual performance, using MAME Reloaded, FPse and N64 to test ram speed and cpu power (GPU as well for FPSe and N64). The HTC View is an apparent exception to the ranking from a single and dual core standard.
Thrive
View (Thrive benches better for some benchmarks, but actual View usage seems smoother- 1ghz vs 1.5ghz).
A510/TF300 Basically a tie (within 100 points).
Prime
Excite
Excite benchmarks about 13% faster than the A510/TF300 and the FPS of the emulators (on average) correlate. That said, ignorance is bliss, so if you do not actually sit with them and nitpick, they are all functionally the same for performance- IMO.
One thing I did notice is in spite of the View being a single core S4, it's 1.5ghz speed and memory channel appears better when downloading apps at the same time than the Tegra 2 and 3. All three SoCs have single channel memory, so none will win multi task battles. BTW, the memory channel is the constraint/bottleneck most users experience when trying to download from the market or other sources while trying to run certain apps that are also using it.
Click to expand...
Click to collapse
No its not the memory channel that's the bottleneck, it's not RAM that performs bad it's the internal storage.
I dont know what tegra2 device you tested but the Transformer1 performs aprox the same as the tf201\tf300 IO-wise (bad).
Sent from my Transformer Prime TF201 using Tapatalk 2
Tempie007 said:
No its not the memory channel that's the bottleneck, it's not RAM that performs bad it's the internal storage.
I dont know what tegra2 device you tested but the Transformer1 performs aprox the same as the tf201\tf300 IO-wise (bad).
Sent from my Transformer Prime TF201 using Tapatalk 2
Click to expand...
Click to collapse
Very true on the flash being a part of the factor! Since a series process (especially with single channel) you are only as good as your weakest link. The only Tegra 2 tabs I have had are the A500 (only a few days) and the Thrive.
I find all the tablets noted above to have some issues with downloading from market and using other apps at same time to be near as bad. Again, the HTC View seems better and probably due to your point.
yep
My score was around 10k... but I'm waiting for some Genius to OC the kernel to blow those scores out of the water... i read in the Dev thread someone is using Faux kernel... (or asked him for help) either way I'm waiting for a Legit ROM with an even more legit OC Kernel
I think 10k is not bad at all for me considering that im using stock ROM. Plus, it really feels fast.
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
omidz said:
I think 10k is not bad at all for me considering that im using stock ROM. Plus, it really feels fast.
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
Click to expand...
Click to collapse
How do you get to 10k using the stock rom? Anything else than the build.prop?
I always match transformer prime or a bit more on antutu benchnark.. also this is not 1.5ghz on performance mode. Every single app atm missreads the tablet for 1.5 when it is 1.2. I have tried overclocking and when set to 1.5 it will score same as set to 1.2.
Dont forget to set performance mode before benchmarking and be on a fresh boot
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
I just read this http://androidandme.com/2013/06/opi...-from-the-best-to-worst-tablet-ive-ever-owned
I was wondering about that as well; aside from continued use, the lags are more evident when it reaches the 9gb mark.
I have a 32 GB ram Nexus 7 and seriously; out of the 32 GB, we actually only have 27.5 GB storage.
I usually need more than 9GB of storage free space so i can have a buttery smooth performance.
Estimate 9+ gb to 10GB which means i have 17GB left to work with.
So far, I only discovered 2 worthwhile apps in battling this Nexus 7 predicament for non root users.
Clean master and Forever Gone.
Clean master is a great app for clearing files and releasing load from the 1 GB Ram.
While forever gone seems to work better than restore factory settimgs., it is somewhat slow and inconvenient.
Anyway, is there any other way to have better performance out of the Nexus 7?
Up to how many GB of free space you leave off before it starts to lag?
There was a discussion about this on AAA last week, but no real conclusions as to what might be causing it. One thing they mentioned was the possibility of the flash based storage "giving out". One guy commented that there are no real garbage collection routines present, so without a lot of free space, storage must be manually erased before being written to, and that can take a lot of time. This is why SSD's have garbage collection or trim routines running at idle to erase unallocated blocks.
It sounds plausible as our tablets are either being used or sleeping, with no real idle time for garbage collection to occur. Of course this is all conjecture at this point. I certainly see more lag in my upcoming 1 year old N7, but nothing like the unresponsiveness originally reported by folks filling up their storage, nor the symptoms described in the article you referenced.
But when mine goes belly up, I plan to move to an iPad. The tablet experience on Android pretty much sucks IMHO, and in the better part of the year in which I have been using it, it hasn't got much better. Of the hundred + apps I have on it, I can count on my fingers how many of them are actually optimized for a tablet.
http://forum.xda-developers.com/showthread.php?t=1971852