[Q] pathetic I/O score from quadrant standard - Samsung Galaxy Nexus

Ever since OTA update from 4.0.2 to 4.0.4, my quadrant score has been consistently sucky. 1957 overall, the worst being database write scores of 950 or so. So I rooted and flashed CWR and GLaDOS (v1.34) kernel to see if that might do anything. Nope. I screwed around with the SD cache and increased it to 1024KB, and that didn't help any either.
I don't know what I'm doing Anybody know how to make this better, or if I even should care... I mean, maybe Quadrant Standard is broken, idk. I tried SD Tools to see if maybe a different app would get a different score but it got a buffer error and refused to run or do anything. Sucks.
GNex 4.0.4 stock IMM76K (CDMA) 3.0.8-GLaDOS-V1.34

You shouldn't care.

Test it with another benchmark to confirm That low of a score will bottleneck performance (if its real). Normally I say ignore it, but in this case you might want to investigate further.
Stop playing with sd readahead. If there was ever a worthless tweak on Galaxy Nexus, this takes the cake.

Does your phone seem laggy. If not don't worry about it.
Sent from rooted Jitterbug running ICS

Firstly, pay no mind whatsoever to benchmarks. They're not fair since they show the best results in perfect conditions such as just rebooted, unrealisticly high overclock, few apps installed and so on so forth.
Secondly, do not use SD readahead like adrynalyne said. The Galaxy Nexus has no external storage. Only the internal storage partitioned and mounted as external, but it's still the internal NAND flash memory.
On phones with an actual SD card in an SD card slot, that tweak would maybe be good.
Real-time use diverts from benchmark use.

I thought tweaking the SD cache would work for the internal SD card too, but considering I had to google readahead imma go ahead and defer to your obviously superior knowledge :3
I just ran AnTuTu and got 647 on I/O, 5615 overall I'm not too sure how this score matches up to Quadrant Standard, but it shows poor me being much lower than the average GNex.
It does seem laggy sometimes and runs hot... 42C sitting there idle, on the charger...

stompsfrogs said:
I thought tweaking the SD cache would work for the internal SD card too, but considering I had to google readahead imma go ahead and defer to your obviously superior knowledge :3
I just ran AnTuTu and got 647 on I/O, 5615 overall I'm not too sure how this score matches up to Quadrant Standard, but it shows poor me being much lower than the average GNex.
It does seem laggy sometimes and runs hot... 42C sitting there idle, on the charger...
Click to expand...
Click to collapse
Thats just it though, there is no internal SD (and even if there was, the tweak wouldn't affect I/O scores).
647 is a decent score on Antutu for i/o, so I would say you are fine.

adrynalyne said:
Thats just it though, there is no internal SD (and even if there was, the tweak wouldn't affect I/O scores).
647 is a decent score on Antutu for i/o, so I would say you are fine.
Click to expand...
Click to collapse
k thx. I thought the folder called mnt/sdcard was like a fake SD card, except with MTP instead of FAT or whatever normal SD cards have... I don't quite get this lower level stuff still, and I been trying to be smarter about it for a while. Every time I learn something they change it

stompsfrogs said:
k thx. I thought the folder called mnt/sdcard was like a fake SD card, except with MTP instead of FAT or whatever normal SD cards have... I don't quite get this lower level stuff still, and I been trying to be smarter about it for a while. Every time I learn something they change it
Click to expand...
Click to collapse
Google was kind of clever with the sdcard. It is actually a symlink to the data partition. Specifically, /data/media is the same directory as /sdcard.

symlinks, I heard of those! I been putting Linux on everything I can get my hands on recently. I'm still in the "breaking everything" stage but I got as far into regex as ln -s.
and it turns out, the heat was from 4G. I forgot about that.
thanks bunches adrynalyne.

As long as its not laggy or slow, you shouldnt care, unless its a competition for you
Sent from my Galaxy Nexus

Quadrant is a joke of a benchmark. Don't worry too much about it unless you notice slow-downs in actual use.

Some kind of SOB got a score 8617 clocked at 1200mhz with 4.0.1.
Can anyone explain how this kind of **** happened???

stompsfrogs said:
Ever since OTA update from 4.0.2 to 4.0.4, my quadrant score has been consistently sucky. 1957 overall, the worst being database write scores of 950 or so. So I rooted and flashed CWR and GLaDOS (v1.34) kernel to see if that might do anything. Nope. I screwed around with the SD cache and increased it to 1024KB, and that didn't help any either.
I don't know what I'm doing Anybody know how to make this better, or if I even should care... I mean, maybe Quadrant Standard is broken, idk. I tried SD Tools to see if maybe a different app would get a different score but it got a buffer error and refused to run or do anything. Sucks.
GNex 4.0.4 stock IMM76K (CDMA) 3.0.8-GLaDOS-V1.34
Click to expand...
Click to collapse
Yes, I got the same...
Looking at the results, everything looks the same as the example Galaxy nexus, but I got an I/O around 1024, when a friend of mine with a galaxy note had around 4000... :S
Overall results are lower than the nexus one because of this lack of IO???
Did they put the results wrong on the Quadrant app or they tested with another ROM? I have rooted stock rom.

Related

[Q] Lag fix as on Galaxy S series

Do we have some kind of a lag fix for MT3GS available for Galaxy S phones? I heard it speeds up the system in more than 2 times as shown in Quadrant.
as Cyanogen showed all the is doing is exploiting one of the things that quadrant tests with a code mod... he was able to score 9,000+ on quadrant showing it isn't a perfect benchmark
Owhhh, you dissapoint me, because I was very pleased while looking at the quadrant results and seeing my device score above N1 2.1 and Galaxy S (with a 844 kernel of course)
Here's the thing.
The Galaxy S line of phones has 16gb of internal memory. Unlike ours, which is a standard flash memory, theirs is an sd card. A pretty slow one.
The lag fix makes everything run off the external sd card, which is much faster.
We don't have an internal sd card.
Sent from my T-Mobile myTouch 3G Slide using XDA App
Thanks, you made it clear.

Android on RAM?

Hi just out of interest would it be possible to let android run from RAM on the desire as a dev made it possible on the HD2? I mean that would really improve performance as RAM is quite a bit faster than NAND and SD and on the HD2 it seems to work well and not even run out of ram (which is quite awesome IMO).
quad2012 said:
Hi just out of interest would it be possible to let android run from RAM on the desire as a dev made it possible on the HD2? I mean that would really improve performance as RAM is quite a bit faster than NAND and SD and on the HD2 it seems to work well and not even run out of ram (which is quite awesome IMO).
Click to expand...
Click to collapse
its only the fs that gets stored on the ram before boot not data, so I/O improvements would be minimal as the device rarely need to R/W from system compared to /data, also HD roms need to run totally from sd card witch is slower than nand obviously, so you gotta ask yourself would you be prepared to loose 150-250mb of ram, just from a minor speed improvement?
Ahh ok makes sense only one noob question xD on HD2 lot's of people experienced quite a performance boost is that because of the different devices?
quad2012 said:
Ahh ok makes sense only one noob question xD on HD2 lot's of people experienced quite a performance boost is that because of the different devices?
Click to expand...
Click to collapse
its prob because the whole rom is booted from the sd card, if you notice, when Multiple process try to access a sd card, each process after the first one gets lower and lower R/W speed, try it with a sd card reader on your pc, if you try to copy 10 seperate files to your sd card at once, the first one coppies at normal speed but the others will be gradually slower and slower, the same thing happens when Multiple android processes would try to access the sd card, leading to obvious lag, a nand chip, and ram dosent have this problem all processes have the same R/W speed.

[Q] Dalvik Cache

By moving the dalvik cache to the sd-ext, does it slow the rom down?
Some people will say yes, some people will say no...
From personal use I'll say I've moved my dalvik cache to my sd card on a couple of rom's and never noticed a difference. I wouldn't recommend it with a class 2 sd card, but with class 4 and above it should be fine.
I never felt any difference and I have been practicing this long before alternate hboot's came into play.
Also card's wear is reduced and supposedly battery life a bit better by a marginal value equal to those of havs/svs, hardkeys backlight on/off etc.
zedmarcus said:
I wouldn't recommend it with a class 2 sd card, but with class 4 and above it should be fine.
Click to expand...
Click to collapse
Again:
Class = minimal sequential write speed -> large files.
Random writes -> dalvik cache.
Hence, cards class only guarantees card will perform at it's rated speed with sequential writes. Unfortunately increase in sequential speed doesn't mean linear increase in random. Or put bluntly, this justifies why some people could run data2sd scripts with lower end cards better than the others who had better cards.
erklat said:
I never felt any difference and I have been practicing this long before alternate hboot's came into play.
Also card's wear is reduced and supposedly battery life a bit better by a marginal value equal to those of havs/svs, hardkeys backlight on/off etc.
Again:
Class = minimal sequential write speed -> large files.
Random writes -> dalvik cache.
Hence, cards class only guarantees card will perform at it's rated speed with sequential writes. Unfortunately increase in sequential speed doesn't mean linear increase in random. Or put bluntly, this justifies why some people could run data2sd scripts with lower end cards better than the others who had better cards.
Click to expand...
Click to collapse
Saw you having a similar conversation with island3r in the insertcoin thread
AGAIN: "http://www.tomshardware.com/reviews/...i,2940-11.html
Scroll down to sequential write. As the class goes up (in direct relation to the card's brand quality) sequential write speeds go up as well.
Of course, for instance, a Kingston class 6 is going to be faster than a Adata class 10.
But as a general rule (especially between the same brand), the higher the class, the higher the speed."
This time I am better prepared, second. Couldn't find it last night for the love of God otherwise that argument would be over quickly.
EDIT:
sibere said:
Card class indicates minimum sequential write speed achievable on the card.
It's good for writing big files like movie files.
But when it comes to writing small random files, it can be the opposite. Many people reported having bad random write speeds with C10 cards whereas others reported better speeds with lower class cards.
It's a matter of sd controler quality and size of erase block size. Bigger and faster SD cards have bigger erase block sizes (up to 256Ko). It means that those cards need to write 256Ko of data even for writing a single byte on the card.
That's why C10 are not the best choice when it comes to DATA2SD or any kind of similar solution to short data storage on the Desire.
Click to expand...
Click to collapse
Now, I don't suppose that creator of data2sd scripts knows anything better than island3r or Tom's hardware, but that's just the way it is. Your choice whom are you going to take for serious.
I suppose you are going to start rant about 720p recording in dalvik thread like he did?
Thanks for the replies. But I think i will leave it where it is on the internal memory because my SD-ext is only 512mb and I am running out of space
erklat said:
This time I am better prepared, second. Couldn't find it last night for the love of God otherwise that argument would be over quickly.
Now, I don't suppose that creator of data2sd scripts knows anything better than island3r or Tom's hardware, but that's just the way it is. Your choice whom are you going to take for serious.
I suppose you are going to start rant about 720p recording in dalvik thread like he did?
Click to expand...
Click to collapse
Nah... I was just messing with you
I'm a noob really and don't understand much of this "data2sd scripts" and all that malarky. From my personal experience I was using a class 4 8gb with the dalvik on the card and found no issues with it. No lag's or pauses or anything. I've since moved to a class 10 16gb and decided to keep my dalvik on the nand because I don't need the extra space for 100's of apps like some people do - my needs for apps are simple
IF I did have a class 2 I'd be a bit worried about putting the dalvik onto it, but like you said, it depends on the random read/write speeds the card can handle.
ps: you should have had that answer the other day with island3r
zedmarcus said:
ps: you should have had that answer the other day with island3r
Click to expand...
Click to collapse
As I said I couldn't find it last night. I have no doubt he will see it here too. This puts a nice finishing touch to who was right and who was wrong so I have no need to reignite that discussion again

SD Booster on Ace 2

Has anyone tried SD Booster on the Ace 2? I found a small mention of it in the gaming thread. I was just wondering if it's worth bothering with. I've got a 4GB Class 4 card. The reviews look pretty positive on the Play STore.
dont use it. i used it and it messed up my phone. i ever created a thread here about it and i still havent gotten my phone working because of it. it will seriously **** your phone up and is not worth it.and if anyone reading this knows how to fix it can they please reply on the thread above or send me a message please
me too
you mean like after intalling it and rebooting after takes a lot and when finally booted. evrything force closes.it happen to me also.
the thing that did was factory resret and recover a backup i have taken before.also only a factory reset will work
Interesting. I think I'll give it a miss then! I'm always wary of these alleged 'performance enhancing' apps. I guess for what might only be a small improvement in performance it's not worth the bother. Thanks for the replies.
Does anyone know of any performance boosters that are actually worth bothering with? I mean my phone runs okay, but it's just that temptation to squeeze every last bit of pwer out of it, ha.
for me the best thing i done was rooted it, removed all the crapware that came with it like the network apps and some of the samsung stuff. i installed an app called setcpu which automatically controls the speed of the cpu. set it as profile: screen off and max: 400000 or 200000 min:20000 priority: 100. it works great for me and i can get great times except when i use a 3g connection
i recommend just upgrading your ext sdcard to a class 10,(and swap storages after) would help. (that is, if you can get one.)
teddytsen said:
i recommend just upgrading your ext sdcard to a class 10,(and swap storages after) would help. (that is, if you can get one.)
Click to expand...
Click to collapse
A REAL class 10 card. You can't believe how many off-brand "class 10" cards are on the market that are slower than the class 4 I bought sometime in 2006...
Ryccardo said:
A REAL class 10 card. You can't believe how many off-brand "class 10" cards are on the market that are slower than the class 4 I bought sometime in 2006...
Click to expand...
Click to collapse
Lol. Mines a sandisk . Ive never encountered fake class 10. Fake ones i saw were class 4 i think..
Sent from my GT-I8160 using Tapatalk
Thanks again for the tips guys, might look into that. ANd I'll be avoiding Sd Booster!

Yep, 633x does work!

OK, after searching for a good SD card... I found 100mb/s reads from top shelf SD cards made for high def recording to work well on our Note 8.0 devices.
Though I did the mistake of moving a few 500+ MB apps to SD, yes they were quick to transfer... just as slow at backing up to SD as before my tests.
One thing that is a hidden bonus... before, when I disabled saving cached images in a xposed module... there was considerable lag in showing album art from mp3 files. Now it is barely noticeable when scrolling at a fast pace. I had over 300MB of cached images before disabling android's built in feature. Now I can have my cake and eat it.
Since many are moving apps to SD, I recommend the fastest speed you can get, not the fastest you can afford.
remember write speeds will very, by burst mode and read/write verification, as with TI Backup. I would assume dumping data would be greater, just have no way to measure quantitatively, just seat of the trousers.
Here is a list of compatible ultra high speed SD cards.
PNY Elite 100MB/s (Able to get them in the UK, US is low availability)
Sandisk Extreme (Plus/Pro) 95MB/s
Toshiba Exceria 95MB/s
Lexar 633x 95MB/s
gooberdude said:
OK, after searching for a good SD card... I found 100mb/s reads from top shelf SD cards made for high def recording to work well on our Note 8.0 devices.
Though I did the mistake of moving a few 500+ MB apps to SD, yes they were quick to transfer... just as slow at backing up to SD as before my tests.
One thing that is a hidden bonus... before, when I disabled saving cached images in a xposed module... there was considerable lag in showing album art from mp3 files. Now it is barely noticeable when scrolling at a fast pace. I had over 300MB of cached images before disabling android's built in feature. Now I can have my cake and eat it.
Since many are moving apps to SD, I recommend the fastest speed you can get, not the fastest you can afford.
remember write speeds will very, by burst mode and read/write verification, as with TI Backup. I would assume dumping data would be greater, just have no way to measure quantitatively, just seat of the trousers.
Here is a list of compatible ultra high speed SD cards.
PNY Elite 100MB/s (Able to get them in the UK, US is low availability)
Sandisk Extreme (Plus/Pro) 95MB/s
Toshiba Exceria 95MB/s
Lexar 633x 95MB/s
Click to expand...
Click to collapse
I forgot to add... I have done extensive redirection of files and folders to SD card. I only moved apps over 200mb in size as their data was a bit large to be taking up system ram.
All my downloads, media, backups and internet cache are stored on SD card. About half of the apps allow default folder redirection, so an xposed plugin was used to set redirect system defaults for all media types and downloads. So far I do not see any noticeable lag as before using my old SD card. Though it was 20mb/s read/write capable. I am now at 95MBps / 45MBps
I am using Trickster Mod to push my cache to 2048 for added speed improvements.
gooberdude said:
I forgot to add... I have done extensive redirection of files and folders to SD card. I only moved apps over 200mb in size as their data was a bit large to be taking up system ram.
All my downloads, media, backups and internet cache are stored on SD card. About half of the apps allow default folder redirection, so an xposed plugin was used to set redirect system defaults for all media types and downloads. So far I do not see any noticeable lag as before using my old SD card. Though it was 20mb/s read/write capable. I am now at 95MBps / 45MBps
I am using Trickster Mod to push my cache to 2048 for added speed improvements.
Click to expand...
Click to collapse
Are they actual speeds, or speeds claimed by the manufacturer? Big difference. I'm not totally convinced the SD-interface itself is capable of such high speeds, but I would love to be wrong on that.
thany2 said:
Are they actual speeds, or speeds claimed by the manufacturer? Big difference. I'm not totally convinced the SD-interface itself is capable of such high speeds, but I would love to be wrong on that.
Click to expand...
Click to collapse
I have no real data as android does not have a good set of tools for proper testing.
I do know the speeds given are maximum. Given that doing random read/write functions will be much lower.
Though I did test a UHS-1 64gb that had 30mb/s read - 20mb/s write. It took 30 minutes to perform a full Ti Backup to SD.
On reading mp3 files, I noticed lag bringing up album art in Samsung music player with cache disabled.
With the 633x chip, and same full backup, I got 17 minutes backup time in Ti Backup.
On reading mp3, hardly any lag with album art in Samsung music player and cache disabled.
I assume the biggest issue is write speeds on the Ti backup test. As the 633x chip has 2x the speed for writes.
Seeing how my chip is working well, you may find a 600x chip will work just as good for less money. After all there is not much difference in speeds between the 600x and 633x chips at this time. Mostly due to write speeds. Getting a 65+ MB/s write speed is more critical than getting 90+ read.
OK, it has been some time since I posted...
Here is my update!
Ti Backup has been my sole source for testing, as my backups have been timed and consistent.
As for placing apps on the SD card, speeds are much better than expected. In some cases faster than having them loaded to the system.
Just there is no way to beat the speeds for backing up from system as there seems to be a bottle neck with the internal data channel and SD card. It shows up even more so with OTG connection. Could be the driver not hardware for SD slot, the USB2 connection will be limited as seen in OTG testing. Once I have 4.4.x installed from OTA, I can make a better run at testing SD slot performance.
GT-N5110 & GT-N5120 - 64GB 633x on board, Status Official on SafeRooted OEM ROMs with Wanam Xposed and RootCloak. The only way to fly 8+ hours!
This badboy don't play with Play & Triangle away!
gooberdude said:
I have no real data as android does not have a good set of tools for proper testing.
Click to expand...
Click to collapse
Android has several great tools for testing SD cards. SD Tools, A1 SD Bench, SD Card Tester. You can test a number of different ways. They're definitely better than just timing a backup.
dontsurf said:
Android has several great tools for testing SD cards. SD Tools, A1 SD Bench, SD Card Tester. You can test a number of different ways. They're definitely better than just timing a backup.
Click to expand...
Click to collapse
Well if you place a full back up on SD and also have apps and large files on SD as well, you are getting the best test as far as I am concerned.
Real world performance vs synthetic benchmarks has always been a pain in my side.
Giving such a review should hold up over numbers from bench mark apps. I scoff at numbers, as they are skewed into one thing or another. Just like Intel getting bigger numbers over AMD, Yet, AMD ran faster for my real world experience.
I'm old school, real world or no world for me! Its a wonder why I bother with the internet, as it is too virtual for any existence. HA!
GT-N5110 & GT-N5120 - 64GB 633x on board, Status Official on SafeRooted OEM ROMs with Wanam Xposed and RootCloak. The only way to fly 8+ hours!
This badboy don't play with Play & Triangle away!
gooberdude said:
Well if you place a full back up on SD and also have apps and large files on SD as well, you are getting the best test as far as I am concerned.
Real world performance vs synthetic benchmarks has always been a pain in my side.
Giving such a review should hold up over numbers from bench mark apps. I scoff at numbers, as they are skewed into one thing or another. Just like Intel getting bigger numbers over AMD, Yet, AMD ran faster for my real world experience.
I'm old school, real world or no world for me! Its a wonder why I bother with the internet, as it is too virtual for any existence. HA!
GT-N5110 & GT-N5120 - 64GB 633x on board, Status Official on SafeRooted OEM ROMs with Wanam Xposed and RootCloak. The only way to fly 8+ hours!
This badboy don't play with Play & Triangle away!
Click to expand...
Click to collapse
The point of benchmarking is to put the system through the same tests under the same circumstances and give equipment and equal footing. That's probably not what you're doing with titanium backup. There's a bunch of variables that could skew your results when you just time it like that. Without a proper benchmark is probably as valuable as reading the max read writes from the packet.
dontsurf said:
The point of benchmarking is to put the system through the same tests under the same circumstances and give equipment and equal footing. That's probably not what you're doing with titanium backup. There's a bunch of variables that could skew your results when you just time it like that. Without a proper benchmark is probably as valuable as reading the max read writes from the packet.
Click to expand...
Click to collapse
Wha??? Now freak'n way dude! I know I have been in the business for 40 freak'n years. Cache, and background services always screw up benchmarks, and there is no way to make equal be equal.
As for my tests... I have 3 gb of apps and data on system. 1 gb of data on SD, and 300mb of apps on SD. Do the math, a full backup on Ti making the backups on SD, will allow both random read writes which are critical, and sequential read/writes as a mix. What I stated from the beginning all being equal moving from a fast 30MB/s card to 633x allowed for cutting my time almost in half.
I have done, 5 tests back and forth and the times are all the same. Seeing is believing since I have already noted, disabling cache made Samsung music player slow at reloading album art, while the 633x chip acts like cache is enabled.
Again do the math... I doubled write speed from chip specs, and got results along with showing reads are much faster by how the apps function.
A whole industry disagrees with you, but whatever.

Categories

Resources