Related
According to this link, the upcoming Epic 4G has an I/O benchmark that is much higher than Vibrant (and therefore the similarly handicapped Captivate and international and Bell Galaxy S i9000). I started wondering whether this is due to actual hardware differences, or if there have been tweaks incorporated into the firmware that are helping. Then I found an Epic 4G system dump here. I haven't personally downloaded the dump from that link, because even if I did, I wouldn't know how to find anything. Could someone with more ability study this dump and possibly learn something useful for us?
I have also heard reports that Epic 4G has a GPS that works well. This is less of an issue for me personally, as I am on JH2, and find my GPS adequate for the occassional use I have put it through. This is another area that might be worth looking at.
Edit:
alternate Epic 4G dump download location
Interesting to say the least. Wonder if they dumped RFS or if they just figured out how to optimize it. The hardware differences between the devices shouldn't be significant enough to show that big of an I/O difference.
Who knows, maybe this indicates that Samsung has an idea what's going on with the performance issues and will eventually filter it down to the SGSs variants
Zilch25 said:
Interesting to say the least. Wonder if they dumped RFS or if they just figured out how to optimize it. The hardware differences between the devices shouldn't be significant enough to show that big of an I/O difference.
Who knows, maybe this indicates that Samsung has an idea what's going on with the performance issues and will eventually filter it down to the SGSs variants
Click to expand...
Click to collapse
QFT .................
I am going to go ahead and download the system dump zip file for archiving purposes. I'll probably put it up at an alternate download location, once I have it.
I'm not 100% sure on this but I think the Epic's total internal memory is NAND (1GB) unlike the other phones that use an internal SD card. Much different memory transfer speeds.
Thats why the one lag fix works so well, it copies data/data to the internal NAND.
I am already running the gps fix from the dump(haven't tested it yet). There is no lag fix as far as I can tell. The epic gets 850+ in quadrant. If it had a lag fix it would get a lot higher score.
I believe derek4484 is right on the 1gig of NAND. I just checked it out on a review site.
Sent from my GT-I9000 using XDA App
Yeah doublechecked derek too... it appears it comes with 1GB ROM only, all other storage is via microSD.
False alarm =P Samsung didn't fix or optimize anything, they just used a cheaper, more shortsighted method to avoid having to give up their precious RFS
Those test results are questionable. You can click on the link under the graph and compare other phones. They have Droid X edging out SGS in linpack: SGS- high 7's; X- low 8's.
Sent from my GT-I9000 using XDA App
tiger4j said:
Those test results are questionable. You can click on the link under the graph and compare other phones. They have Droid X edging out SGS in linpack: SGS- high 7's; X- low 8's.
Sent from my GT-I9000 using XDA App
Click to expand...
Click to collapse
I dont think they are right either. If the phone had that fast of a file system it would kill in quadrant. The epic has been reported to only get 850+ in quadrant.
Unless you're actually running an epic, the GPS fixes out there only dump the GPS files, so it wouldn't do anything to indicate the speed of the actual Epic in terms of file system performance =P
why would they let sprint have the nicest version of the galaxy s ugh -___-
Zilch25 said:
Unless you're actually running an epic, the GPS fixes out there only dump the GPS files, so it wouldn't do anything to indicate the speed of the actual Epic in terms of file system performance =P
Click to expand...
Click to collapse
Two different issues regarding GPS locking and file system I/O performance
rajendra82 said:
Two different issues regarding GPS locking and file system I/O performance
Click to expand...
Click to collapse
I know, I was commenting on sheps post =P He seems to be saying that using the gps fix from the epic would alter file system performance.
No I am saying people with epics have tested them with quadrant and are getting 850+. I cant see it getting so high scores from laptopmag.com under the file sytem benchmarks but only get 850+ in quadrant. laptopmag.com says the droid x file system benchmarks the same but it gets over 1200 in quadrant with slower cpu and gpu.
shep211 said:
No I am saying people with epics have tested them with quadrant and are getting 850+. I cant see it getting so high scores from laptopmag.com under the file sytem benchmarks but only get 850+ in quadrant. laptopmag.com says the droid x file system benchmarks the same but it gets over 1200 in quadrant with slower cpu and gpu.
Click to expand...
Click to collapse
Ah okay Thanks for the clarification. The posts were a little confusing as to what you were saying.
shep211 said:
No I am saying people with epics have tested them with quadrant and are getting 850+. I cant see it getting so high scores from laptopmag.com under the file sytem benchmarks but only get 850+ in quadrant. laptopmag.com says the droid x file system benchmarks the same but it gets over 1200 in quadrant with slower cpu and gpu.
Click to expand...
Click to collapse
FYI, Quadrant scores MEAN JACK
andy2na said:
FYI, Quadrant scores MEAN JACK
Click to expand...
Click to collapse
Unless youre comparing the same phone. Especially when one with lag vs one without it get the same score. Even though quadrant isn't the best thing to use you can still use it to show an improvement after a fix or a mod from the base number. So his question was why do the phones get the exact same score but one has lag.
Sent from my GT-I9000 using Tapatalk
Clienterror said:
Unless youre comparing the same phone. Especially when one with lag vs one without it get the same score. Even though quadrant isn't the best thing to use you can still use it to show an improvement after a fix or a mod from the base number. So his question was why do the phones get the exact same score but one has lag.
Sent from my GT-I9000 using Tapatalk
Click to expand...
Click to collapse
Quadrant only tests throughput bandwidth not latency, etc. The internal NAND may be just as fast in terms of MB/s but it could have lower latency or other charecterists that Quadrant does not test.
So DK28 is supposedly 2.2.1 but I ran Quandrat with the stock rom and it comes up 873, almost exactly the same as Galaxy S 2.1, leading me to believe that JIT is not in the rom. Correct?
avenger213 said:
So DK28 is supposedly 2.2.1 but I ran Quandrat with the stock rom and it comes up 873, almost exactly the same as Galaxy S 2.1, leading me to believe that JIT is not in the rom. Correct?
Click to expand...
Click to collapse
I may be wrong but quadrant:
1) Isn't the best way to base performance
2) Doesn't really show improvements that JIT provides
Download linpack if you want and your score should be around a 14 or so, compared to the 8.5-9 on 2.1
Quadrant isn't a real life benchmark, and is highly flawed. Check out Quantum's "warp speed edition" ROM for proof.
Sent from my SPH-D700 using XDA App
I got 1205 with stock dk28 but I have noticed others reporting lower scores. Variability in testing with a bad benchmark perhaps?
I've got 1839 lol... For some reason I doubt that's accurate
I did a quadrant test with a stock epic 2.1 and an epic on dk28 and the scores were pretty much the same every time. The major difference was that the Froyo epic ran the cpu benchmarks much faster but hung during the first I/O test. The 2.1 epic ran everything about the same speed and completed the whole test much faster than the froyo epic.
I also noticed this and posted in another thread...
to quote myself
I was getting around 8.3xx in 2.1, and now I'm getting 13.7xx in 2.2.
(I used the 1506bdf2e04b.update-SPH-D700_DI18ToDK28.zip renamed to update.zip method.)
Saw a video the other day showing a nexus 1 scoring 20+ on 1st run and 40+ on 2nd.. My friends Evo (with 2.2) scored @ 38 or so..
Quadrant still shows me well under all of the 2.2+ phones listed.
Click to expand...
Click to collapse
I expected a bit more of an increase..
I too am wondering if the JIT was left out of our Froyo (if that's even possible, I'm still a total nube to andriod stuff).
What exactly is JIT, does it speed the phone up?
Sent from my SPH-D700 using XDA App
Just In Time compiling...
From what I know about .Net (I'm a C# coder), it allows for compilation of the app for the specific CPU during the 1st run.
Someone please correct me if this differes from andriod's JIT...
the epic has fast computation hw between the cpu and gpu, it could be the case that the JIT benefit doesn't show in the benchmark margins.
the device is a fast one, all compared.
Ok so ive rooted this toy, flashed the 1.2ghz kernal and ran quadrant on it. I hit anywhere from 1080 to 1180. Have setcpu on demand at 800 min and 1200 max so why on earth is my rooted nook color running android 2.1 and a flashed 1100mhz kernal hitting wayyyy higher scores? (Around 1280) doesnt seem logical at all. Specially cause this is an actual tablet and thats just a rooted e reader. Did I do something wrong?
Sent from my SCH-I800 using XDA App
Quadrant is a bad benchmark. It weights i/o far too heavily.
The Tabs RFS filesystems bring its Quadrant score way below where it should be, though in actual use it's very fast.
If you move to EXT4, your Quadrant scores will shoot up by around 80%.
Regards,
Dave
O sweet love of mary. 80% you say. Ok ive seen some posts redarding this ext4 thingymabob. Ill check it out. So if im understanding you correctly by changing to this my tab will perform even better than it does now? Or ill just see a genaric increase on this quadrant test that really means nothing? Sorry im noobish lol
Sent from my SCH-I800 using XDA App
Well, my tab is currently on ext4 and my quadrant scores completely destroys my rfs score taken earlier.
However, that said, i don't really feel any performance improvements except the satisfaction knowing that my toy is now on an open and faster file system as opposed to a propriatory and slower one ;-)
Sent from my GT-P1000
Ok so ive looked at some threads containing info but I see no post telling me how to do this magic. Could one of you nice, young, kindhearted souls link me to where I need to go to put this thing on my verizon cdma galaxy tab? Please and thank you sirs?
Sent from my SCH-I800 using XDA App
Quadrant scores aren't as accurate as people think you know...
Sent from zombie infested Gingerbread.
There's an app called OCLF that will transparently add an EXT4 area on the RFS filesystem without needing to convert the whole thing to EXT4, giving you EXT4 IO performance in an easily installable and reversible way. Probably your best choice for trying it out. Otherwise, for the complete conversion, you could try this Modaco ROM/Kernel, which includes a complete conversion to EXT4.
As far as performance differences go, it makes a huge (10x) difference to the Quadrant IO score (measured before and after figures), but no measurable difference to actual performance (measured real-world usage figures). Some people claim to notice a subjective difference, but I never did.
Im using the modaco rom and kernel with ext4.
And it does make a huge difference. I get around 1800 in quadrant now.
conan1600 said:
Ok so ive looked at some threads containing info but I see no post telling me how to do this magic. Could one of you nice, young, kindhearted souls link me to where I need to go to put this thing on my verizon cdma galaxy tab? Please and thank you sirs?
Sent from my SCH-I800 using XDA App
Click to expand...
Click to collapse
hi,
please be aware that the modaco kernel (rfs/ext4) will not work on CDMA Tabs.
from modaco kernel thread
This kernel is ONLY tested on a UK Galaxy Tab. It MAY work on other GSM Tabs, it almost certainly WON'T work on a CDMA Tab. If you have a non UK Tab and want to test (and know how to flash back to a regular version), then go ahead and report your results. At your own risk.
Click to expand...
Click to collapse
regards,
mike
robertsydbrink said:
Im using the modaco rom and kernel with ext4.
And it does make a huge difference. I get around 1800 in quadrant now.
Click to expand...
Click to collapse
That's the point!
It makes a huge difference in Quadrant, which is a synthetic benchmark, but far, far less difference in real world usage - hence it is a crap benchmark!
Pre-EXT4 my Tab would pull around 1000 in Quadrant, whereas my Desire HD would pull around 1800. However in real usage, they seem to perform pretty much the same which is not too surprising as they are similarly specified. On EXT4, my Tab pulls around 1800 now, but still performs much like my DHD.
I so wish people would stop bandying about Quadrant scores because they are meaningless.
Regards,
Dave
If you look at the scores in quadrant , they are split by colour so you can see how good the graphics capabilities are for example and compare to other phones. The colour codes are at the bottom of quadrant by the way
Linpack is a better benchmark. Not perfect, but better
Not really - Linpack only tests floating point performance.
Regards,
Dave
Wow, I started a heated discussion lol. Well ive downloaded the one click lag fix but have not applied the ext2 tools as yet. Want to do more reading about it first. Obviously I want my yab to be the best it can be but I surely dont want to make it genericly better at the expence of my video grafix as one user said he suffered in that thread.
More reserch required
Sent from my SCH-I800 using XDA App
Taking Quadrant scores aside, when I made the switch from CF-Root to the Ext4 MCK, I really did notice a huge improvement in real world usage. I'm not sure if it is really because of the change in file system, but nevertheless thats the only big thing present in that kernel aside from CWM.
Everything said make sense and seems to be valid but let me share my experience.
I have Samsung Galaxy Vibrant and T-mobile Tab. Now both are using EXT4 file system and have fully functional recovery allowing for flashing straight from the phone.
Both now have Quanrant score around 1700-1800 and run very smothly.
Is I/O speed important? I think it is very important because task switching requires reading of huge chuncks of memory. Until read operation completed the user is essencially suspended. Multitasking is the major distinction of Android and lags associated with the tasks switching might be the most noticeable issue since its used so much. Converting of the RFS file system to EXT4 practically reduced lags to unnoticeable level. I don't need any better.
Another critical area for I/O is playing video and especially capturing HD video. The latter works only if I set internal storage and shut down all tasks.
This is work in progress but it seems it reached level of usability when most critical bugs eliminated.
It should be noted that the Tab is flashed with Rotohammer KM2 v1 ROM and Paul Obrien's kernel on a top. The kernel contains scripts converting the file system and flashing recovery. This combination works well, no issues so far.
Well I decided to give it a try and after install my quadrant is 2556 and linpack is 16.865 mflops at 1200mhz. Good scores but just numbers. I do however believe im seeing a bit of snap that I didnt have before when accessing my library. Still really unsure if this is a good thing as im not sure if I can use apps to sd anymore so I may uninstall at some point in the future but thanks to all you who helped the old man out. Atm im quite happy
Now lets get ta craka lackin on a 1.5 ghz update for our tabs
Sent from my SCH-I800 using XDA App
Of course you can use a2sd. Froyo does it by default
Sent from my Legend using XDA App
I think you are pushing!
Overclocking will raise processor/RAM heat dissipation and thus might lead to a catastrophic failure. It would be great to have temperature sensor on the board or at least measure current consumption from the battery but it is not easy.
Of course im a pushin lol. Only way to achieve is to try. Course ive read about the tab proccessor being able to handle 1.4 stable and can handle 1.6... Not that id know about these things first hand. Just taking the words of better men.
Just an old man with a little time to kill and a dream to be able to play facebook cityville on my tab haha. Well that and I have always enjoyed souping up my toys lmao.
Sent from my SCH-I800 using XDA App
Hi guys,
Since this group has been very receptive to Smartbench 2010, I'm hoping Smartbench 2011 will be accepted just as well.
I've been working on the new version of Smartbench for the past few weeks. This version is still BETA, so you won't find it in the Android Market yet.
Here's what's new:
- Dual-core processor support! (Ok, not relevant to you, but you can see how well Atrix and Optimus 2X performs on CPU intensive tests now... ). 'Pi', 'Mandelbrot' and 'String' tests now spawn 4 threads which are ran simultaneously. So on 1 core phone, each thread runs 4 times slower, dual-core phone will run each thread about twice as fast, etc. This also means, Smartbench 2011 is quad-core ready.
- Added "String Test". This test manipulates large strings and also performs some regular expressions.
- REMOVED File I/O tests. This test was somewhat inconsistent not just on Galaxy S phones but also on other phones. I don't think it's worth while to keep it.
- Auto-submission of results! At the end of the test run, all test scores are submitted to the server (well, assuming you have the internet connection).
- Result chart is now REAL-TIME! Watch the result chart grow in real-time when you re-visit the chart screen! Oh, for now, I am limiting the display of the results to 50 entries and this is based on 50 most popular device/configurations. Different clock speed counts as separate entries, and so does different Android versions.
- Donated users can see results without running the complete test first.
IMPORTANT: Do NOT compare 2011 results with 2010. They have different baselines and apps have been tuned differently. You WILL get lower Games Index scores because now it is base-lined on HTC G2 instead of Nexus One.
Here's the link to the latest BETA:
[Link removed]
As soon as you run the test, I will see results on my DB.
Please let me know if you find any issues. I'll try to get the final version up in the Market sometime this week.
Thanks much!
EDIT: Smartbench 2011 v0.7 has been released. It will now allow you to sort results by either Average, Productivity or Games scores. Please re-install this version.
EDIT2: Smartbench 2011 v0.8 has been released. It now offers a filtering option - you can see all results or just stock clock speed results. Mandelbrot test also has been extended for better precision. Again, please re-install this version.
EDIT3: I have just published Smartbench v1.0 in the Android Market - please install this version moving forward. Results submitted by the BETA version will no longer be stored in the server DB. Your support during the BETA period has been very helpful!
Just ran it. The Atrix's results look beast!
Sent from my SPH-D700 using XDA App
Overstew said:
Just ran it. The Atrix's results look beast!
Sent from my SPH-D700 using XDA App
Click to expand...
Click to collapse
Yup I can see your results - thanks.
Tegra 2 is a dual core, and it is Cortex A9 based, so it should perform very well on CPU tasks. But as you can see Hummingbird is doing rather well on Gaming Index.
I've defined the correct name mapping for Epic now. It should show the human-friendly name on the chart.
jus ran....
Acei said:
Yup I can see your results - thanks.
Tegra 2 is a dual core, and it is Cortex A9 based, so it should perform very well on CPU tasks. But as you can see Hummingbird is doing rather well on Gaming Index.
I've defined the correct name mapping for Epic now. It should show the human-friendly name on the chart.
Click to expand...
Click to collapse
Ya, cant wait until Sprint releases a dual-core phone.
Sent from my SPH-D700 using XDA App
I think Epic's result looks ok from my perspective given other devices on the list already...
I see couple of results at 1.2GHz. You guys are not running the stock ROM, correct?
FC's on my ACS Fozen rom
apatcas said:
FC's on my ACS Fozen rom
Click to expand...
Click to collapse
Where did it FC?
EDIT: My bad. I was trying to fix a small bug but introduced FC in the process. Try downloading it again - that version should fix the FC issue.
I'm on syndicate Rom.
Sent from my SPH-D700 using XDA App
Ditto on FC on Frozen (1.2ghz) ideas?
Closes right after GLJellyfish, can see the menu for a split second then FC.... hope it helps...
EDIT: new one works....1081/2600.... nice
Overstew said:
I'm on syndicate Rom.
Sent from my SPH-D700 using XDA App
Click to expand...
Click to collapse
Ouch. Looks like my code to detect the name of the custom ROM isn't always working. Not sure if there is a reliable way to get this info... Hmmm...
ungovernable1977 said:
Ditto on FC on Frozen (1.2ghz) ideas?
Closes right after GLJellyfish, can see the menu for a split second then FC.... hope it helps...
Click to expand...
Click to collapse
Yup this bug should be fixed on the new version which I posted few minutes ago.
Running the new ACS Frozen ROM (at 1.2Ghz)
My phone shows up as a SPH-D700 running Stock ROM on Android 2.2,
this is not the case as this is a 2.2.1 phone not running the stock rom. It is not grouped under the rest of the Epics.
whoa..
ran again oc'd..big increase on the productivity index, nice!!
Acei said:
Ouch. Looks like my code to detect the name of the custom ROM isn't always working. Not sure if there is a reliable way to get this info... Hmmm...
Click to expand...
Click to collapse
Ya, just to verify, im not getting any fcs on either version.
Sent from my SPH-D700 using XDA App
Testing on ACS Frozen
Just tried out the 2011 version on ACS Syndicate frozen with their OC kernel. Ran the test with setcpu set to 200-1000 and conservative governor as well as 1000-1200 and performance governor. No FC or errors at either setting. Cool to see more new devices in comparison.
You are going to see FAR MORE devices to compare with Smartbench 2011 in the future. Imagine, there will be tens of thousands of results if not more. I just need to figure out a sensible way to display all that in some meaningful way.
I can see someone ran Evo Shift 4G and HTC Bee!
does it detect the actual clock speed, or does it just know its an OC kernel? Cause somebody in here is dragging down our scores!
How about uploading your score being optional? like if say, ridiculous example, someone was listening to pandora and ran the benchmark on a whim...
ungovernable1977 said:
does it detect the actual clock speed, or does it just know its an OC kernel? Cause somebody in here is dragging down our scores!
Click to expand...
Click to collapse
It was mmeeeeeee.
Sent from my SPH-D700 using XDA App
A call for all the benchmark whores, this is your time to shine. Your e-peen will grow like no other. You can post your new quadrant scores, without anyone telling you that it's outdated!
Article link
Market link
AndroidPolice said:
When it comes to benchmarks, one name usually stands above all others: Quadrant. Even though it has been proven to be easily faked, there's just something about running it and see a 3000+ score show up at the end to make you feel your device is untouchable.
Despite its popularity, Quadrant has been missing one key feature: multi-core support. That all changes with an update that was pushed to the Market earlier today, which brings Quadrant up to version 2.0, though. Not only does Quadrant now support multiple-core chips, but the update also brings ICS support, GUI improvements, support for Polish, and a fix for slow frame rates on PowerVR SGX 540 GPUs.
Click to expand...
Click to collapse
If it would matter, mine would be 3547.
It also occurred to me, that I might have not posted in the correct subforum. If so, I kindly ask Scotty to beam me up and back down a bit to the left.
And in the meantime it also occurred to me, that someone else had posted a similar thread already. I can only blame the forums for not giving me relevant search results.
It's still crap, because it doesn't display per-test scoring so you can know WHAT changed a score.
For example, disabling per-file sync in the filesystem code gains you a few hundred points in Quadrant - in AnTuTu it's clear that this improvement is in database write tests. (It also doesn't skew the scores nearly as much. Quadrant's weighting sucks.)
With ICS 'LPB' i got 3914.
I got 4004 on i9100 with ICS XXLPB
However...
Quadrant reports that my device has only one core, which is quite an important mistake, and some of the graphics tests keep hitting the 60fps ceiling, which ruins the integrity of the test, assuming it is based on measuring the maximum possible framerate.
hedgepigdaniel said:
I got 4004 on i9100 with ICS XXLPB
However...
Quadrant reports that my device has only one core, which is quite an important mistake, and some of the graphics tests keep hitting the 60fps ceiling, which ruins the integrity of the test, assuming it is based on measuring the maximum possible framerate.
Click to expand...
Click to collapse
The s2 framerate is capped at 60 fps because of vsync, it can't be changed..
And Quadrant results are more accurate on quallcomm device.. Other than that results may vary and could even be manipulated, CMIIW
I suggest you to run nenamark2 benchmark instead.. It's free on the market and gives a comparable results too..
Sent from my GT-I9100 using xda premium
Guys, is the planet part of the test scores only about 38+ fps?
Sent from my GT-I9100 using xda premium
3875 on XLB remix 9.2
Using omega's ics rom...
Sent from my GT-I9100 using xda premium
Gorjess said:
The s2 framerate is capped at 60 fps because of vsync, it can't be changed..
And Quadrant results are more accurate on quallcomm device.. Other than that results may vary and could even be manipulated, CMIIW
I suggest you to run nenamark2 benchmark instead.. It's free on the market and gives a comparable results too..
Sent from my GT-I9100 using xda premium
Click to expand...
Click to collapse
The 60fps limit can't be changed, but quadrants graphics test can be made more challenging so that the limiting factor is power rather than the arbitrary fps limit. It's not an accurate test if its hitting the 60fps ceiling.
ephraim033 said:
Guys, is the planet part of the test scores only about 38+ fps?
Sent from my GT-I9100 using xda premium
Click to expand...
Click to collapse
mine stuck @ 45 fps for the same test, are you in ICS rom?
edit:
between got 4285 in LPB
4346 here and planet test is at 38 fps.
Sent from my GT-I9100 using XDA App
3510 on CM9 latest
bala_gamer said:
mine stuck @ 45 fps for the same test, are you in ICS rom?
edit:
between got 4285 in LPB
Click to expand...
Click to collapse
Nope, i'm using Hyperdroid ROM
Sent from my GT-I9100 using xda premium
Hmm... it seems to me that CM9 gets lower scores than other ICS ROMs. That, I think, can be explained by the lack of full kernel support. Let's hope for the better, when they will get their hands on the kernel sources.
Please don't discuss Quadrant scores in this forum. Search "quadrant" in GSII General for the 7 or 8 threads we've closed for the reasons why.
Thanks