I'll admit this is not the must scientific of tests but the best I can do.
ROM is Virtuous 3.0.1 on Droid Incredible. Quadrant scores were higher for King kernel than custom. Setcpu turned off. Lower quadrant speed was on custom.
Edit:
Well it seems the XDA app for Android will only let me upload 1 pic. Sorry about that everyone.
FWIW the Virtuous Custom kernel Quadrant speed (which didn't upload) was 1085.
Sent from DInc running Virtuous ROM
New:
Just flashed Redemption ROM and ran Quadrant against the stock kernel installed and first run it came to 1314. I rebooted and ran it again and it came back as 1142. So I'm not seeing a real improvement against Virtuous but I have to say I do like the Redemption theme and "clean sense" of it. Nice job Lou!
Are you planning on testing the battery life, too? (Of course, that is subject to a lot of variables, too.)
Bahill: I plan to do a battery test with both of them at some point though as you rightly said there are so many variables with batteries I'm not sure what I'd be able to show. I'll let you know when/if I do it and what it looks like.
Im running Redemption Rom 2.0.4 with King #5, no setcpu at stock speed, NO OC!
Ran 3 Quadrant tests
1) 1433
2) 1463
3) 1475
and battery life is BY FAR the best out of ANY of these kernels!
Here's the Virtuous custom ROM Quadrant Score image.
DroidHam said:
Im running Redemption Rom 2.0.4 with King #5, no setcpu at stock speed, NO OC!
Ran 3 Quadrant tests
1) 1433
2) 1463
3) 1475
and battery life is BY FAR the best out of ANY of these kernels!
Click to expand...
Click to collapse
Awesome! I've been meaning to try that one out.
Here's a screenshot, like I said, this is at STOCK SPEED with NO SETCPU and battery life is awesome!
I was running King's #5 as soon as I flashed the latest Virtuous (3.0.1). I've finally narrowed down my problem of "snappiness" is that it happens with King's kernals. I'm using the exact same setcpu settings on stock, custom Virtuous, and King's, and any time I'm running king's kernals, I have nothing but problems. Scrolling in the app drawer is not smooth. And the stock and custom aren't always 100% smooth, but I'm talking to get top to bottom, I can do in 1-2 swipes with custom or stock. King's takes 3-5. And within the first minute or two is 10 times worse. I do have a profile to underclock the cpu while screen is off, but stock and custom still do a MUCH better job waking up the phone. Battery life was marginally better with King's, but performance is > than battery imho. There's times I have to wait to dial a number on King's. Not sure if I'm doing something wrong, but I've gone back to custom because it's still snappy.....
Yeah I'd like to see batt life and just overall smoothness.
bahill said:
Are you planning on testing the battery life, too? (Of course, that is subject to a lot of variables, too.)
Click to expand...
Click to collapse
...like which ROM you're running...
I've had the same problem with King's kernels. My phone seemed laggy as hell running them... even when I tried overclocking.
The custom Virtuous one is smooth as silk. From what I can tell, both are pretty good on battery life. Neither of them jump out at me as being drastically better than the other.
And with the Virtuous kernel, I don't have to reselect Swype as my input every time I reboot... with King's it doesn't start back up properly each time and I have to select stock keyboard, then back to Swype.
douger1957 said:
...like which ROM you're running...
Click to expand...
Click to collapse
Right, but the OP stated Virtuous 3.0.1, so one can assume that stays constant in his 'test'.
http://adrynalyne.us/files/kernels/battery.zip
This kernel by adrynalyne has given me the best battery life and so far it has been completely stable for me. I'm pretty sure this is an experimental kernel so YMMV.
DroidHam said:
Im running Redemption Rom 2.0.4 with King #5, no setcpu at stock speed, NO OC!
Ran 3 Quadrant tests
1) 1433
2) 1463
3) 1475
and battery life is BY FAR the best out of ANY of these kernels!
Click to expand...
Click to collapse
Unfortunately, I think quadrant isn't the best way to go about comparing the kernel's speeds. Kernels can manipulate the benchmarking tests without actually providing any more day-to-day speed:
"Cyanogen posted a troubling tweet about the flaws of Aurora Softworks’ Quadrant application. He specifically pointed to the I/O portion of the Quadrant benchmark test. The I/O test is an assessment of the efficiency of “filesystem access and database operations” on a device. In Cyanogen’s tweet, he says he was able to exploit the flaw to get obscenely high Quadrant benchmark scores by mounting “a tmpfs over quadrant’s data directory.” A tmpfs, or temporary file storage system, can artificially inflate a device’s I/O score and provide unreal benchmark improvements."
Given that the kernel was running at stock speed and came up with abnormally high levels, I would say there is some quirk in the kernel that produces the higher score. This isn't to say that the kernel isn't faster than the stock kernel or hyrda kernel, but maybe not to the extent that the quadrant scores indicate.
I too have had issues with kings #5 and Virtuous 2.7 and 3.0.1 as far as swype, and general smoothness is concerned.
I haven't noticed any significant battery savings over the adrenalyne virtuous kernel.
Related
I have Darchs latest Froyo build and it seems to run great without jit enabled.
Who's running without jit enabled and who with? Also post what rom you are using.
I'm just curious as to whether some of y'all benefit having jit enabled.
Sent from my Froyo Hero using the XDA app
I have JIT enabled on dd 2.8 but I rarely see any increase in speed. My linpack scores even run about the same (4.7+).
mbobino said:
I have JIT enabled on dd 2.8 but I rarely see any increase in speed. My linpack scores even run about the same (4.7+).
Click to expand...
Click to collapse
You can get 4.7 in linpack without JIT?
Yeah but I have 50+ apps and I stay around 2.9-3.1 mflops on Froyo without jit.
Sent from my Froyo Hero using the XDA app
seriously who gives a **** about linpack scores? no i don't use JIT, too unstable.
markus_del_marko said:
seriously who gives a **** about linpack scores? no i don't use JIT, too unstable.
Click to expand...
Click to collapse
Linpack is a common and accepted way to benchmark. Just because you don't care about it doesn't negate it's overall effectiveness at what it does.
Edit: And I was wrong on my scores. I get about 3.3 w/o JIT and still don't notice any difference.
Now that Froyo is in a RC build who runs jit enabled on it?
Sent from my Froyo Hero using the XDA app
Tried the RC1 with jit and had pretty good results, no FC's except for the common one in the market. Running the 818, jit enabled, with the uncapped kernal and never could I believe this phone would be this fast. Still the FC in the market though
So does jit drain the battery more?
Sent from my Froyo Hero using the XDA app
i have jit enabled on 8/16 build and im only getting 2.9 mflops...stock kernal...maybe i need to try the uncapped kernal for cm6??
CM6 8/18 build. 8/18 #7 Decad3nce 691mhz uncapped w/ JIT enabled - 4.7 linpack
Currently figuring out my battery life. It did drop severely from the old Fresh 2.3.3 I was running with the 2.6.34 uuV kernel. Just wiped the battery stats and applied the collin battery tweak. We'll see how it goes. Unfortunately, I did not take any note of battery life before enabling JIT - I pretty much did it immediately after dropping 2.2 on.
james33440 said:
I have Darchs latest Froyo build and it seems to run great without jit enabled.
Who's running without jit enabled and who with? Also post what rom you are using.
I'm just curious as to whether some of y'all benefit having jit enabled.
Click to expand...
Click to collapse
I'm running nfinitefx45's latest deoxed rom, urushiol's vUV2 691 kernel with Zen's kernel on top of that, and jit enabled.
Apparently, jit has no effect on my battery life, because I'm getting absolutely astounding battery life ... better than I've ever gotten before, in fact, and keep in mind that I'm the one who posted the thread a few weeks ago about getting 110 hours on a single charge!
I'm running AutoKiller with 82-20-98 settings, and I'm using SetCPU carefully dialed in to deliver the best compromise between performance and battery life. I can give you all the specifics on my SetCPU settings if you want them. For the most part, my settings have me effectively running between 558 and 576 mhz max and 19 mhz min during screen-on, and 245 max and 19 min during screen-off.
With jit enabled my best Linpack scores are in the 3.9 range, and with jit disabled I get a best Linpack of 2.4, so jit definitely makes a difference there. It also undeniably makes a difference in how snappy my machine feels.
With the way I've got SetCPU set up, the fact that my min speeds are set to 19 mhz has no negative effect at all on system performance. The phone wakes immediately from even a deep sleep, with no lockscreen lag at all, and touchscreen response when I'm using the phone is instant.
Suffice it to say, I'm extremely happy with my setup.
Cheers and best of luck in your efforts!
Ok after bricking one phone, i got a new one and loaded voodoo and dirkks SV1200 kernal. also oc'd to 1200. everything has been perfect and had scores in the 18 or 1900s. this morning i was using my phone and noticed it was very laggy. i checked quadrant scores again and it scored in the 14's. as i ran the benchmark, i noticed it hangs on the I/O 3of4 for longer then usual. i have restarted the phone, no luck. even reflashed the kernal, no luck. I cant think of any new programs or changes to the phone. any ideas?
Jeff5886 said:
Ok after bricking one phone, i got a new one and loaded voodoo and dirkks SV1200 kernal. also oc'd to 1200. everything has been perfect and had scores in the 18 or 1900s. this morning i was using my phone and noticed it was very laggy. i checked quadrant scores again and it scored in the 14's. as i ran the benchmark, i noticed it hangs on the I/O 3of4 for longer then usual. i have restarted the phone, no luck. even reflashed the kernal, no luck. I cant think of any new programs or changes to the phone. any ideas?
Click to expand...
Click to collapse
Well the 3rd I/O test does take the longest to complete. If you are having I/O issues this certainly would give you a fluctuating Quadrant score since so much is I/O affected especially on a voodoo'd SF. Have your other benchmarks taken a dip as well? Does the phone lag in regular use? If its just a Quadrant issue I would not overly panic. Cross check with other benches
My phone did that with voodoo too, I installed dirrks lv 1200, and when I upgraded it started acting up. I did remove the fix before I updated, did everything right, but it still lagged so I just glassed back to stock and did all the steps for remove lagfix. Still on stock kernel now but will install birdman optimized tonight.
Adrynalyne is the man BTW. Ok, as to the OP, have you reset the CPU in SetCPU after you restart your phone? If I don't, I get a score of 1400 ish too.
that would make sense as I believe it resets to 800 instead of 1200
thestigwins said:
that would make sense as I believe it resets to 800 instead of 1200
Click to expand...
Click to collapse
Yes, it resets to 800. 1.2 ghz is nice, I would like to see 1.6. G2 OC'ed at 1450 I believe... Ugh, one day, one day.
No luck with OC kernels
I have tried a few, everytime I get issues. Seems like setcpu just does not play well with this phone. I get 1800 on stock voodoo .2. Don't see a need for the extra conservative 200mhz that an OC can provide. Good luck with the kernels though.
Anyways, as stated, make sure set CPU is set to 1.2 GHz and not 800 MHz which is where it defaults. I use the widgit and crank up the speed any time I'm using the phone, else, I leave it at 800 when charging / idle etc.
Also, go into System Panel (best app killer around) and do an END ALL - it will kill everything... THAN run quadrant. If your scores skyrocket, don't be alarmed. Quadrant likes to have a very clean system to run on, if you have a lot of cached or running applications, Quadrant seems to suffer, I'm assuming due to the delay between it requesting more RAM from Android and the time it takes for it to release it. Seems to make sense as to why it would slow down to me at least based on what I know about how Android handles RAM.
Just make sure to reboot your phone after to restore your normal app library of running apps.
With a 1.2 GHz overclock on jt's voodoo kernel most of us are/were getting about 1925 . I've not ran dirrk's LV kernel, the undervolting is very minimal and I'd rather error on the side of stability, YMMV.
Accused: Everyone flaming & veering off-topic (You know who you are)
Charge: Violation of XDA Anti Spamming, Trolling & Flaming Rules
Verdict: GUILTY
Resolution: All violating posts have been deleted
Warning: Further transgressions might result in a BAN!
COURT'S ADJOURNED!
Hi guys,
I have been running Fresh since I first rooted my phone months and months ago. I would say, that on average, I would get high 1400s to low 1500s in quadrant with either net's or king's kernels. Lately, however, as I was finding time to update the benchmarks on my sig, I found that my quadrants are landing at about 1100s. I don't remember them being that low with net's 4.3.1 kernel and below. Anyone else having issues with his latest (4.3.2) kernels? Everything other than the low quadrants are fine; including fps2d and linpack.
Also, I tried CFS and BFS with the same results.
Any ideas?!
Thanks in advance...
Quadrant tests don't mean ****. Does your phone run OK? If it does, don't even worry about it.
Did you upgrade you Quadrant? The new one reports lower scores.
jstalford said:
Did you upgrade you Quadrant? The new one reports lower scores.
Click to expand...
Click to collapse
Indeed! That might be it. How come it reports lower now? Shouldn't they have adjusted the other phones listed as well then?!
I can't remember exactly, but I think they fixed something that was causing scores to be falsely reported as higher than they actually were. Not sure why they didn't adjust the scores in the graph, maybe they were unaffected by whatever was fixed?
Well, anyone else experiencing the same thing? Lower quadrants?!
unless you love stroking your ego, these scores mean nothing. as long as the phone is snappy and responsive for you, don't worry about "lower" scrores
Yep I get a lot lower scores on all roms and kernels I use.
Sent from my PC36100 using XDA App
Just tested on fresh 3.5.0.1 with stock kernel and got 1014.
I couldn't imagine a phone running any smoother than this and I have no need to overclock or tweak or anything. Easily get 16 hours on a charge with moderate use.
brokentoaster24 said:
unless you love stroking your ego, these scores mean nothing. as long as the phone is snappy and responsive for you, don't worry about "lower" scrores
Click to expand...
Click to collapse
I understand that it isn't accurate, but it is still the most used method of comparing between and within devices. As long as the 'innacuracy' is constant across the board, what does it matter? It also, somewhats, can give you a glimpse of whether a ROM or kernel is better or worst than others. Which is what I mostly use it for. A 300-400 drop is significant IMHO and I don't feel I'm crazy for inquiring about it.
Bielinsk said:
Just tested on fresh 3.5.0.1 with stock kernel and got 1014.
I couldn't imagine a phone running any smoother than this and I have no need to overclock or tweak or anything. Easily get 16 hours on a charge with moderate use.
Click to expand...
Click to collapse
Good to know I'm not the only one. I didn't notice any change in performance on my end too. Just thought the sudden drop was weird.
Thanks!
from my experience, there's tons of things you can do to improve quad scored but do nothing in the realm of real world gains. look up data2ext for serious I/O boosts
edit: just for fun, i loaded up the same set up. overclocked and set to performance governor and i get about 1300's
Is it just my phone that seems to run as smooth or smoother in real world use at the undervolted 1ghz setting? I haven't seen any difference in opening apps and when playing Mega Jump it stutters at 1.2ghz just as much if not more than 1.0ghz. I'm not complaining in the least as my e-peen has no relation to the speed of my phone. The undervolting is also an added bonus of running the 1ghz setting. Coupled with Juice Defender and a few battery concious settings my phone is lasting crazy long. I'm running Bonsai's stock Genocide kernel at 1ghz conservative. When I was on ACS I think 1.2ghz my have been a bit smoother.
Yes i believe its just the kernels that are out right now...
i think they recommend 1000-1200 IF you want to OC but im not sure
As Randy pointed out in one of the Bonsai threads, much more battery is saved by lower operating freqs than by undervolting... 1000-1200 will drain the battery faster than 200-1200, even with the undervolt at 1000.
Sent from my mind using telepathitalk
I swear no one ever takes the time to fully read my posts on this forum :facepalm
stir fry a lot said:
I swear no one ever takes the time to fully read my posts on this forum :facepalm
Click to expand...
Click to collapse
I have noticed just the opposite of your OP. Running at 1000 I find that some of my games are a tad choppy at times. Going up to 1200 and they are smooth.
To be completely honest, I run at 100-800MHz on the SFR rom, no hiccups or lag differences between 800MHz and 1000MHz that I can tell, so why not run it at 800MHz.
what are your OC scaling?
Ondemand or Conservative?
100-1ghz Conservative (as recommended in the Genocide thread)
100-1.2ghz On Demand when I was running ACS.
I have personally found that setcpu is my phones nemesis. I get waaaaay better battery life and crazy better performance running genocide 0.2a 100-1200 "stock" than running any combination of settings / parameters, other versons of same kernel, or even another kernel for that matter.
Seven Sisters EPIC using XDA App
Bonsai rom uses genocide kernel? I thought it was it's own custom kernel, I'm on bonsai 3.0.1 with the bonsai 3.0.0 kernel and gpu performance is much improved over stock, running @ 1.2Ghz and much improved battery life over stock eb13.
Its an optimized version of Genocide from my understanding
I have personally found that setcpu is my phones nemesis. I get waaaaay better battery life and crazy better performance running genocide 0.2a 100-1200 "stock" than running any combination of settings / parameters, other versons of same kernel, or even another kernel for that matter.
Click to expand...
Click to collapse
I set it to 1.2ghz and deleted SetCPU. My phone is flying now. Thanks for the advice.
stir fry a lot said:
Its an optimized version of Genocide from my understanding
Click to expand...
Click to collapse
Ok, I've never read that anywhere before and my first thought was.....you switched out the kernel for genocide and thus caused your own performance issues.
No bonsai is not the same as the genocide kernel at all. Its their own kernel that they made.
Sent From My Evo Killer!
Well in that case I'm running the Bonsai kernel.
I am running the Twilight Zone kernel, and have it set at 200-1000 with ondemand
My battery life has been AWESOME, and the I have noticed no lack of response in anything I have run
stir fry a lot said:
I set it to 1.2ghz and deleted SetCPU. My phone is flying now. Thanks for the advice.
Click to expand...
Click to collapse
Apologies in advance if I misinterpreted your statement (it's a bit ambiguous).
If you set it in SetCPU, then got rid of SetCPU, the change will not stick. Any settings done through SetCPU do not make permanent changes to startup scripts, but are instead applied by the SetCPU application. To make them persist, the "Set on boot" checkbox must be checked, and when this is the case, SetCPU simply does its thing quietly in the background as a service.
If, instead, you modified startup scripts to set CPU speed and governor, well, then, never mind
dwallersv said:
Apologies in advance if I misinterpreted your statement (it's a bit ambiguous).
If you set it in SetCPU, then got rid of SetCPU, the change will not stick. Any settings done through SetCPU do not make permanent changes to startup scripts, but are instead applied by the SetCPU application. To make them persist, the "Set on boot" checkbox must be checked, and when this is the case, SetCPU simply does its thing quietly in the background as a service.
If, instead, you modified startup scripts to set CPU speed and governor, well, then, never mind
Click to expand...
Click to collapse
No, I got you. I've heard of people changing scripts in lieu of SetCPU. I just had read somewhere by one of the devs that if you set it to the 1ghz setting and remove SetCPU that the undervolting would stick. I though it would be the same for clocking. Either way my phone seems t be running very smooth without it so I'm ok with that.
Hello, first of all this is my first post and thread. I have read the rules and searched for a similar thread. I am not stupid and I know how forums work. Now my question is: Which is the most performant kernel for the MT4G?
Know to be more specific, I am looking for a kernel which a very high OC and UV rate. I want something that can easily manage clock speed when it needs it the most. I am interested in genereal boosts of speed, higher I/O managing speeds, better gaming experiences and a longer battery life. I know that the kernel doesnt do all of that but it has a major impact those aspects. I am running CM7 nightly 144. Im hesitating between Faux123 Cm7/AOSP 2.6.35.13 kernel (http://forum.xda-developers.com/showthread.php?t=1113697) and thederekjay 2.6.35.13::demonSPEED INSANITY kernel (http://forum.xda-developers.com/showthread.php?t=1113554). I appreciate anykind of constructive feed back
I personally use demonspeed-insanity. It'll clock up to 1.8 atm. Couple that with the SMARTASS governor and you get a pretty decent balance of battery life to performance.
can't go wrong with either kernel, i've used both and their both outstanding in performance and battery life.
Sent from my HTC Glacier using XDA Premium App
Thanks for the feedback guys . I'll probably try both of them. I'll post some quadrant benchmarks and battery life results.
Faux 123 1.0.7 is the best IMO, great uv and quadrant scores, got 2600+ on stock and battery life is great.
I have used both kernels and got great battery life from each one. I prefer Faux's kernel for AOSP roms, and thederekjay's kernel for MIUI roms. Faux updated his kernel to OC @ 2.03ghz and its blazing fast
Faux's all day!!!!
He has put in the most work and has the best performance by a wide margin.