Being quite happy overall with the prime I couldn't help feeling certain things are very slow, mainly IO actions. Restoring titanium backups takes ages for example.
So I decided to do some extensive benchmarking using the i9100 and TF201 (for now, will add the TF101 later as well) to see how exactly the prime performs. I used the following benchmarks and ran them 3 times each (TF201 in performance mode, CFQ scheduler):
RL Benchmark - SQLite performnace
Androbench
CF-Bench v1.1
Antutu Benchmark
Benchmark (from softweg)
And I'm mostly blown away by the pathetic sqlite performance (and related, the disk performance). See screenshot for a summary (the .xls contains all the results).
1000 inserts taking 68 secs on the tf201 vs 0.6 secs on the i9100!!!
drop table taking 2+ secs on the tf201 vs 0.1 on the i9100
seq. read: ~20mb/s on the TF201 vs 52+ on the i9100
sqlite (updates and deletes): 18-ish secs on the TF201 vs 6 ish on the i9100.
Is the disk of the TF201 such utter crap? Seems to be... Or are there other ways of improving this. Any other results confirming this?
Looking into possibilities of disabling fsync() btw (http://forum.xda-developers.com/showthread.php?t=903507) but nothing really yet.
I think it's a well established fact that Primes have a poor disk performance. Is there a solution, though? I don't know.
roccodaone said:
I think it's a well established fact that Primes have a poor disk performance. Is there a solution, though? I don't know.
Click to expand...
Click to collapse
I firmly firmly believe it is software related. Even if Asus put in the worst flash possible it would not perform but maybe 1/2 to 1/5 as fast. Only time will tell, once Asus/Nvidia has time to focus on things other than critical issues (sleep of death, gps, ect ect ect) maybe disk performance can get hashed out. I think even if Asus did put crappy flash memory in there, they would be smart enough to only go so bad to where it at least contended with some of the competitors. They have been catering to the enthusiast crowd for quite some time and should understand by now that EVERY thing will be benched by its buyers LOL. Sometimes I feel like the EEEpad line is a completely different company than Asus its self though.
Odd thing is my droid razr maxx SMOKES my prime in androbench but does about 10% worse on antutu.
Sent from my EEEPAD using Tapatalk 2
I certainly hope it's software related. But what makes me skeptical about this is, how come it hasn't been resolved by Asus yet.
Disk performance doesn't change in CM9 or AOKP roms, so I don't think it's software related.
If it is not hardware it will most likely be to do with firmware/drivers.
Im sorry guys. I didnt mean software rom wise i mean firmware for the controller. What i find to be odd is my prime wipes and flashes waaay quicker than my razr maxx.
Also like i said in antutu i score higher which is odd.....
We will see though just seems off for some reeeason.
Sent from my DROID RAZR using Tapatalk 2
Is sqlite installed ? I can't find it in documented Android locations and sqlite3 entered as a terminal command has no effect. If it isn't installed then Java "sql" must be being used and it has significantly worse performance.
peterk-1 said:
Is sqlite installed ? I can't find it in documented Android locations and sqlite3 entered as a terminal command has no effect. If it isn't installed then Java "sql" must be being used and it has significantly worse performance.
Click to expand...
Click to collapse
Hmmm, good one... didn't see it in the usual place. There are some libsqlite.so files and stuff but not the sqlite3 in the xbin folder like there is on the i9100 for example.
Might just try and see what happens if I copy it from the i9100 heh. Tried copying the libsqlite.so and libsqlite_jni.so files from the i9100 to the TF201 but didn't improve anything (was actually surprised I didn;t get errors everywhere, but ok).
Will check it out some more later...
peterk-1 said:
Is sqlite installed ? I can't find it in documented Android locations and sqlite3 entered as a terminal command has no effect. If it isn't installed then Java "sql" must be being used and it has significantly worse performance.
Click to expand...
Click to collapse
You might be onto something here.
Sent from my HTC Sensation Z710e using xda premium
Here is a thread that that could be properly used for the I/O discussion clogging up the "so what about the upgrade"
This tablet is a joke...
RL Benchmark SQLLite:
1.0GHz (Bal) TFP - 124.9 seconds
1.0GHz (Perf) TFP - 120.9 seconds
1.2GHz (Bal) TFP - 112.8 seconds
1.3GHz (Perf) TFP - 116.2 seconds (slower???)
1.6GHz (Perf) TFP - 117.4 seconds (slower again???)
1.0GHz Desire HD - 93.4 seconds
1.5GHz Desire HD - 82.8 seconds
Can someone explain why SQLite is so important to my everyday use of the Prime? How will these poor benchmark results manifest in how we use the tablet?
clenz said:
Can someone explain why SQLite is so important to my everyday use of the Prime? How will these poor benchmark results manifest in how we use the tablet?
Click to expand...
Click to collapse
SQLite is the way apps store data and retrieve it in android, it's a database system/layer and used just about all the time. So if this is messed, you'll notice.
Here are my results on RL Benchmark. (On .28, performance mode)
EDIT: Sorry for the low res pics, here are better ones.
Sent from my Transformer Prime TF201 using XDA Premium HD app.
Is it me or is the .28 firmware getting even worst results than before.
Does any have the same graph .21 or S2 or S3 or better yet a One X?
EDIT I see yumms posted a graph for his Prime.
SteveG12543 said:
Here are my results on RL Benchmark. (On .28, performance mode)
Sent from my Transformer Prime TF201 using XDA Premium HD app.
Click to expand...
Click to collapse
Now I have lost all hope!
I am sure a HTC ChaCha is faster than that.
SGS3 first try overall 20.762. There seems to be some serious IO bug.
http://db.tt/TFYUAdjk
leppie said:
Now I have lost all hope!
I am sure a HTC ChaCha is faster than that.
Click to expand...
Click to collapse
As Demandarin said,
demandarin said:
Remember this also. As far as I know, asus/nvidia never really got to address i/o issues yet. wasn't at top of list of things to take care of. most complaints publicly came from gps n wifi. then sod, reboots, lockups, etc.. They've been busy addressing other issues first. they will address i/o soon.
Click to expand...
Click to collapse
Hopefully he's right and they can fix it!
Sent from my ADR6425LVW using XDA Premium.
odd... i am getting way way better sqlite performance than any I have seen recently; on locked TFP .21; fresh from factory wipe. On balanced power.
Related
I know I'm not comparing apples to apples, but my dual core Epic 4G Touch (Galaxy S2 Phone) running Gingerbread feels much more responsive / less lag prone than the quad core Transformer Prime running ICS.
Sometimes in the browser and during any semi to heavy I/O operations such as installing an app from the market the entire Prime UI seems to freeze up until it finishes that particular I/O task. It seems like the operation is synchronous. Unlike my Epic 4G Touch, where when an I/O task occurs the UI/OS are still responsive. This effects being able to do simple things like scroll a web page while it's loading. Or do any muti tasking while an I/O operation occurs in the background of any significance. Maybe this can be tweaked by custom ROM/drivers by people smarter than me. Any thoughts from anyone else on this?
Edit: Some intrepid fellow users have found out that turning on the "completely fair queue scheduler" seems to fix the I/O lock up issues, the fix is described here (must be rooted):
http://forum.xda-developers.com/showthread.php?t=1461583&page=4
i agree with that. i have the same lockup issues.i don't have an answer but I would like to second that. also i would like to add that a lot of the time when i watch youtube videos in hd it buffers a lot. i have a wireless n router, 40/5Mb connection, 3 computers and 2 iphones and none of the videos buffer like they do on here.
so on top of his question i would like to add my question to see if anyone else is having those youtube issues.
Yep
I ran a benchmark after experiencing high loading times on some games and HD videos...
8 MB/s read
7 mb/s write
Didn't want to believe it, made a factory reset and tested again. Speed was the same.
Maybe it's getting faster with ICS, I didn't get the update yet.
I just got an Epic 4g and I also noticed that it is faster than ATP is. I ran a quadrant benchmark and the Epic 4g get 3440 and the ATP get 2420. Before ICS, the ATP got a much higher number.
I'm glad to say I haven't experienced any lag on my TP. Got the skyrim live wallpaper loaded, downloaded loads of apps when I got it and was using the tablet browsing through the UI admiring how slick it is even with live wallpaper. Might seem obvious but everytime the prime turns on from a power off its in balanced mode, so make sure you change to performance mode
Sent from my Transformer Prime TF201 using xda premium
vincepg13 said:
I'm glad to say I haven't experienced any lag on my TP. Got the skyrim live wallpaper loaded, downloaded loads of apps when I got it and was using the tablet browsing through the UI admiring how slick it is even with live wallpaper. Might seem obvious but everytime the prime turns on from a power off its in balanced mode, so make sure you change to performance mode
Sent from my Transformer Prime TF201 using xda premium
Click to expand...
Click to collapse
What mode its in doesn't really matter as far as UI n lag goes. I don't experience this lag mentioned here. I have put my prime in powersavings mode n UI n everything was still very fast, fluid, n smooth. Surprised me actually. Balanced mode is Prime default mode. So after a reboot it'll revert back to it as its the best of both worlds. Good performance and good battery life.
They say alot of times if you experiencing lag n such like that, then it's usually caused from an app you recently downloaded or a widget.
I get this lag too. I will be moving most of my io functions to my Samsung class 10 32gb card when it gets here. Hopefully this is a driver issue. There is another thread about this too.
Sent from my Transformer Prime TF201 using Tapatalk
If you root, there's a fix described here that a few are reporting fixes the issue:
http://forum.xda-developers.com/showthread.php?t=1461583&page=4
this should be brought to gary keys attention imo, so Asus can look at implementing a fix since a fair number of users (including me) aren't rooted. If asus OTA push an I/O fix i'll be very happy, my Iconia A500 runs rings around the TFP in terms of storage performance, and that has to be about the only (noticeable) issue mine seems to have
First, I KNEW the thread title would get your attention. Second, i hear 95% of you calling BS, and I would too. In fact, I am still not sure what to make of it, and it makes ME think the number I am getting is BS.
Here's the deal. Got my replacement Prime back from repair, and like more than a few (read the RMA thread) I was lucky enough to get a replacement instead of a repair. I had sent in my Prime, originally received on 1/3 (shipped on 12/23... GROUND, from Best Buy, though I had paid for overnight. Serial was: BCOKAS**5153
Sent if for repair mostly for a cosmetic issue, scratches on screen border, but I had among the worst GPS compared to others I have seen, no sats indoors, maybe 3-4 outside, never a lock. After the 1/19 update (the one that we still haven't seen a change-log for) I started getting reasonably crappy GPS, LOL as in 4-6 sats inside, and 9-12 Outside, occasional locks, but nothing usable. My Wifi was decent, actually quite good I would say. Bluetooth did not dropout when streaming to BT headset watching Netflix.
The support folks offered to open the replacement and test it before sending, and I said: **** yeah: Test everything: GPS, Wifi/BT check for stuck pixels, make sure the serial shows, also check for light bleed. I was promised they would do all these things. Then it was overnighted to me.... All in all I felt it was a reasonably GOOD RMA/support/repair experience, especially in light of the horror stories, and stories of general incompetence we have seen.
New unit arrived, looked cosmetically perfect. One spot on back, it was glue or tape adhesive from assembly no doubt. Screen appears perfect, with just the tiniest bit of light bleed. Wifi is decent, not outstanding, but close to what I had. No stuck pixels, have yet to test BT streaming. GPS? I got another dead one. I have yet to see a SINGLE FREAKING SATELLITE, Outdoors or in, though it's cloudy here today, LOL (If it's so bad that clouds kill it, it's pretty ****ed up)
So that was a bummer. Yes, it's true I will probably not use it in the car (though I did use my OG that way once in a while, but I refuse to accept something that does not work. It is not in me to say: I can do without that.... even if I CAN, LOL. They are gonna get me another one, cause I am NOT RMA'ing this and going without it a week again... not for GPS anyway. If the WiFi was borked, it would be a different story.
One additional improvement. The power button on this one works correctly. I never knew how bad mine was, but compared to the 3 TF101's I had, it sucked. You never felt it bottom ot, or any detectable motion al all. It was just stiff and stuck, and I had to just press on it super hard to get any effect from it. The new one is perfect, the spring is stronger than the TF101 switches were, but at least it has a positive motion, with a solid detent, and you can tell when it hits bottom. Man, I can't believe how bad what I had was, and I had fully planned to live with it, LOL!
Okay so now for the weirdness. I am putting the new one through tests today, and among the, I ran Antutu. The numbers I got were nothing all that amazing, but the Processor speed threw me for a loop: 1600 Hz. WTF?????
Serial is C1OKAS**1793 I thought I was hallucinating. This thins is NOT ROOTED. I have NO performance tweaks, and nothing changed other than my 126 installed apps... How weird is this???
The numbers that I got were 10222, and a couple more in that range. This was running in Balanced mode. I switched it to Performance Mode (what used to be "normal" back in the Honeycomb days" ) rebooted, and ran Antutu again. Best number to date is 11089. Like I said, not up to par with folks that have REALLY overclocked, but somewhat better than my original... Anyone want to take a stab at explaining this to me? Maybe I am missing something super obvious?
Attached the Antutu result, and one other thing: The box sticker from a "C" serial number, for those who are curious: Check it: NO GPS LISTED!!!!I forgot to take a picture of my original box before RMA, screwed up there. Took dozens of the tablet to protect myself against claims of "CID" Customer Induced Damage, and forgot the box... grrrr.
GOOD NEWS: At least for the present the BOX LABEL still shows GPS. As many of us have discussed, as long as it SAYS GPS, we can still break chops to get them to fix it (or try to, I suppose) but what is with the question mark? Does anyone know the significance of that? That is kind of weird. I am pleased they left GPS spec on the box, expecially since I now have a completely dead GPS setup. I guess it's time to practice my Vulcan Nerve Pinch, but the few squeezes I gave it so far near the GPS Pogo Pins have been completely useless. Not a SINGLE SATELLITE HAS APPEARED YET.
An the Asus Support guy went on about how the Jan 19th patch was on this one. I said: as long as you see a few sats. send it along, He said (laughably) that it would get GPS performance as good as my TF101 did. Hmm. Well, to do that, it would need to start by getting ANY performance, LOL. Then it would need to work, in the car, with CoPilot live, all the way up the East Coast... Somehow, I don't see it happening. Sorry for the long post, but I was up late last night reading the huge tear down thread, and I had no energy for posting to a dozen separate threads today, on my latest "interesting" mix of features and performance with my new Prime. These feel like hand built devices.. every one is unique, LOL!
Hmm i have a C serial too and my GPS is fine. Little worse than my dhd but fine.
Get something like CPU Spy & see what it shows in the "unused cpu states" list at the bottom.
It's not overclocked at all, since ICS the Prime has 1.5 and 1.6 ghz in the kernel, however, they are both disabled. This is the only reason we currently have over clock scripts, with root we can enable these settings.
Sent from my HTC Desire HD using xda premium
Doktaphex said:
It's not overclocked at all, since ICS the Prime has 1.5 and 1.6 ghz in the kernel, however, they are both disabled. This is the only reason we currently have over clock scripts, with root we can enable these settings.
Sent from my HTC Desire HD using xda premium
Click to expand...
Click to collapse
thisssss
TF201 | XDA Premium
Antutu says mine is 1600 too.
Doktaphex said:
It's not overclocked at all, since ICS the Prime has 1.5 and 1.6 ghz in the kernel, however, they are both disabled. This is the only reason we currently have over clock scripts, with root we can enable these settings.
Sent from my HTC Desire HD using xda premium
Click to expand...
Click to collapse
This was what I was going to show with the cpu spy app
I just checked my prime, 1.6ghz with the latest update, no modifications.
^
Mine too
BCOKAS serial number
Lock-N-Load said:
This was what I was going to show with the cpu spy app
Click to expand...
Click to collapse
Exactly. If you all use cpu spy like lock says you will see it never enters the 1.6 state. The processor has those steps built in but aren't enabled in the firmware
People in here are funny..lmao. we already went thru this when ICS first dropped. All primes were showing 1.6ghz in the stats n benchmarks. But this isn't the case. WE KNOW THIS FOR A FACT. If you take CPU spy app n look it'll show the 1.5 n 1.6Ghz speeds disbled. MEANING YOU ARE NOT OVERCLOCKED. Only real overclock you can have is if your rooted and Running ViperControl mod or using SaturnDe ATP tweaks app or running script manually from directory /system/etc..
I was the first one to assume what all of you are saying also when ICS first came out. Then we dug in n researched further n found out this wasn't the case. It did lead us to a true overclock though since we knew it was in the kernel but just disabled. Me, I'm running a "TRUE" 1.6Ghz overclock using ATP tweaks app. Mines is confirmed through CPU spy. Plus my Antutu scores are well over 13,000. So 10,000 would be bad for an overclock like that if you really had one.
SORRY TO RAIN ON YALL PARADE BUT YOURE NOT OVERCLOCKED. Only through Root n methods I listed would you "Truely" be overclocked. CPU Spy will confirm your false claim. Don't feel bad, we thought so to at first when ICS first came out but quickly realized the real truth.
No Biggie
demandarin said:
People in here are funny..lmao. we already went thru this when ICS first dropped. All primes were showing 1.6ghz in the stats n benchmarks. But this isn't the case. WE KNOW THIS FOR A FACT. If you take CPU spy app n look it'll show the 1.5 n 1.6Ghz speeds disbled. MEANING YOU ARE NOT OVERCLOCKED. Only real overclock you can have is if your rooted and Running ViperControl mod or using SaturnDe ATP tweaks app or running script manually from directory /system/etc..
I was the first one to assume what all of you are saying also when ICS first came out. Then we dug in n researched further n found out this wasn't the case. It did lead us to a true overclock though since we knew it was in the kernel but just disabled. Me, I'm running a "TRUE" 1.6Ghz overclock using ATP tweaks app. Mines is confirmed through CPU spy. Plus my Antutu scores are well over 13,000. So 10,000 would be bad for an overclock like that if you really had one.
SORRY TO RAIN ON YALL PARADE BUT YOURE NOT OVERCLOCKED. Only through Root n methods I listed would you "Truely" be overclocked. CPU Spy will confirm your false claim. Don't feel bad, we thought so to at first when ICS first came out but quickly realized the real truth.
Click to expand...
Click to collapse
Uhm... No rain on MY parade. I think you misread my meaning I don't Have any great desire to overclock my ATP. Maybe later, for now I AM interested in the Kernel Module for Background I/O. Cranking up the processor does diddly when the Prime lags, and freezes every so often, especially under any type of background disk activity. 1.4 single. 1.3 ghz Multicore speeds are more than sufficient. People that overclock do so often for the wrong reasons. Why should I be cranking along at 1.6 ghz to browse a news article? If I had ultimate control I would like to control WHEN the additional core kick in, not how fast they tick along, doing a lot of nothing. I did miss your discovery that with ICS, it shows these rates. Right now I would give my right arm to know how, if you do all your posting with the prime, you deal with the cursor misbehavior in text boxes, like when it jumps to the first character position in the box, or insists on jumping to a spot and only the arrow keys will override that. And why is it "mines" do you have multiple Primes?
Edit: Also, I wasn't stating that "11, 089" was a enormous score, if you read I was stating that it was a BETTER score than my returned Prime, but not what I would expect if overclocked. I'm glad your numbers are so impressive. Maybe someday MY Prime will have a Antutu score of over 13,000! That's really fast! And that helps exactly how?
Also, since the processor has those steps built in it technically isn't overclocking to enable the 1.6GHz step. You are only overclocking when you go above that.
SmartAs$Phone said:
Uhm... No rain on MY parade. I think you misread my meaning I don't Have any great desire to overclock my ATP. Maybe later, for now I AM interested in the Kernel Module for Background I/O. Cranking up the processor does diddly when the Prime lags, and freezes every so often, especially under any type of background disk activity. 1.4 single. 1.3 ghz Multicore speeds are more than sufficient. People that overclock do so often for the wrong reasons. Why should I be cranking along at 1.6 ghz to browse a news article? If I had ultimate control I would like to control WHEN the additional core kick in, not how fast they tick along, doing a lot of nothing. I did miss your discovery that with ICS, it shows these rates. Right now I would give my right arm to know how, if you do all your posting with the prime, you deal with the cursor misbehavior in text boxes, like when it jumps to the first character position in the box, or insists on jumping to a spot and only the arrow keys will override that. And why is it "mines" do you have multiple Primes?
Edit: Also, I wasn't stating that "11, 089" was a enormous score, if you read I was stating that it was a BETTER score than my returned Prime, but not what I would expect if overclocked. I'm glad your numbers are so impressive. Maybe someday MY Prime will have a Antutu score of over 13,000! That's really fast! And that helps exactly how?
Click to expand...
Click to collapse
You are correct. Stock speeds are more than sufficient for everyday usage. Overclock just makes everything even more faster n snappier. Web pages load up faster, apps open up faster, etc.. Its not for everyone. But for me its so easy to do and has no ill effects on my tablet. I'm like why not harness the extra power. Its in the kernel anyways and newer tegra3 tablets down the line will have exact same chip but the 1.6Ghz speeds enabled Stock. I do find myself going back to stock speeds alot because I truly appreciate how fast this device already is stock and ill get better battery life on stock. I do get pretty good battery life also on 1.6Ghz overclock.
Also, yes, I DO ALL MY POSTINGs, from the prime. I don't experience the major slowdowns or lags some seem to experience. Mostly only occurred sporadically in stock browser. I used ATP settings to change scheduler from noop(stock) to cfq. It seemed to help for a while but then came across an issue before, where I never had before, where the whole stock browser seemed to just freeze up or stop loading. Before it might freeze n then prompt come up saying to wait or close. After the switch it didn't do any of this. So i switched back to noop I/o scheduler n everything has been running great for me. I mainly just use the ATP tweaks app now for quick access to overclocking. Less steps needed vs. Having to go into terminal emulator and pulling up vipercontrol. With ATP all I do is open up the app n press Turbo2 n I'm overclocked. I did try out the deadline I/o scheduler also but didn't seem to make much of a difference. Actually acted kinda funny at times so i switched back to Noop. As Noop is the best out of all of them. Its just some people switch to other schedulers based on their needs for prime to PC file transfers n such. So its really a preference thing and what you experience with your personal usage and benefit the change makes for you.
AS far as bench scores go, I just said that cuz you mentioned yours n wanted to show another form of proof you weren't overclocked. Now for the cursor jumping around, I know what you mean..lol. but I use thumb keyboard, split mode, n very use to using the arrow keys to correct anything. Yeah the cursor does jump around at times but it still goes where you want most of the time through touching where u want it. Plus copying n pasting is very easy for me now that o got the technique down packed. So all of this is working pretty well for me and not a hassle at all as majority of my post count had been made with my prime alone. Before I got the prime, I was posting from my Ipad n Atrix4g. I can type very fast using thumb keyboard. Love it. Also, I only have one prime. The "mines" must've been a typo.
---------- Post added at 11:52 PM ---------- Previous post was at 11:50 PM ----------
rand4ll said:
Also, since the processor has those steps built in it technically isn't overclocking to enable the 1.6GHz step. You are only overclocking when you go above that.
Click to expand...
Click to collapse
Yeah yeah..lmfao but we can't enable it stock. So its still "Technically" considered overclocked since it is higher than allowed out the box. Only if you running a "true" overclock using vipercontrol mod or ATP tweaks.
It's all good.
demandarin said:
You are correct. Stock speeds are more than sufficient for everyday usage. Overclock just makes everything even more faster n snappier. Web pages load up faster, apps open up faster, etc.. Its not for everyone. But for me its so easy to do and has no ill effects on my tablet. I'm like why not harness the extra power. Its in the kernel anyways and newer tegra3 tablets down the line will have exact same chip but the 1.6Ghz speeds enabled Stock. I do find myself going back to stock speeds alot because I truly appreciate how fast this device already is stock and ill get better battery life on stock. I do get pretty good battery life also on 1.6Ghz overclock.
Also, yes, I DO ALL MY POSTINGs, from the prime. I don't experience the major slowdowns or lags some seem to experience. Mostly only occurred sporadically in stock browser. I used ATP settings to change scheduler from noop(stock) to cfq. It seemed to help for a while but then came across an issue before, where I never had before, where the whole stock browser seemed to just freeze up or stop loading. Before it might freeze n then prompt come up saying to wait or close. After the switch it didn't do any of this. So i switched back to noop I/o scheduler n everything has been running great for me. I mainly just use the ATP tweaks app now for quick access to overclocking. Less steps needed vs. Having to go into terminal emulator and pulling up vipercontrol. With ATP all I do is open up the app n press Turbo2 n I'm overclocked. I did try out the deadline I/o scheduler also but didn't seem to make much of a difference. Actually acted kinda funny at times so i switched back to Noop. As Noop is the best out of all of them. Its just some people switch to other schedulers based on their needs for prime to PC file transfers n such. So its really a preference thing and what you experience with your personal usage and benefit the change makes for you.
AS far as bench scores go, I just said that cuz you mentioned yours n wanted to show another form of proof you weren't overclocked. Now for the cursor jumping around, I know what you mean..lol. but I use thumb keyboard, split mode, n very use to using the arrow keys to correct anything. Yeah the cursor does jump around at times but it still goes where you want most of the time through touching where u want it. Plus copying n pasting is very easy for me now that o got the technique down packed. So all of this is working pretty well for me and not a hassle at all as majority of my post count had been made with my prime alone. Before I got the prime, I was posting from my Ipad n Atrix4g. I can type very fast using thumb keyboard. Love it. Also, I only have one prime. The "mines" must've been a typo.
---------- Post added at 11:52 PM ---------- Previous post was at 11:50 PM ----------
Yeah yeah..lmfao but we can't enable it stock. So its still "Technically" considered overclocked since it is higher than allowed out the box. Only if you running a "true" overclock using vipercontrol mod or ATP tweaks.
Click to expand...
Click to collapse
I am glad that I am not alone with the cursor thing. A bit distracting, but for the most part I use the Hardware Keyboard in the dock when doing any serious typing. Like with my TF101 I like that the dock acts as a "stand" and a cover/case. and at home I tend to do a lot of wen/newsreader browsing, and posting, as you know.
I know you mentioned that battery life seems decent when running at full clock speed (okay, we know it's technically not "overclocked") so I say "full speed to avoid being corrected, LOL! I am curious about battery performance with it running at a true 1.6, and before you answer that it is not bad, can you say for certain, as in have you tested the discharge times with the Turbo2 enabled full time? Or, if you can at least say that it's not too bad, what would you peg the loss to be? 10% or more? I agree that if I had an app to easily toggle it on/off like the build in app that prioritizes processors, I would like to use it at times.. The built in one I think does the following:
Power Saving: Runs all Cores at reduced speed, I think its 600 MHz, 700 MHz when three are active, and 1 GHz when one or two are active
in "Balanced mode" - the cores are capped at 1.2 GHz
and in Performance Mode (used to be normal mod, just to keep it sounding exciting, they changed it to "Performance with ICS, LOL!) a single core runs at 1.4, or all 4 cores can run at full clip of 1.3
Not sure where they came up with these choices, but I will bet it was only after Nvidia ran them every which way, and came to these numbers as the best compromise of power and battery life. I think the Tegra 3 SOC also controls the video brightness and depth on the fly as well. All told a nice implementation. I still see lags occasionally, and games like Riptide GP freeze in a "Stutter Frame" kind of lock, till I exit to home screen, and resume the game, and it's good again.
Anyone else get that?
SmartAs$Phone said:
I am glad that I am not alone with the cursor thing. A bit distracting, but for the most part I use the Hardware Keyboard in the dock when doing any serious typing. Like with my TF101 I like that the dock acts as a "stand" and a cover/case. and at home I tend to do a lot of wen/newsreader browsing, and posting, as you know.
I know you mentioned that battery life seems decent when running at full clock speed (okay, we know it's technically not "overclocked") so I say "full speed to avoid being corrected, LOL! I am curious about battery performance with it running at a true 1.6, and before you answer that it is not bad, can you say for certain, as in have you tested the discharge times with the Turbo2 enabled full time? Or, if you can at least say that it's not too bad, what would you peg the loss to be? 10% or more? I agree that if I had an app to easily toggle it on/off like the build in app that prioritizes processors, I would like to use it at times.. The built in one I think does the following:
Power Saving: Runs all Cores at reduced speed, I think its 600 MHz, 700 MHz when three are active, and 1 GHz when one or two are active
in "Balanced mode" - the cores are capped at 1.2 GHz
and in Performance Mode (used to be normal mod, just to keep it sounding exciting, they changed it to "Performance with ICS, LOL!) a single core runs at 1.4, or all 4 cores can run at full clip of 1.3
Not sure where they came up with these choices, but I will bet it was only after Nvidia ran them every which way, and came to these numbers as the best compromise of power and battery life. I think the Tegra 3 SOC also controls the video brightness and depth on the fly as well. All told a nice implementation. I still see lags occasionally, and games like Riptide GP freeze in a "Stutter Frame" kind of lock, till I exit to home screen, and resume the game, and it's good again.
Anyone else get that?
Click to expand...
Click to collapse
See I don't even have a dock n typed all that with no problem. Thumb keyboard makes typing so easy n fast on the prime. All I use is 2 thumbs to type n can do it very fast with the split keyboard mode.
AS for battery life if you ever ran the built in performance mode full time, then you will know it drains the fastest out of all modes because its running the higher speeds. Now with 1.6Ghz it drains about the same or a lil faster than performance mode. I've noticed lately I've been getting battery life even better than performance mode on my overclock. For the quick implementation of it that's easy. It can't get no easier then this. Check this out..lol. if you rooted, all you have to do is go into prime developement section. Go to ATP tweaks thread. Download and install ATP app into prime. Then open it. Allow superuser permissions. Then just press the Turbo2 tab n bam! Now you St 1.6Ghz overclock. The real one. You can use CPU spy to confirm that 1.6ghz speed is enabled.
hen if you want to disable it, just reboot tablet. Then whenever u want. All u have to do is open up that app n press turbo2 tab. There's a turbo1 also n that's for 1.5Ghz.. There is a step to do a more hardcore overclock to where you can run ALL 4 CORES @1.6GHZ at the same time all the time. Of course this burns up the battery the fastest. I doubt u ready for that yet though... lmfao just take it slow..lol use the regular overclocks. Get ATP tweaks app in developement section.
For the games, I don't have no stutters or freezes even on stock speeds.
Just found this article:
http://www.androidpit.com/nexus-7-overclocked-to-1-64-ghz-destroys-tablets-priced-3x-more
The article mentions something about a "Trinity" kernel. Look at those write speeds! Anyway, is there anything we can do like this to our Primes?
Moist said:
Just found this article:
http://www.androidpit.com/nexus-7-overclocked-to-1-64-ghz-destroys-tablets-priced-3x-more
The article mentions something about a "Trinity" kernel. Look at those write speeds! Anyway, is there anything we can do like this to our Primes?
Click to expand...
Click to collapse
Who cares about quadrant benchmarks? Btw the nexus 7 has basically the same chip as the prime (or almost) so this could be done to our tablet as well i guess. But the currently released kernel sucks ass. So NVIDIA should finally release something good for us!
I don't particularly care for benchmarks, but when something almost identical to our Primes performs much better than them, I'd like a piece of that pie...
Sent from my Transformer Prime TF201 using Tapatalk 2
The trinity kernels are available at derkernel.com. I don't know much about android development, but would it not be possible for someone to port this kernel or build one for our prime? Benchmarks mean very little, but I'd live to know how this kernel impacts our real life performance, especially the write speeds.
Sent from my Desire S using xda app-developers app
benchmarks and especially from quadrant suck .. my prime loads things like browser pages apps etc faster than nexus 7 , (tested it myself ). also the prime in the benckmark shown in your link is just full stock without tweaks etc ,
as i said my prime loads things faster so who cares about a stupid benchmark
ps im running jellybean not ics
You should also take into consideration that the N7 runs Jellybean, which brings performance updates.
Is the prime getting official jb?
Sent from my SGH-T999 using Tapatalk 2
Are you kidding, every tablet/phone rapes Prime when it comes to disk speed. And Asus didn't even bother to fix this in Infinity.
Nobody wants to point out that this exact behaviour (I/O scores going insane) happens on other devices when you overclock?
It's just Quadrant freaking out and giving wrong scores. I/O scores can't magically double because of a 200mhz overclock to the cpu.
Nothing to see here, move along.
Namidairo said:
Nobody wants to point out that this exact behaviour (I/O scores going insane) happens on other devices when you overclock?
It's just Quadrant freaking out and giving wrong scores. I/O scores can't magically double because of a 200mhz overclock to the cpu.
Nothing to see here, move along.
Click to expand...
Click to collapse
They say they use a different kernel, it's not JUST an overclock. It's different mounting options, other optimalizations. Still, the io score is insane and I guess there's some trick to it. (Different mounting options on the prime jb roms lead to an 2 or 3-fold improvement already).
I've been tracking benchmarks, the N7 doesnt beat the TF. It's not stable at certain OC speeds and won't complete the test. I was able to test it at a "stable" OC speed and see results. I can't unlock my prime so it's basically stock (except for a I/O fix) and it beats the N7 OC 1.6GHz linaro kernel. When I tried the Trinity kernel it wouldn't finish the test, so it's not ready yet. Both are on regular settings, I didnt try to get high benchmarks, since that would require lots of tweaking and draining of the battery, I tried to get normal, everyday results. Stock vs stock shows the TF beating the N7. OC vs stock shows the TF almost beating the N7 despite it being OC. If I was able to OC the prime, then it would be a better result for the prime, OC vs OC.
https://docs.google.com/spreadsheet/ccc?key=0AnAJoa0-ynONdDZIUk1EZFZfay1mN3RScU42VnNia2c
bbua88 said:
Is the prime getting official jb?
Sent from my SGH-T999 using Tapatalk 2
Click to expand...
Click to collapse
Yes it is though there is no definitive word on when. In my mind the two main issues surrounding comparisons at this point are that the N7 has the new version of the OS and also that the new OS uses Chrome as it's default browser. The reviewers typically use whatever is the default browser on the device for bench marking browser scores. So to do an apples to apples comparison we would have to wait for the TF201 to get its 4.1 update (hopefully soon) and compare when both devices are using the same version.
It's more than just a ROM. I'm on JB rom and it performs like ass. Obviously the people who managed to get the benchmark, made some tweaks that aren't public yet. A benchmark that shows impressive results is still showing what that thing can do. My Prime has never performed like that on any rom/kernel that I have tried...
Sent from my Transformer Prime TF201 using Tapatalk 2
I/o improvements on trinity kernel is real for nexus 7. im running it now. on rl sqlite bench i get 16secs. overall. motley latest kernel on nexus7 gives same great results. on that kernwl we have gpu overclocked to 520mhz. like it is on prime. on nenamark2 i get 64.7+ Fps.
my app store apps downliad and install nearly instantly. reason trinity kernel i/o so high is because if you run trinity toolkit app, it automatcally loads a script to improve i/o. then also has option to run with fsync set to faster. those users with highest I/o scores use both i/o improvement scripts. it does translate to real world oerformance though.
---------- Post added at 03:03 PM ---------- Previous post was at 02:50 PM ----------
dude2k5 said:
I've been tracking benchmarks, the N7 doesnt beat the TF. It's not stable at certain OC speeds and won't complete the test. I was able to test it at a "stable" OC speed and see results. I can't unlock my prime so it's basically stock (except for a I/O fix) and it beats the N7 OC 1.6GHz linaro kernel. When I tried the Trinity kernel it wouldn't finish the test, so it's not ready yet. Both are on regular settings, I didnt try to get high benchmarks, since that would require lots of tweaking and draining of the battery, I tried to get normal, everyday results. Stock vs stock shows the TF beating the N7. OC vs stock shows the TF almost beating the N7 despite it being OC. If I was able to OC the prime, then it would be a better result for the prime, OC vs OC.
https://docs.google.com/spreadsheet/ccc?key=0AnAJoa0-ynONdDZIUk1EZFZfay1mN3RScU42VnNia2c
Click to expand...
Click to collapse
trust me, developement for nexus 7 smokes prime. i have both also. except my prime is unlocked and oc. nexus 7 still beays it by largw margin. reason being nexus 7 has better/newer kernel base. plus nexus 7 kernels more performance oriented.
so oc vs. oc. nexus 7 wins. why? because although nexus7 in its infancy developement wise, they already made larger strides in performance improvements. stock vs. stock, prime wins in "most" categories. like gpu or cpu intensive benching. even then, stock, nexus 7 still beats out prime in other areas, like i/o performance.
developement scene on nexus7 has more to work with. they abke to do more. we have franco kernel, faux kernel, morrific der kernel, motley, etc... all the hardest hitting devs. are developing for it. tons of custom roms. whether original or kanged.
prime performance and benching will dramatically improve once its on jelkybean and newer kernel & drivers. bigger improvements can be made dev. wise once primw updates to jb.
---------- Post added at 03:05 PM ---------- Previous post was at 03:03 PM ----------
Moist said:
It's more than just a ROM. I'm on JB rom and it performs like ass. Obviously the people who managed to get the benchmark, made some tweaks that aren't public yet. A benchmark that shows impressive results is still showing what that thing can do. My Prime has never performed like that on any rom/kernel that I have tried...
Sent from my Transformer Prime TF201 using Tapatalk 2
Click to expand...
Click to collapse
if you run i/o script and use fsync on orime, your i/o scores will dramaticalky improve along with i/o performance.
Why is quadrant still even used? It was already proved to be heavily outdated and is overall terrible.
There are so many better benchmarking apps out there.
@demandarin Do you think I have not tried those? goodness...
Sent from my Transformer Prime TF201 using XDA Premium HD app
Gotta say.. seeing stuff like this has been making me chomp at the bit for a new release on the Prime, or to simply unlock and install a ROM on the device... the only reason I've held out this long, is due to the fact that Gary occasionally calls out for locked devices for beta testing.
We have a Nex7 at the house, and everything it does is just faster than the Prime... *sigh* Quite depressing, really.
I agree, there isn't much dev work on the Prime at all compared to the Nexus 7. But I think that's because the price point and it's from "Google" so it's more...universal and updated. I love both devices for separate reasons. The Prime and Nexus 7 are basically the best android tablets out there imo (except for the TF infinity, but that only has a better screen). My Evo LTE isn't too far behind either (does better on other tests) but dev work kinda sucks for that as well. I'm hoping the JB update will speed things up for the Prime. On the OG transformer it actually works pretty well.
Moist said:
It's more than just a ROM. I'm on JB rom and it performs like ass. Obviously the people who managed to get the benchmark, made some tweaks that aren't public yet. A benchmark that shows impressive results is still showing what that thing can do. My Prime has never performed like that on any rom/kernel that I have tried...
Sent from my Transformer Prime TF201 using Tapatalk 2
Click to expand...
Click to collapse
If you are on JB now, then by definition you are on a beta level ROM at best. Every one of them come with a list of the things they don't have working yet - which is not a poke at the developers but only to say that whatever JB ROM you are running now is not what the rest of us will be getting when the real update comes out.
Screen size makes a massive difference to performance.
Rewriting data to the larger estate takes extra grunt away from performance figures.
Check it out here, I am going to test it out see how it goes.
jgaf said:
Check it out here, I am going to test it out see how it goes.
Click to expand...
Click to collapse
I just tried it out and it does work! Recent apps menu loads way faster, the whole device seems faster!
I can't see the difference on my TF300
Thanks jgaf, so far a tremendous reduction in lag but have only been using this for a few hours. Any other feedback or issues?
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app
I'm not sure if its working for me, but it hasn't gotten worst. Haven't tried gaming, maybe I should try playing spiderman or batman and see if it improves the gameplay. I'm so used to my phone (GS3) that the pad will always seem slow There's a lot of debate on the original thread, if it really works or not and how much change does it do. It is quite impossible to read the whole thread, since it grows exponentially by each hour.
jgaf said:
I'm not sure if its working for me, but it hasn't gotten worst. Haven't tried gaming, maybe I should try playing spiderman or batman and see if it improves the gameplay. I'm so used to my phone (GS3) that the pad will always seem slow There's a lot of debate on the original thread, if it really works or not and how much change does it do. It is quite impossible to read the whole thread, since it grows exponentially by each hour.
Click to expand...
Click to collapse
Your on an optimized rom and kernel, so I would guess the difference probably isn't as dramatic as with stock. It never ceases to amaze me that the snapdragon can outperform the tegra 4, but that extra gig of ram on the s3 is a big part of it.
I spent some time looking through that thread and comments seemed to lean towards better performance with a reduction in battery life. I can live with that on my tablet as I am not often on the go with it anyway. Good find bro!
Sent from my SGH-I747M using XDA Premium HD app
https://play.google.com/store/apps/details?id=com.lcis.seeder
http://forum.xda-developers.com/showthread.php?t=1987032
Has anyone tried it on their rooted Prime? I would love to try it but mine is gone for yet another RMA.
apsamrit said:
https://play.google.com/store/apps/details?id=com.lcis.seeder
http://forum.xda-developers.com/showthread.php?t=1987032
Has anyone tried it on their rooted Prime? I would love to try it but mine is gone for yet another RMA.
Click to expand...
Click to collapse
Nope, but I'm just about to. I'll report back. Nice spot. Thanks for sharing.
Installed on my Stock but rooted prime now. Appears to eliminate all the stock browser lag I used to have. It's now blazing fast, and I can scroll while its still loading stuff which I was never able to do before. Holy mother of all fixes it seems like:good:
I installed it last night on my prime. It does lot for my prime. I also did some of the early investigation to what was really being fixed by this hack. I am glad Cyanogen has confirmed some of my theories. If you are rooted I would highly recommend installing it.
-l4k
Sent from my Transformer Prime TF201 using xda app-developers app
fatski said:
Nope, but I'm just about to. I'll report back. Nice spot. Thanks for sharing.
Click to expand...
Click to collapse
Tried this and also read up a bit on what it does. Unfortunately I don't think what this tries to do is an issue on the Prime. This basically makes sure there is always random data available for apps from a location that will block if there's none available. However, when I ran a command that was constantly consuming all the random data (cat /dev/random > /dev null), I could see that the random data was getting consumed but it wasn't causing a perceptibly adverse effect on anything else.
I also just ran the app for a while, and I didn't notice anything being more responsive. It will consume extra battery though.
Take look at my write up on the Google bug. That is what the author originally thought the problem was. What is probably going on is kernel lock contention due to the code that generates entropy when the pool is depleted.
Installed it just now, no discernible effect. Quick reboot... Chameleon launcher, app switching, Chrome, all smooth as butter. I am gobsmacked! I think have a feeling this works well for me because of the number of connections made by Chameleon, logged in accounts (I have 9 accounts logged in at all times) and apps syncing that I regularly exhaust the block device. Add to our already slow IO, any blocking at the kernel level will be multipled.
Going to monitor it some more and report back on battery life and any further observations. Thanks for the share! :good:
To all using this: please post some real before & after data. Anecdotes do nothing but cause confusion and misinformation.
xxbeanxx said:
To all using this: please post some real before & after data. Anecdotes do nothing but cause confusion and misinformation.
Click to expand...
Click to collapse
it has no effect on any benchmarks so its hard to quantify. for my use it does not make a discernible difference in anything I do with my prime is about the best data you can get.
It has no effect whatsoever: https://plus.google.com/u/0/115049428938715274412/posts/GWr72W9zmY2
Will give it a try.
xxbeanxx said:
To all using this: please post some real before & after data. Anecdotes do nothing but cause confusion and misinformation.
Click to expand...
Click to collapse
pandaball said:
It has no effect whatsoever: https://plus.google.com/u/0/115049428938715274412/posts/GWr72W9zmY2
Click to expand...
Click to collapse
I have done some benchmarks and it makes no difference to those whatsoever. This bothers me because I can't understand if this is supposed to do nothing, how does my device run orders of magnitude snappier all the time while running this compared with intermittent lag that I usually get (for example: waiting 1-3 seconds for the recent app list to show normally vs instant with this mod) . The CPU governor can't be it: I set cpu to max and use the performance governor and I still have massive lag. I have tried numerous things for I/O: different schedulers, different mount options to reduce writes on reads and improve write speed ie. noatime, nodiratime, discard, and nothing works! I have tried to narrow down a single process that causes the lag, no dice. I know I only have just under 2 days of testing behind me and yes I only have anecdotal evidence but I have been running this a long enough time with absolutely zero lag that I can't say it makes no difference for me. I'll continue testing but so far it has been too long to be placebo IMO.
gunzy83 said:
I have done some benchmarks and it makes no difference to those whatsoever. This bothers me because I can't understand if this is supposed to do nothing, how does my device run orders of magnitude snappier all the time while running this compared with intermittent lag that I usually get (for example: waiting 1-3 seconds for the recent app list to show normally vs instant with this mod) . The CPU governor can't be it: I set cpu to max and use the performance governor and I still have massive lag. I have tried numerous things for I/O: different schedulers, different mount options to reduce writes on reads and improve write speed ie. noatime, nodiratime, discard, and nothing works! I have tried to narrow down a single process that causes the lag, no dice. I know I only have just under 2 days of testing behind me and yes I only have anecdotal evidence but I have been running this a long enough time with absolutely zero lag that I can't say it makes no difference for me. I'll continue testing but so far it has been too long to be placebo IMO.
Click to expand...
Click to collapse
Haha benchmarks dont equate to real life performance. Benchmark scores have been decreasing since gingerbread even though the new software makes everything faster
Sent from my SAMSUNG-SGH-I727 using xda premium
The reason why your device *feels* snappier is because constant I/O activity ensures the CPU will remain at rather high clocks, producing the effect of a smoother experience at the expense of massive battery drain. The explanation of increasing the entropy does nothing by itself, you could cause a lot of I/O activity to anywhere in the emmc and it would produce a similar effect.
Sent from my HTC One X using Tapatalk 2
ian1 said:
Haha benchmarks dont equate to real life performance. Benchmark scores have been decreasing since gingerbread even though the new software makes everything faster
Sent from my SAMSUNG-SGH-I727 using xda premium
Click to expand...
Click to collapse
Indeed. That was my point, benchmark results were unchanged but user experience is different. Hrrrm
Didn't do anything for me. Uninstalled it.
Thanks for finding this
I was skeptical as many are about this. I did not see any improvement in Quadrant scores. I installed this for one reason only and it works!
I stream .m4v movies from my NAS drive to my TV through my TP. Before this was installed I would have problems with stuttering no matter what player I tried. Sometimes I would have problems right away, and depending on the player/movie I could watch about an hour before the stuttering/lag begins. I have been beyond frustrated, especially since my wife has an IPad 2 that plays the same movies flawlessly.
Since installing this I can now play my movies without a hitch. I don't know why, but it works. I only use it for this purpose as my other testing showed no major difference. I turn it off when I am not watching movies as I have read that the battery drain is increased.
I streamed 3 movies this weekend without a problem and 2 of them were back to back and I only used approx 35% battery.
Running AndroWook 1.51 and with 1.6 oc kernal.
f'n around said:
I was skeptical as many are about this. I did not see any improvement in Quadrant scores. I installed this for one reason only and it works!
I stream .m4v movies from my NAS drive to my TV through my TP. Before this was installed I would have problems with stuttering no matter what player I tried. Sometimes I would have problems right away, and depending on the player/movie I could watch about an hour before the stuttering/lag begins. I have been beyond frustrated, especially since my wife has an IPad 2 that plays the same movies flawlessly.
Since installing this I can now play my movies without a hitch. I don't know why, but it works. I only use it for this purpose as my other testing showed no major difference. I turn it off when I am not watching movies as I have read that the battery drain is increased.
I streamed 3 movies this weekend without a problem and 2 of them were back to back and I only used approx 35% battery.
Running AndroWook 1.51 and with 1.6 oc kernal.
Click to expand...
Click to collapse
+ 1 here. Running last JB update + rooted, wasn't able to play heavy games since the update to JB when we lost the OC kernel from ICS, games were laggy, freezing,almost unplayable. After seeder.apk installation, <i am again able to play NOVA 3, Batman, Spiderman, Asåhalt 7, without any lag and high FPS, seems even better than when my Prime was on ICS OC to 1600. I don't care about what people say but what I experience o preformance, this is something I've learnt after all those years using Android devices. Give it try expecially if you run heavy apps. Battery is a bit worse, but who cares when the prime isnot overheating and run a lot better.
Sent from my Transformer Prime TF201 using XDA Premium HD app
I want to install it too, but as I understand there is no root for current JB yet, or am I wrong?!
as the uninstall?