I know that 3.0 is mostly UI improvements but I'm wondering if Google updated the JIT or made any other performance improvements with Gingerbread.
I know there was a CES presentation and we have a gingerbread ROM already... does anyone know if there were speed improvements made?
Oh I got 3.0 and gingerbread confused. I guess I'd be interested in hearing about both! lol
edit:
JVM speed. For Java developers, 2.3 adds a number of speedups, most notably a concurrent garbage collector. According to Google garbage collection pauses will be under 3ms, which is small enough not to be noticed in a 30fps or even 60fps game. New JIT optimizations make Dalvik code run even faster than before.
I remember when 2.2 was coming out we were able to use the JIT on our 2.1 ROMs... will we be able to backport 2.3's JIT?
FWIW, I compared CELBv3.9 to KGBv8 using Linpack @ min=max=748Mhz
(governor set to performance in both cases, but that shouldn't have any effect when min=max)
If you recall, that particular benchmark was/is pretty sensitive to whether JIT is operational.
The results? Over about 10 repeated trials each,
CELBv3.9: 4.75 - 5.05*
KGBv8: 4.68 - 5.28
There is an observable, but modest increase in the peak (maximum) value of the benchmark score in KGBv8 - roughly a 5% increase. But at the same time, note that the "scatter" in the benchmark actually worsened. I take that to mean that in real-world performance, it is unlikely that you will be able to notice a perceptible difference. Hard to know, really - synthetic benchmarks are pretty crappy predictors of the "pleasure of use" for an arbitrary application mix.
* Peak quadrant score was 429 for CELBv3.9 @ 748Mhz - quadrant currently fails in the 3D graphics tests in KGBv8
Thank you bftb0, very interesting results.
I think it's fair to attribute that scatter to the lack of optimization currently in place with that ROM. It is nice to see that it is capable of more though. Certainly when you use that ROM it's easy to see that the stable 2.2 ROMs are still going to give you better real world performance.
And idea as to whether we can backport the 2.3 JIT?
Hi everyone,
Like most people I've been trying to find that optimum balance between performance and battery life in my Aria. I'm running CM7.0.3 and the latest version of drowningchild's kernel. With that kernel I switched setCPU from ondemand to smartass, and I can already see the CPU spending most of it's time in either 400 or 480 MHz, but when playing a graphic intensive game it'll go up to 806 MHz and run great.
But, here's the problem. With things like less graphic intensive games, or playing videos, smartass doesn't want to scale up to a useful clock speed, and things get choppy. I switch back to ondemand and everything smoothes out again. I don't want to keep it in ondemand all the time though, because then it's spending half the time in 806 MHz and eating up battery unnecessarily.
So, is there a better governor option for me? Or maybe a way to "whitelist" apps within setCPU, to designate apps that always get the max clock speed?
tl;dr: smartass isn't that smart in certain situations. Is there a better way?
Check out the app "Tasker" in the market. It will allow you to reconfig the governors automatically based on a slew of complex conditions.
If you want smartass roll back to 7-5 of my kernel and it should be fixed as I tweaked it to jump around less in newer versions
I'm still messing around with the new kernel released & will add smartass very soon & hopefully optimized
drowningchild said:
If you want smartass roll back to 7-5 of my kernel and it should be fixed as I tweaked it to jump around less in newer versions
I'm still messing around with the new kernel released & will add smartass very soon & hopefully optimized
Click to expand...
Click to collapse
Hey yea you're right, I didn't notice you had a new kernel out. I'm running 7-5 still but I'll upgrade once smartass is working in the new version.
I've been using "conservative" setting. Although when it initial opens a resource intensive app, its stutters a little, and then compensates by clocking up. I haven't had an issue where a game didnt want to run smooth (also at 806mhz).
I run on demand at 320/768. This to me seems to be a very good combo. The 320 min makes a noticeable difference in responsiveness without much sacrifice to battery. I've found the Max cpu setting doesn't make too much of a difference in most situations.
Sent from my cm7 Aria.
ANTUTU BENCHMARKING :- The "OPEN" Discussion (11/12/12) 10125 BEST SCORE SO FAR
First things First
I HAVE THE EXTENDED BATTERY
Secondly
TO ME
Benchmarks DO gauge the ACTUAL Performance of MY phone (when pushed)
right your still reading
HELLO
BENCHMARKING
I've had my phone since day one of the uk release, and have flashed many many many roms, and kernels, and hboot's , experimenting with different oc programs, governors, frequencies and voltages, and ui in comparison to smoothness is greatly improved the higher the antutu score
REGARDLESS OF BATTERYLIFE.
that said.
the best concoction I cooked up, with the best ingredients from many experienced chefs (so far) is
Rom:- HTC Stock ICS LATEST
Kernel:-Yoda 10.1
Tweaks:- Pedja kernel tweak app v3.3.3/ Chainfire 3D/ Pimp My Rom
oc'd @ max and min 1.836 ghz
Uv @ 1312 mv
performance CPU Governor
I/0 Scheduler SIO @ 2048kb sd cache
No vsync
my score was]
10125
Please leave detailed Descriptions of
ROM
KERNEL
PATCHES
CPU GOVERNOR
I/O SCHEDULER
SD CLASS/SIZE
OV/UV CHANGES
THANKS IN ADVANCE
mOAr voltz
argument invalid, benchmark numbers does not equal performance. research how to trick benchmarking apps.
gav-collins1983 said:
first things first, I HAVE THE EXTENDED BATTERY. secondly, TO ME, benchmarks DO gauge the performance of MY phone, ......
right your still reading, hello,
benchmarking, I've had my phone since day one of the uk release, and have flashed many many many roms, and kernels, experimenting with different oc programs, governors, frequencies and voltages, and ui in comparison to smoothness is greatly improved the higher the antutu score, regardless of battery life.
that said.
the best concoction I cooked up, with the best ingredients from many experienced chefs (so far) is
leedroids rom v 5.3.0
leedroids kernel v4.1.0
system tuner pro 1.863 Max 192 min (added 0.25mv to 1.863 frequencie because it seemed to struggle)
and leedroids ondemand forced dual core patch
my score was 7004
rant over , lol
Click to expand...
Click to collapse
Jesus Christ is this what this site has came to?
bloodrain954 said:
argument invalid, benchmark numbers does not equal performance. research how to trick benchmarking apps.
Click to expand...
Click to collapse
it's not an argument, I clearly state, that this is to "ME" on "MY" phone,
also, I'm not trying to trick my benchmarking app, what I've seen over 28 weeks is a clear indication, to "ME" on "MY" phone , that ui is smoother and faster on higher benchmark scores, simple as that
Well duh, having an overclocked CPU results in higher benchmarks because the phone can complete things faster. This translates to a smoother UI.
The problem is that benchmarks are affected by so many things that might not reflect accurately the performance of the phone. And comparing different phones is useless because of skins and different hardware performing better in tests but worse in real life.
Faster CPU = higher benchmarks and smoother UI. That doesn't mean that higher benchmarks = smoother UI.
Sent from my HTC Evo 3D using XDA Premium App
SoraX64 said:
Well duh, having an overclocked CPU results in higher benchmarks because the phone can complete things faster. This translates to a smoother UI.
The problem is that benchmarks are affected by so many things that might not reflect accurately the performance of the phone. And comparing different phones is useless because of skins and different hardware performing better in tests but worse in real life.
Faster CPU = higher benchmarks and smoother UI. That doesn't mean that higher benchmarks = smoother UI.
Sent from my HTC Evo 3D using XDA Premium App
Click to expand...
Click to collapse
it does to me from what I've seen, I've made that white clear,
what's with all the hating on evo 3d , we never had this in the desire hd forums.
gav-collins1983 said:
it does to me from what I've seen, I've made that white clear,
what's with all the hating on evo 3d , we never had this in the desire hd forums.
Click to expand...
Click to collapse
Lololol your not in Kansas anymore
Sent from my SPH-D710 using XDA App
one time, at band camp...
sent from Evo
SoraX64 said:
Well duh, having an overclocked CPU results in higher benchmarks because the phone can complete things faster. This translates to a smoother UI.
The problem is that benchmarks are affected by so many things that might not reflect accurately the performance of the phone. And comparing different phones is useless because of skins and different hardware performing better in tests but worse in real life.
Faster CPU = higher benchmarks and smoother UI. That doesn't mean that higher benchmarks = smoother UI.
Sent from my HTC Evo 3D using XDA Premium App
Click to expand...
Click to collapse
No, not, duhhhhh, overclocking also causes instability in some cases, more so if incorrect/not optimized voltages per frequency and governor in kernel, and smooth ness is affected
gav-collins1983 said:
first things first, I HAVE THE EXTENDED BATTERY. secondly, TO ME, benchmarks DO gauge the performance of MY phone, ......
right your still reading, hello,
benchmarking, I've had my phone since day one of the uk release, and have flashed many many many roms, and kernels, experimenting with different oc programs, governors, frequencies and voltages, and ui in comparison to smoothness is greatly improved the higher the antutu score, regardless of battery life.
that said.
the best concoction I cooked up, with the best ingredients from many experienced chefs (so far) is
leedroids rom v 5.3.0
leedroids kernel v4.1.0
system tuner pro 1.863 Max 192 min (added 0.25mv to 1.863 frequencie because it seemed to struggle)
and leedroids ondemand forced dual core patch
my score was 7004
rant over , lol
Click to expand...
Click to collapse
People will hate me for this, and I waited a long time to post this on this thread, but I just have to do it.
Cool Story Bro.
tgruendler said:
People will hate me for this, and I waited a long time to post this on this thread, but I just have to do it.
Cool Story Bro.
Click to expand...
Click to collapse
:thumbup:
From the Beast : Galaxy Note
gav-collins1983 said:
first things first, I HAVE THE EXTENDED BATTERY. secondly, TO ME, benchmarks DO gauge the performance of MY phone, ......
right your still reading, hello,
benchmarking, I've had my phone since day one of the uk release, and have flashed many many many roms, and kernels, experimenting with different oc programs, governors, frequencies and voltages, and ui in comparison to smoothness is greatly improved the higher the antutu score, regardless of battery life.
that said.
the best concoction I cooked up, with the best ingredients from many experienced chefs (so far) is
leedroids rom v 5.3.0
leedroids kernel v4.1.0
system tuner pro 1.863 Max 192 min (added 0.25mv to 1.863 frequencie because it seemed to struggle)
and leedroids ondemand forced dual core patch
my score was 7004
rant over , lol
Click to expand...
Click to collapse
7000 score in Antutu @ 1.86 GHz? That seems low... I right around the same score at 1.65GHz
gav-collins1983 said:
first things first, I HAVE THE EXTENDED BATTERY. secondly, TO ME, benchmarks DO gauge the performance of MY phone, ......
right your still reading, hello,
benchmarking, I've had my phone since day one of the uk release, and have flashed many many many roms, and kernels, experimenting with different oc programs, governors, frequencies and voltages, and ui in comparison to smoothness is greatly improved the higher the antutu score, regardless of battery life.
that said.
the best concoction I cooked up, with the best ingredients from many experienced chefs (so far) is
leedroids rom v 5.3.0
leedroids kernel v4.1.0
system tuner pro 1.863 Max 192 min (added 0.25mv to 1.863 frequencie because it seemed to struggle)
and leedroids ondemand forced dual core patch
my score was 7004
rant over , lol
Click to expand...
Click to collapse
*Sips his Earl Grey* You like high numbers? You clearly did not like high numbers when you tipped me for my "bedroom" services!
gav-collins1983 said:
No, not, duhhhhh, overclocking also causes instability in some cases, more so if incorrect/not optimized voltages per frequency and governor in kernel, and smooth ness is affected
Click to expand...
Click to collapse
Anybody who knows anything about overclocking knows to make sure it's stable..
Seriously, if you think your high quadrants means your phone outperforms the leaked scores of a quad core Tegra 3 phone by almost double, then you're in denial.
Let me put this in words you can understand.
Faster clock speed = faster graphics and calculations.
Faster clock speed = higher benches.
Higher benches != better performance. They're just linked.
Sure, if you're pulling over 4000 in quadrant, your device is performing quite well. But that doesn't make your Evo 3D a super phone twice as good as the rest of ours.
There are so many factors involved that simply saying "my quads are high, and that indicates performance" is ignorant and naive. Once you can explain to me how your phone "performs" better than a quad core device, then maybe I'll spare your thoughts my time.
Sent from my HTC Evo 3D using XDA Premium App
I can't wait to see where this thread goes from here. Also, my phone is completely usable, fast, smooth, and has amazing battery life and I just scored an 1100 on Antutu. Benchmarks are worthless except to raise your eEsteem. See my thread about GeekBench from a few nights ago for proof.
Sent from my PG86100 using Tapatalk
Yes, as I stated in the other thread, quadrant is simply the equivalent of measuring mobile phone penis size and nothing more.
From the Beast : Galaxy Note
7000? Pics or get out.
Twitter: @knowledge561
Blog: knowledgexswag.tumblr.com
my score was OVER 9000!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
What! Over 9000! THERE'S NO WAY THAT CAN BE RIGHT!
Hi girls and guys!
I've been searching for recommendations on optimizing the settings in the Boeffla Kernel http://forum.xda-developers.com/oneplus-2/orig-development/kernel-boeffla-kernel-05-10-2015-t3218164 through Andi's app.
And because I couldn't find any thread about it, I decided to create one.
So please feel free to post your settings and maybe tell us if you are more keen on battery life or performance, or both, and how you use your phone (gaming, browsing, maybe even making calls!? )
Thank you!
Theese are my values..
i use Interactive on both cluster with preference set to Interactive-Battery,
I/O on SIO
Boeffla Tweak enabled
so far so good, UI is smooth, battery is good, still 35% after 4 hours of screen time...
otonieru said:
i use Interactive on both cluster with preference set to Interactive-Battery,
I/O on SIO
Boeffla Tweak enabled
so far so good, UI is smooth, battery is good, still 35% after 4 hours of screen time...
Click to expand...
Click to collapse
That system tweaks option is deprecated, says so on boefflas website. Although I don't know what's @Lord Boeffla take on these tweaks in oxygenOS.
The following presets are available:
Off: No system tweaks in place, everything on stock configuration (I would recommend this)
Boeffla tweaks: this will load the Boeffla-Kernel system tweaks - smoothness and performance (nowadays I would not recommend this anymore)
Speedmod tweaks: this will load the same tweaks as used in Hardcore's Speedmod kernel
Mattiadj tweaks: Tweak set mainly for settings of the Dalvik VM created by user Mattiadj, specifically for playing heavy games
Boeffla system tweaks were recommended by me for a very long time, however Samsung did a quite good job in improving both firmwares and kernels, so nowadays I recommend to not use system tweaks anymore for daily usage.
Click to expand...
Click to collapse
I SUGGEST EVERYONE TO USE SCHEDULER NOOP AS IT IS FOR EXTREME BATTERY LIFE., also I suggest to decrease the read ahead buffer speed of SD card to 1024 as it will not drain the battery hence prolong the battery.
Press thanks if helped
Expecting 6hrs sot with these settings if all goes well. My color profile tries to mimic true rgb but it may look different on another since not all calibrations are the same per device (even with same model).
These are my settings. Quite smooth and good battery life...
I am having cpu governor to battery and the hotplug of a53 to min2max4 from the a57 min0max4.
Best kernel there is!
I go good with stock boeffle settings. Between 4h-6h SOT..
How is the benchmark scores you guys getting ? I'm on stock boeffla setting in OOS 2.1.2, mine never crossed 52K in Antutu
codeVerine said:
How is the benchmark scores you guys getting ? I'm on stock boeffla setting in OOS 2.1.2, mine never crossed 52K in Antutu
Click to expand...
Click to collapse
My result with stock settings...
Cheers
Clark
[/COLOR]
Clark789 said:
My result with stock settings...
Cheers
Clark
Click to expand...
Click to collapse
Why is this happening man My score is 52K with stock settings. Which OOS version are you using ? Is there any changes u made in boeffla config or with any other app ?
codeVerine said:
[/COLOR]
Why is this happening man My score is 52K with stock settings. Which OOS version are you using ? Is there any changes u made in boeffla config or with any other app ?
Click to expand...
Click to collapse
It depends on the mood of the OP2... I saw already results of 52K as well... The benchmark isn't very reliable.
I'm on OOS 2.1.2 too. No other app or something else.
Cheers
Clark
I feel that benchmark results are some cosmetic numbers which actually dont matter in real world. People with 52k score in antutu still get super smooth UI and smooth & lag free gaming with a good battery life (thats all everyone wants) . So thats the power of Boeffla kernel
Has anyone seen battery or performance differences with latest xposed installed on top of OOS 2.1 and Boeffla beta10?
Sent from my ONE A2003 using Tapatalk
CounterC said:
Has anyone seen battery or performance differences with latest xposed installed on top of OOS 2.1 and Boeffla beta10?
Sent from my ONE A2003 using Tapatalk
Click to expand...
Click to collapse
For me it's better than the past one. I got sobre good results with interactive in big and little, battery extreme and battery for profile, io zen... As someone here has recommended and a new thermal config as @i9100g user has made and brougth to us in the dev thread of boeffla
benchmark with Boeffla beta 10
i got 68k in AnTuTu with Boeffla Kernel beta 10, stock settings
i. imgur . com/ eLrZ2wH.png
I just today flash this kernel and atm trying these settings:
A53 and A57 interactive battery extreme
IO: noop 1024
CPU A53 max 1344 2/4
CPU A57 max 1440 1/2
GPU max 510
Touch boost 1248
System tweaks: off (dunno should i use boeffla tweaks ?)
Display:
Saturation 281
Display value 251
Contrast 260
mostly want good battery life lets see how these settings are working
tadessi said:
I just today flash this kernel and atm trying these settings:
A53 and A57 interactive battery extreme
IO: noop 1024
CPU A53 max 1344 2/4
CPU A57 max 1440 1/2
GPU max 510
Touch boost 1248
System tweaks: off (dunno should i use boeffla tweaks ?)
Display:
Saturation 281
Display value 251
Contrast 260
mostly want good battery life lets see how these settings are working
Click to expand...
Click to collapse
System Tweaks are outdate as Boeffla says on his homepage. I think you could underclck gpu until 390, touchboost 960 ad use sio as io, what do you think?
Will testing it. Thanks for your suggestion.
Gesendet von meinem ONE A2003 mit Tapatalk