Linpack Scores vs. Other Phones - myTouch 4G General

How come the G2 beats the mytouch 4g
http://www.greenecomputing.com/apps/linpack/linpack-top-10/
http://www.greenecomputing.com/apps/linpack/linpack-by-device/
we get lower scores even when overclocking to 1.8ghz

the one thing I noticed was from the looking at the notes only one of the results appears to come from a MT4G running "vanilla" android, even highly optimized Sense takes up alot of our RAM and such... perhaps our Linpak scores will starting coming up when we get our AOSP ROMs more stable and optimized, this is NOT a knock on any of our kickass developers.

but even with CM 6.2 i get ~60's.
X.x and we have a lot more RAM that the G2

Bleh, bench tests. You like your phone? GREAT! Enjoy it.

Related

Is Android 2.2 Froyo really all that its cracked out to be?

From the Droid X results and benchmark it seems that Froyo did not do to the Droid X as it did for all the phones that are running snapdragons.
Some people are saying that it's because the JIT was made to run on a snapdragon.
So what is up with the benchmarks? Are we not going to see our CPU speeds increase 3 to 5 folds? Is anyone else curious about this topic?
http://www.greenecomputing.com/apps/linpack/linpack-by-device/
i went from linpack for max 9.6 with oc kernal on 2.1 plus voodoo lag fix to 13.9 on froyo, no lag fix etc. so i say its faster.
nephets0 said:
i went from linpack for max 9.6 with oc kernal on 2.1 plus voodoo lag fix to 13.9 on froyo, no lag fix etc. so i say its faster.
Click to expand...
Click to collapse
Yeah, but how do we score in the 40s and 60s? I hope our ROM authors do a better job than Samsung has.
The Droid X hasn't even been optimized yet, for some reason 2.2 isn't doing what it did to the Nexus one for Droid X and Galaxy.
Voice Search is pretty sweet: "text jenny hey what's up?"... bam, done.
Flash support is neat for certain websites.
Auto-updating apps...
There is alot to froyo than just a speed improvement (although we wish it was more of a speed improvement...)
With JI6(Froyo), OCLF(Lagfix), and rooted I'm pretty much as happy as I can be until cyanogenmod comes out.
I know benchmarks are useless, but the quadrant score I got more than doubled the stock 2.1 score at around 1900.
I haven't played around enough yet to notice any bugs, but nothing significant has jumped out at me. USB tethering and Mobile AP both work.
NO issues here...I don;t use synthetic benchmarks wither in ,my PC builds or on these devices. I go strictly by real word applications. In that regard the user experience shell and speed is better on all counts. GPS showed me 10 Sats today. I never could see more than 6-8 on the best day.
The user interface is 10X better..........
Haha oh yeah, I really want to find out if the Leak (Beta) update is really worth it, just seriously might wait for the actual update. I read in the thread that some people were reporting a CPU clock of 400 Mhz, and just bugs, which "should" be fixed by the actual update
I was really hoping that our CPUs would boost 3 to 5 times as much.
The snapdragons with 2.2 have unbelievable to near incredible results. Their CPU benchmarks have more than triples in most cases.
I'm just really trying to get to the bottom of why the Hummingbird and OMAP seem to lack this potential.
Well it's still just a leak. Hopefully the full Froyo update will roll out at the end of October. A post on the Androidforums has me thinking that end of October is when it will be released.
Regarding the Samsung Dock:
If you search really hard on the Samsung site, you find that that 3.5mm jack is "line out" and won't be functional until the Android 2.2 release.
Click to expand...
Click to collapse
And from looking around for the dock, I found two sites that have it for pre-order with estimated availability towards the end of October.
As a former Nexus one owner, I can honestly say that the froyo benchmarks are in no way indicative of real world performance. My linpack went from 7 to 40 and I saw exactly zero speed up in real world daily use. There very few fringe apps that see any increase, mainly games and only a select few of those.
Sent from my SAMSUNG-SGH-I897 using XDA App
I had another experience. The performance boost FroYo gave me in Nexus One was very noticeable in "real world" situation.
However, I do fail to see that ratio on Captivate.
Linpack score: 14.031
Froyo be damned, the improvements in the Samsung version of it make it more than worth it.

[Q] Will Gingerbread be optimized for the Galaxy S phones?

Now that Gingerbread has been written by google for a Galaxy S phone that is very similar in specs to the Galaxy S line, will it be fully optimized?
Will we see big jumps in benchmarks like the Nexus One did when it received Froyo? Is the reason why the Nexus One and its variants received a big benchmark boost was because Android wrote Froyo for it?
I dont think 2.3 will come to captivate....And also, would the GPS been fixed??? Im going to sell my captivate and buy some HTC....
That is ridiculous situation...we are a several months waiting for 2.2 and so far nothing...We are several months waiting for some GPS fix...and so far nothing yet....
AT&T does not care about the customers if they are confortable with the 2.1 only or not....Neither Samsung....WE ARE LOST....hurt me to say that...
But thats how i feel...IM FEEL LIKE A FOOL...
sorry, dont get me wrong....
I doubt with Samsung's record of slow updates and bug fixes we'll ever see an official port of Gingerbread to the Galaxy S. Custom ROM's much more likely but I guess they will be a while.
As for speed increase... Don't hold your breath. I've got the official Froyo 2.2 running on my UK Galaxy S and it boosted my Quadrant benchmark score from 800+ (Eclair) to only 900+ (Froyo) (Nexus One with Froyo 1250). Custom ROM's apparently give better scores. Problem? Samsung again... they use a different file system on the Galaxy S (and I believe the American equivalents) than the official Android one and this slows the device and causes most of the lag problems on it. Watching the Quadrant benchmark the Galaxy flies through the graphics and CPU tests and grinds to a halt when it tries to complete the write to memory test... Froyo didn't help and I doubt an official Gingerbread update would either. I'm afraid flashing an unofficial ROM is the only way to get speed. It can be risky though...
Fizzig said:
I doubt with Samsung's record of slow updates and bug fixes we'll ever see an official port of Gingerbread to the Galaxy S. Custom ROM's much more likely but I guess they will be a while.
As for speed increase... Don't hold your breath. I've got the official Froyo 2.2 running on my UK Galaxy S and it boosted my Quadrant benchmark score from 800+ (Eclair) to only 900+ (Froyo) (Nexus One with Froyo 1250). Custom ROM's apparently give better scores. Problem? Samsung again... they use a different file system on the Galaxy S (and I believe the American equivalents) than the official Android one and this slows the device and causes most of the lag problems on it. Watching the Quadrant benchmark the Galaxy flies through the graphics and CPU tests and grinds to a halt when it tries to complete the write to memory test... Froyo didn't help and I doubt an official Gingerbread update would either. I'm afraid flashing an unofficial ROM is the only way to get speed. It can be risky though...
Click to expand...
Click to collapse
Quadrant is a poor indicator of overall performance, which is why an ext2-loopback lagfixed rom will show astronomical scores (2300+) in Quadrant - because it is essentially telling quadrant exactly what it wants to hear. The score is artificial and doesn't reflect how the device will actually perform (not to imply that the lagfixes don't make the phone more responsive or anything, simply making the point that quadrant is a poor benchmark).
The differences you'll see between 2.1 and 2.2 are an increase in processing power and battery life due to the JIT. It's a bit difficult to test this due to outside battery-eating variables (things syncing over the network in the background, quality of reception and radio power levels, etc.), so your ability to notice a difference may vary depending on your phone usage style and environment.
Also, you can use linpack to get an idea of the processing power increase. You'll notice that it doubles between 2.1->2.2. I should warn you that it's not really comparable to the Nexus One, simply because the cpu architecture is different and linpack is geared to take advantage of it - the n1's snapdragon has a 128bit simd fpu whereas the hummingbird has a 64bit fpu, so the increase in speed will show up as 4x-5x for the n1 between 2.1->2.2. Again, this does not translate into a 4x increase in real world performance. I only mentioned linpack to demonstrate the relative speed increase between stock 2.1 for the galaxy S and 2.2, and to show that the JIT is indeed boosting the processing speed.
SlimJ87D said:
Now that Gingerbread has been written by google for a Galaxy S phone that is very similar in specs to the Galaxy S line, will it be fully optimized?
Will we see big jumps in benchmarks like the Nexus One did when it received Froyo? Is the reason why the Nexus One and its variants received a big benchmark boost was because Android wrote Froyo for it?
Click to expand...
Click to collapse
To my understanding, 2.3 is essentially a more optimized 2.2. I doubt that the results of the optimizations will be as large as the introduction of the JIT was in 2.2, but every little bit helps - and look on the bright side, it won't be any slower than 2.2.
I don't think we'll really know how well the Nexus S roms will run on our phones or how easy they'll be to port over until we actually get our hands on an NS rom (still don't know what filesystem it uses or how big of an obstacle the filesystem will present). However, since the fundamental architecture is so similar, I don't really expect many problems and I expect the 2.3 builds to run great.
Edit: I don't expect Samsung or ATT to release 2.3 for our phones. I actually wouldn't be surprised if ATT refused a gingerbread update for our captivates, even if Samsung offered it. What I meant above was that I expect whatever custom roms we cook up based off of the NS builds to run great.
When do we start speculating about Honeycomb?
alphadog00 said:
When do we start speculating about Honeycomb?
Click to expand...
Click to collapse
When someone in our family (Galaxy S Line) gets it, which will definitely be the Nexus S.
SlimJ87D said:
When someone in our family (Galaxy S Line) gets it, which will definitely be the Nexus S.
Click to expand...
Click to collapse
Why wait... it is all guessing anyway. I think anything we say about Honeycomb has an equally good chance at being accurate as what we say about Gingerbread.
alphadog00 said:
Why wait... it is all guessing anyway. I think anything we say about Honeycomb has an equally good chance at being accurate as what we say about Gingerbread.
Click to expand...
Click to collapse
I don't think you really understand my question.
The nexus one was a developer phone that had a snapdragon in it. Google, the creator of Android, directly engineered their software to be optimized on the device because it was their developer device.
Now that a Galaxy S phone is a developer phone, I was wondering if anyone knew anything about the software engineering side to the Nexus one to guess if google built 2.3 from the ground up for the Nexus S, or can/could heavily optimized the code for it.
Now you're question is to speculate about Honeycomb, speculate what? There's nothing to discuss about it, but my question is legit from an engineering stand point. So I'm still left wondering if this is the case or not, I wonder if there is anyone that can enlighten me.
Would be a nice pipe dream for an offical update.
Sent from my SAMSUNG-SGH-I897 using XDA App
SlimJ87D said:
I don't think you really understand my question.
The nexus one was a developer phone that had a snapdragon in it. Google, the creator of Android, directly engineered their software to be optimized on the device because it was their developer device.
Now that a Galaxy S phone is a developer phone, I was wondering if anyone knew anything about the software engineering side to the Nexus one to guess if google built 2.3 from the ground up for the Nexus S, or can/could heavily optimized the code for it.
Now you're question is to speculate about Honeycomb, speculate what? There's nothing to discuss about it, but my question is legit from an engineering stand point. So I'm still left wondering if this is the case or not, I wonder if there is anyone that can enlighten me.
Click to expand...
Click to collapse
And you don't understand my point: No one knows. We might as well speculate about honeycomb. The Nexus S has a different momory mudule iNand not moviNand. What impact will this have? No one knows until they have phones in hand.
Sent from my MB520 using XDA App
alphadog00 said:
And you don't understand my point: No one knows. We might as well speculate about honeycomb. The Nexus S has a different momory mudule iNand not moviNand. What impact will this have? No one knows until they have phones in hand.
Sent from my MB520 using XDA App
Click to expand...
Click to collapse
What about the CPU though? The memory for the Nexus One and Droid Incredible were different, but yet because they shared the same CPUs they received similar benchmark scores in that department.
What does it matter everyone is going to flash custom ROMS of Gingerbread when/if it ever somehow leaks for our phones.
Sent from my axura phone with Gingerbread keyboard.
Fizzig said:
As for speed increase... Don't hold your breath. I've got the official Froyo 2.2 running on my UK Galaxy S and it boosted my Quadrant benchmark score from 800+ (Eclair) to only 900+ (Froyo) (Nexus One with Froyo 1250). Custom ROM's apparently give better scores. Problem? Samsung again... they use a different file system on the Galaxy S (and I believe the American equivalents) than the official Android one and this slows the device and causes most of the lag problems on it. Watching the Quadrant benchmark the Galaxy flies through the graphics and CPU tests and grinds to a halt when it tries to complete the write to memory test... Froyo didn't help and I doubt an official Gingerbread update would either. I'm afraid flashing an unofficial ROM is the only way to get speed. It can be risky though...
Click to expand...
Click to collapse
This is one of the things that I am optimistic about. Android 2.3 adds support for devices with large internal storage capacities - my understanding is that it was Samsung's poor attempt to hack that support into 2.1 that introduced the issues you mention above. I'm hopeful that this means 2.3 would eliminate the need for lag fixes, and that the better support for some of the cutting edge hardware in the Galaxy S Phones being built into Gingerbread will make it much easier for Samsung to push 2.3 out for our phones.
AdamPflug said:
This is one of the things that I am optimistic about. Android 2.3 adds support for devices with large internal storage capacities - my understanding is that it was Samsung's poor attempt to hack that support into 2.1 that introduced the issues you mention above. I'm hopeful that this means 2.3 would eliminate the need for lag fixes, and that the better support for some of the cutting edge hardware in the Galaxy S Phones being built into Gingerbread will make it much easier for Samsung to push 2.3 out for our phones.
Click to expand...
Click to collapse
Can you elaborate? Coz AFAIK RFS has nothing to do with Samsung's attempt to make a pitiful 16 GB work as internal sd card..
From what I understand, the movinand flash which Galaxy S , when used in RFS performed horribly when doing sync operations (I think I got it from the thread ryanza posted), so, the new flash might as well be Samsung's attempt to correct that error, instead of replacing the (seemingly crap) RFS ..

[Q] (Q) I9000 2.3 kernel for the cappy?

Is there an JV1 or other gingerbread kernel from 19000 for the cappy, or is there someone working on it. CM7 isnt that great with battery and many other things i figer the I9000 kernels for 2.2 kernels work as well if not better for cappy.
was hopeing to try I9000 gingerbread on my cappy.
thanks pm or post here.
the source code for the i9000 gingerbread has not been released yet since there hasn't even been an official release of the i9000 gingerbread. without the source code, the kernel cannot be reoriented to have the buttons working and the correct headset logic.
Sent from a phone, playing a phone, disguised as another phone.
I see this makes sence, so its a waiting game still! thanks
you can remap the buttons after you flash the rom and disable auto rotate, that will fix the ui stuff but you will not be able to use your headphones.
if you want to try it, people were doing this with with i9000 roms since day one, there are fewer active people that are experimenting now but it should still be possible. i would if i had a second device but i like haveing music on my headphones.
if i feel up to it i might do it this weekend to write an idiot proof procedure and post it in general section but i doubt ill keep it.
it is a shame we dont have the source because from the benchmarks i see with cm7 it seems that the hummingbird responds very well to gingerbread, this chip keeps getting better as android catches up, infact in a few ways it is faster than some dualcore platforms, maybe not overall but there are areas that the hummingbird is untouchable. forget that omap chip in the driod x and the snapdragon. the hummingbird excells in areas that are not measured in market apps, but still holds its own against everything else in other areas, and beats them when matched with gingerbread.
it is funny that every time i see a benchmark that shows the strength of the hummingbird in a tech article someone brings up quadrant and the omap chip and say that the benchmark is inacurate but if you pay for quadrant you see that the droid x only beats hummingbird in areas that dont involve the cpu! dumb asses dont know when they are beat. i hate motorola and there dishonest advertising. hummingbird is weak in floating point operations and that's about it.

[Q] Most recommended phone in terms of ROM support?

I'm currently on a N1 and previously had a Hero (GSM) and I love the love that these two phones have gotten from the devs
Just quite curious with regards to the current phones though, which are the ones that are just as similarly loved by the developers in terms of a wide range of 3rd party roms?
I'm incredibly happy with my vibrant. I had the g1 for like 2 years, and the development was awesome. Then I went to the Mytouch slide and the devving on that sucked ass. That was pretty much a year of cyanogenmod. Now I'm on the Vibrant and this phone is a beast. I don't see the hummingbird chirp getting outdated anytime soon. The only complains I have about the Development is that the roms are very similar, but they're awesome. I used to be a daily flasher coming from the g1, and I was at first with the vibrant, but after flashing Bi-winning V3.0 I have no desire to flash something else. I get approx 2100's Oc'ed to 1.2 ghz, and my battery lasts easily 15 hrs with use
I actually think the desire and desire HD. In the future i guess that will be de sensation

New Kernal for Inspire 4G in the works Along with a new theme

Hello everyone! I'm so glad to see all of the development on the android xda. I used to develop for WebOS, it's gone nearly completely quiet, even with WebOS Internalz R.I.P Palm. (Though Im still working on getting my Pre 3 testing device moved over to CM7. anyway, I've managed to benchmark, Now 25 times with Quadrant, Smart 2011, and a few others. Quadrant and Smart gave very similar results, Though you probably won't believe me until I'm able to upload screen shots.
Running CM 7.1.1
Kernel: K311wOS .9.5.8 <customized kernel I began work on along time ago, just finally found all the mods and programs and most importantly CM 7.1.1 that can handle it. >
OC 2.18ghz @ 92.2 F degrees (never gone above 98, unless charging.) And thats with me pushing it to reset.
Quadrant Low Benchmark: 2870 Quadrant High Benchmark: 4230 (Yes 4230) Average Benchmark: 3540
Well, Having only one snap dragon 1ghz, in comparison to my pre 3 running a qualcomm 1.4ghz stock and my 64GB Touchpad..running 2. with uber kernals lol, but hey it's a start to my new found, android digital life.
But yeah THANKS to the guys over at CM! Hopefully, I'll getover my n00b status on the XDA forums and can move to the developers..
But If there are better ways to test and confirm speeds, etc..please let me know, I'm trying to run this Inspire through a gauntlet because after seeing the rankings, It was Me, surrounded by Galaxy's and Zoom's, and Nexus lol I kinda find it hard to believe..
Thanks! and Hope to be getting all of you Inspire/Desire users a new kernel soon, however, I need to find a desire to test with as well, or if any developer is willing to test around with it, let me know!
Kobos311
Sounds good....SHOW ME THE MONEY!
yea... this thread is worthless without pics! lol
Welcome Kobos from one Palm refugee to another.

Categories

Resources