Related
Sent from my PC36100 using XDA App
http://forum.xda-developers.com/forumdisplay.php?f=652
I've had my phone overclocked 15% for months now...I pre-ordered and picked it up on June 4th (the day they were released to the public) and it's never been replaced. I rooted it right away using the engineering hboot method, and I immediately went to a rooted stock rom with a kernel that'd allow overclocking the very first time I saw one posted on this forum. I've never been able to run it at anything higher than 1152 mhz, but I've also never had any stability issues at 1152 mhz. I also allow it to go as low as 128 mhz all the time.
Since then I've played with various roms, and various kernels, but the first thing I've done when changing any rom or kernel, is always set the min and max CPU speeds again, and decide which governor to use. I've also always played with the undervolting strategies, from static to HAVS, and I've always been able to get away with the most agressive stuff posted without any stability issues.
Your mileage may vary, but thats been my experience with an overclocked Evo. I will admit I can barely notice the performance difference from 998 mhz to 1152 mhz, but I actually notice a battery life improvement...get it done faster so the CPU can go back to idling at a low frequency as soon as possible mentality I guess.
please watch what you say here. its not going to get you any help to curse and swear at other members. last warning
@MikeOD, which governor and what governor parameters have you found to work best for you?
I think the whole overclock boils down to what you do with your phone. If nothing overly cpu intensive, then there's likely to be little gain in the amount of saved time.
I actually have mine under clocked at 921 Mhz (came that way in the rom initially). UI was fluid enough and everything still seemed to work well/responsive. I get slightly better battery life too. Noticeable in the rate the batt % declines during active tasks (web browsing).
As the title says whats the max o/c can you get out of your defy?
Im on standard 2.1 and can get 1300mhz @ 80vesl and get 1388 in quadrant.
So if your a o/c nutter and try an get the max out of everything tell us what ya can get!
Sent from my Defy using XDA App
I'm getting 1348 at the stock 800 mhz :/ Got about the same when overclocked. So, meh.
1350/80 stable, 1409 in quadrant. (stock rom)
Sent from my MB525 using XDA App
Pays ROM v5
Quadrant: 1405
setVsel: [email protected] (haven't tried higher than 1100Mhz yet)
I got 14xx (no trick) with 1000 @ 50 and 16xx (no trick) with 1200 @ 62 on finland 2.2.
For how long and how hot did your Defy get?
it ran for about an hour at 1350, and the max temprature was 39°C as i recall. based on abc's results, that jit in 2.2 is a big performance hit, looking forward to the official update
Sent from my MB525 using XDA App
Just ran 1400mhz @ 95vsel
1508 on quadrant
Ran stability test for 2hrs30mins @28°c (had it sitting on an icepack haha)
Stock 2.1 and its very snappy.
Sent from my thumbs
just run [email protected]
quadrant 1709 1st score, 1671 2nd score.
2.2.1 froyo, 3.4.2-155
i got 1589 with 60/1200
Can running such high frequencies and vsels regularly damage the CPU? I'm kind of scared of toasting it, so i stopped at 1100/60 (scaled with setvsel)
iamnik said:
Can running such high frequencies and vsels regularly damage the CPU? I'm kind of scared of toasting it, so i stopped at 1100/60 (scaled with setvsel)
Click to expand...
Click to collapse
i dont think so... did the stability test for around 15 minutes and the temp stays at 38 C
60/1200 with threshold at 80%, everything is so smooth
samoelito said:
i dont think so... did the stability test for around 15 minutes and the temp stays at 38 C
60/1200 with threshold at 80%, everything is so smooth
Click to expand...
Click to collapse
Hmm, i know have it at 62/1200, with threshold at 60%. 80% makes it feel a little laggy.
Hopefully the processor will last the life of the phone at these settings!
What's a safe level to overclock my phone to? I'm running the latest MIUI (1.9.16) with the lastest Tiamat kernel (1.1.3) I'm at about 1344 MHz right now. Can I go much highter?
Sent from my rooted, MIUI DInc2
From my testing, I've found there isn't to much to be gained going over 1.6ghz.
For a safe level I keep mine at 1.2 ghz. This has worked out well for me. Try different frequencies and see what works out for your needs.
I'm running JDKernel with the max set at 1.5 ghz, using the smartass governor. When I look at the "time in state" info, it usually shows about 75% of my time spent at 768 mhz. Seems to be working great.
Sent from my Incredible 2 using XDA App
Well, I tried bumping it up to 1.6, and I locked my phone and the screen turned off, but the keys stayed on, and it wouldn't come back up. So I pulled the battery and went back to 1.3 like it was before. Now its working great. I'm on smartass v2
Sent from my rooted, MIUI DInc2
If your having lockups, try installing Incredicontrol from the market and bump up the voltage on the higher frequencies 25-50mv, this should keep you from locking up. I do this when I use the performance governor and want to run Quadrant or Linpack .
PacerguyDon said:
If your having lockups, try installing Incredicontrol from the market and bump up the voltage on the higher frequencies 25-50mv, this should keep you from locking up. I do this when I use the performance governor and want to run Quadrant or Linpack .
Click to expand...
Click to collapse
I've been using SetCPU, cuz I couldn't understand Incredicontrol..... What's ideal if I'm running around 1.6 for voltage?
On the Jd Kernal, I'll bump it up to 1450 and go from there.
PacerguyDon said:
On the Jd Kernal, I'll bump it up to 1450 and go from there.
Click to expand...
Click to collapse
So, you don't know about for Tiamat? I'm using that one.
Plus, would anyone recommend using his Incredikernel as well?
I'll go about the same on the Tiamat. I usually will start at 50 mv over whatever the stock setting is on the kernel when running performance runs.
I didn't know that Chad had done a Kernel for Aosp romed Dinc2's. I know his Dinc1 Kernels were great.
I accidently had my phone at 1.9 for a day running jdkernel, no crashes or anything
Atm when my phone is charging, it gets bumped up to 1.4 and i have no issues at that speed.
I wanted to know some more about undervolting and was hoping someone could help and explain a bit. Basically, what I'm seeing is when I undervolt my device on the max side of things, eventually the device freezes and needs a battery pull or reboots. When I don't touch the max voltage values and only drop the min values the phone seems more stable and doesn't reboot. Can someone explain if my thinking of only undervolting the min values is correct and why this may be happening. Also, what would be the proper way to undervolt the phone? I have been using system tuner to do so on deck's 1.3 ROM using the Lee +sbc kernel. Thanks for taking the time to read this and for all the helpful responses.
Sent from my Decked Evo running Lee+sbc... It's easier this way.
I have this same question! I was trying to figure this out becuase the freedom kernels have something to do with under and over voltage to allow better OC? not sure on why the you would need and over voltage to over clock?...
What is your ROM/Kernel setup?
Remember that undervolting and underclocking are two different things. Over/underclocking, via apps like setcpu, raise/lower the frequency at which your processor operates. Over/undervolting raises/lowers the power voltage available to your processor at any given frequency. This can potentially cause problems if you have a kernel set to undervolt your processor while running an app like setcpu to overclock it. This may be why you are seeing issues with your phone.
pmacevad said:
[snip]
I have been using system tuner to do so on deck's 1.3 ROM using the Lee +sbc kernel.
[snip]
Sent from my Decked Evo running Lee+sbc... It's easier this way.
Click to expand...
Click to collapse
Concordium said:
What is your ROM/Kernel setup?
Click to expand...
Click to collapse
OP clearly states he's running Deck's 1.3 and Tiamat 3.3.7 (aka LEE kernel)
OP should also state Deck's ROM version but I'm assuming 1.3d as it's most popular next to 1.3 Stable.
I'm running the same setup - Deck's 1.3d with LEE SBC. I'd like someone to explain (as the OP also asked) the difference between the MIN and MAX sides of the undervolting equation please.
(I do understand underclocking AND that it is a separate and different animal and for the record, my EVO is UC'd to 128/499MHz and using Smartass governor)
I'm just not sure what I'm looking at in the ST interface (what's the first number and what's the number in [] brackets?) and also not sure just why there is both a min and max voltage for each frequency and how it affects performance.
Thanks much for any inputs.
J.
I use vipermod and terminal emulator for aosp roms its much more stable than using system tuner and you can drop all voltages with a single command it all depends on your particular hardware but typically around -100 the phone will lag or freeze i usually stick to -50 at stock voltages and -25 on overclock i never overclock over 1075mhz as anything higher and my phone tries to cook my fingers however now im oc to 1.6ghz on my et4g with a 200mhz min and it stays nice and cool like it was clocked at 1.2ghz stock the exynos is capable of clock speeds up to 2 ghz but the phone doesnt handle those speeds well similar to the evo not liking anything over 1.19 at 1.2 on the evo theres a good chance of frying the logic board or bursting the battery from heat if you overclock dont undervolt too much as it makes the processor get really really hot due to it having to work harder from lack of energy to power it
We are legion, for we are many
Sent from my Anonymous DeathStar in the depths of GalaXy S2
To answer the min/max at each freq min is the lowest voltage that freq will run while idle max is the highest voltage the freq will run while in use the voltages are depenedant on the load running at the particular freq ie how many process' are running the more process' the higher the voltage required to power the processor
We are legion, for we are many
Sent from my Anonymous DeathStar in the depths of GalaXy S2
Is there an app the will show actual voltage the CPU is using in real time?
Sent from my Synergized Aggressive Lionfish Evo using the XDA app
bmw pro will u can set the mv on the widget or set recording in the app and it will log the info
We are legion, for we are many
Sent from my Anonymous DeathStar in the depths of GalaXy S2
I had a similar issue with undervolting and underclocking. I was setting my CPU min/max to 128-499 MHz and undervolting to -75 or -100 using Lee with Smartass v1. But I was getting too many random reboots and freezes on each ASOP rom I tried (e.g. Deck 1.3d, CM7).
I've been playing around with these settings. I changed my CPU to 128-576, no undervolting, and SZ 2.2.1 with SavagedZen governor. I haven't had any random reboots and my batter life is about the same...plus my phone is MUCH snappier. I'm still getting a little screen on lag but not nearly as much as I was at 128-499 and undervolted.
As a note, I normally run Deck 1.3d as my daily driver but for the last couple days I've been running KWIQ. It's definitely giving 1.3d a run for its money as becoming my daily driver.
Anywho, are there any rules of thumb with undervolting and underclocking? Like is 128 MHz is too low with -100 undervolting? Or are these things pretty much phone dependent? I know each phone seems to respond different to these things.
-EViL-KoNCEPTz- said:
To answer the min/max at each freq min is the lowest voltage that freq will run while idle max is the highest voltage the freq will run while in use the voltages are depenedant on the load running at the particular freq ie how many process' are running the more process' the higher the voltage required to power the processor
Click to expand...
Click to collapse
Thank you. That made good sense to me. Question then: is it suggested/typical to drop min and max by equal amounts or is it typical to drop min more (so there's juice for the proc under load but maybe being more aggressive with the sleeping/idle side of things or can the proc run at a given frequency at different voltages)?
jackfrost909 said:
I'm just not sure what I'm looking at in the ST interface (what's the first number and what's the number in [] brackets?)
Click to expand...
Click to collapse
Are you familiar enough with ST to shed some light on this? Please if so. Thank you.
detcup4evr said:
Is there an app the will show actual voltage the CPU is using in real time?
Click to expand...
Click to collapse
-EViL-KoNCEPTz- said:
bmw pro will u can set the mv on the widget or set recording in the app and it will log the info
Click to expand...
Click to collapse
Though I have to recheck this I don't remember BMW showing the voltage being used at each frequency just the overall voltage draw. If you know otherwise or another tool that does, please share
Thanks for all the help.
J.
--
Tapa Tapa Tapa..
Let's see.. My phone flies, my battery doesn't suck.. and your point was.... ?
OG EVO - Deck'd and LEE'd because it's just better this way.
Lionfish kernel
Has anyone tried undervolting with lionfish kernel on SENSE Roms, is it stable?
einstein562 said:
Has anyone tried undervolting with lionfish kernel on SENSE Roms, is it stable?
Click to expand...
Click to collapse
I use aggressive and it works well but some phones don't like it, hopefully I can get vipermod to work, and that will allow you to adjust each frequency independently
Sent from my Synergized Aggressive Lionfish Evo using the XDA app
I actually just got rid of lionfish aggressive on MikG3.11
I had a lot of lagginess, but battery life was really good... Back to anthrax for me, lionfish was based at one point on freedom kernels I thought, maybe one of those would be better...?
What's the best thing to use to control undervolting on sense roms?
Sent from my PC36100 using Tapatalk 2
Is the battery life improvement substantial when undervolting?
I have governators set at 240-650max and am getting decent life on my ics savoca kernel.
What % improvements are obtained?
Well I don't have logcats and scientific evidence, but I saw about 12 hour days, with moderate use, everything syncing, lots of text a few calls, hotspot for an hour or so, some web browsing, but the random revolts and lagginess on that was horrible, freedom was better.... It's worth it to underclock
I have everything set to 384-1152.
Different governors for profiles, screen off @ conservative because I have a lot to sync
Sent from my PC36100 using Tapatalk 2
What is your highest score in antutu using custom rom?
21,871 Mahdi rom 09-02-2014
@soulzero, care to share your settings with that benchmark score?like cpu settings etc...
i only got 18367 with stock cpu settings on vanir rom.
Sent from my Optimus G using XDA Developers Android App on Vanir 4.4.2.020614
Newest Mahdi Rom + dyn.Kernel
I'm usually getting between 24k and 25k, depending on the rom. Just tried now for the sake of this thread, with CM11, stock kernel, Interactive governor :
Mercvtio said:
I'm usually getting between 24k and 25k, depending on the rom. Just tried now for the sake of this thread, with CM11, stock kernel, Interactive governor :
Click to expand...
Click to collapse
what is interactive governor?
I always use interactive /row.
Edit to add, using cm11 nightly and Demetris V1 kernel.
Sent from my LG-V500 using XDA Premium HD app
My best is 26777. I'm on latest Mahdi, using Demetris kernel and the optimized dalvik. I actually tested several times on Mahdi, then flashed the optimized dalvik and tested. Got a good increase. Then flashed Demetris' kernal and tested. Again an increase. Averaged 23847 on Mahdi, 25647 with optimized dalvik, and the 26766 after the kernal.
knightsh said:
My best is 26777. I'm on latest Mahdi, using Demetris kernel and the optimized dalvik. I actually tested several times on Mahdi, then flashed the optimized dalvik and tested. Got a good increase. Then flashed Demetris' kernal and tested. Again an increase. Averaged 23847 on Mahdi, 25647 with optimized dalvik, and the 26766 after the kernal.
Click to expand...
Click to collapse
I usually use the optimised files as they have been the highest scoring for me, though I check with quadrant.
That score I posted above was running Art as I changed over last night to see how it runs and ran antututututu out of curiosity.
When I get a chance I'll switch to optimised dalvik and bionic and do a few runs.
Sent from my LG-V500 using XDA Premium HD app
24761 run once, on the V510 out of the box using the sleekai kernel. Nothing else is optimized on my device. I really don't know if the stock kernel would be any different. Maybe. They still ramp up to maximum cycles during the test. Stock default is Performance and CFQ, whereas sleekai default is On-demand and deadline. I do think the deadline is better at handling scheduler requests. I get a smoother scroll action I attribute to the deadline scheduler. Also a way better battery life.
I don't really think benchmarking is a good way of judging the performance of the device within a 5000 point range, but thats just my opinion, and may be wrong. If your cpu heats up too much during the test, your device will slow it down.
Edit* Newest Antutu Benchmark on the 510 out of the box using the sleekai2 kernel is 27052
As a side note, The gamma problem has been resolved entirely using this kernel. No more need for Trickster.
Haha, using dyn kernel r3and mahdi rom
Sent from my LG-V500 using XDA Premium HD app
whoamigriffiths said:
Haha, using dyn kernel r3and mahdi rom
Sent from my LG-V500 using XDA Premium HD app
Click to expand...
Click to collapse
oooh!!!! Now you have to find out whether it is the Rom or the kernel!!! Better switch back to stock and run it again lol. I'm betting his kernel is pretty good!
whoamigriffiths said:
Haha, using dyn kernel r3and mahdi rom
Sent from my LG-V500 using XDA Premium HD app
Click to expand...
Click to collapse
How? I'm using the same but don't seem to be able to get over 24 000. Usually at only 22000-23200 :/
general_z said:
How? I'm using the same but don't seem to be able to get over 24 000. Usually at only 22000-23200 :/
Click to expand...
Click to collapse
As I said in another thread , the secret sauce is this http://forum.xda-developers.com/nexus-4/general/4-4-4-4-x-qualcomm-optimized-dalvik-t2546120
I'm guessing the v510 doesn't need these libraries for its high score but we do.
It's all moot for me. I'm currently using ART, which scores in between the two when benching, but in reality I don't notice any difference between the three approaches in daily use.
Sent from my LG-V500 using XDA Premium HD app
whoamigriffiths said:
This is how I tested.
I installed Antutututu and quadrant.
I ran both several times to get base results.
Then I flashed the dalvik file and wiped dalvik and cache and ran both benching tests several more times and noted those results.
Then I reflashed my backup and right after flashed the Bionic file only.
I wiped cache and dalvik again, booted up and re-ran benches.
After noting those results I flashed the dalvik, so I now had both flashed. I wiped dalvik and cache again and ran more benches.
I've done this on cm11 and mahdi.
All the best rich
Just an addendum,
Whilst I have ran these tests, it has been for a bit of fun,
As I said in the Antutututututututututu thread, I'm currently running Mahdi and ART.
It scores around 4k less in antutu, but in practice I see no change in behaviour In my gpad.
I guess the gpad has enough power that the extra oomph that these files are supposed to deliver, is just a small ripple in real world use. Even if it's 4k in antutu.
Click to expand...
Click to collapse
Yeah, tried ART as well but didn't notice any sort of improvement. What i'm mostly after is solving the "thing" that makes UI animations a bit less then 60 FPS. It's more like its 30-60 FPS and one can definitely notice a drop here and there. My old Nexus 7 (2012) is more consistent in that regard and it's UI (stock, unrooted) actually responds faster to touch events which is a bit odd due to its old Tegra 3 processor and non-modded/rooted OS.
---------- Post added at 09:44 PM ---------- Previous post was at 08:53 PM ----------
Ran a bunch of benchmarks. All in all it's way too inconclusive since the difference in average pretty much is in the margin for error (the highest average only differ around 1000 points from the lowest average).
I don't seem to be able to go beyond an average of 24.348 so it's odd to see some of you post scores that is almost 3000 points higher then that.
What could i have missed?
Code:
[SIZE="3"]Antutu X Benchmarks[/SIZE]
LG Gpad v500
- Paus 1-2 minutes between each run to let the pad cool slightly
- Airplane mode
- 50% brightness
- All apps killed + wait 3 minutes for the ones "needed" to start
[B]In test[/B]
- Mahdi 2.5 rom
- dyn.Kernel v3
- Dalvik /Bionic mod
[B]Mahdi 2.5 (02/17)[/B]
Ondemand governor
Bench 1: 23.194
Bench 2: 23.237
Bench 3: 23.863
Bench 4: 22.471
Average: 23.191 (92.765/4)
[B]Mahdi 2.5 (02/17) + Dalvik /Bionic mod[/B]
Ondemand governor
Bench 1: 23.459
Bench 2: 23.522
Bench 3: 25.357
Bench 4: 25.055
Average: 24.348 (97.393/4)
[B]Mahdi 2.5 (02/17) + dyn.Kernel v3 + Dalvik /Bionic mod[/B]
Performance governor (no ondemand available)
Bench 1: 22.996
Bench 2: 23.346
Bench 3: 25.499
Bench 4: 24.379
Average: 24.055 (96.220/4)
I'll try Antutu when I get home but I noticed when I switched my i/o scheduler to deadline my score was almost 1500 points higher with 3DMarks Icestorm Unlimited benchmark.
EDIT: so I got 21448
I was hoping to beat at leas the HTC One and S4
My CPU and RAM are "general"
How did some of you get "good"? ART?
general_z said:
I don't seem to be able to go beyond an average of 24.348 so it's odd to see some of you post scores that is almost 3000 points higher then that.
What could i have missed?
Click to expand...
Click to collapse
I would say your governor might be the cause : you are using OnDemand, have you tried swithching to Interactive ? In my experience, it would solve most of my stuttering launcher issues.
If you want better scores, you might wanna try Performance, but I wouldn't recommend it for everyday use.
I wrote a long reply last night and lost it so today I will do it again but a bit at a time
Ok as already mentioned I also prefer to use interactive and if the kernel allows then it's always interactive /row.
If row is not allowed by the kernel then I use deadline.
I would like to suggest that it's probably not just one thing that needs to be done, but a number of small tweaks that accumulate to achieve a better bench score.
Now some of these tweaks are only for benching whilst others will give you a zippier feeling tablet.
This would be a typical setup for me to bench and I'll point out which I used day to day to speed up my old N7.
First flash the dalvik and Bionic files (obviously not tried on the tegra N7)
After booting up go to developer settings and uncheck the three animation settings, then scroll down and check force gpu rendering. (this mod is good for daily use for a bit more zip, I used this on my N7 2012.
Next go into whatever app you use to adjust cpu.
Adjust the min and the max cpu frequency to 1.728 ghz (you may get reboots with the minimum set at this, so just dial it back slightly or increase the voltage value for 1.728 ghz)*this is most definitely not for day to day use *
Next go to the gpu governer and change the setting from ondemand to performance (I believe this has quite an impact and I reckon that it's a good one to try for day to day use)
It's worth spending a bit of time having a play with sd read ahead value.
I set mine to 2048, more from habit than any testing on the gpad. It is definitely the way to go on my phone but ymmv. This can be used all the time if you see a worthwhile difference.
To a degree I too stop some apps before benching but in the past I noticed if not careful this can be detrimental to antutu.
I use clean master for this as you can go into settings and modify the ignore list.
I think that is just about all that I do. I will update this if I think of something I missed.
All the best rich
Sent from my LG-V500 using XDA Premium HD app
Mercvtio said:
I would say your governor might be the cause : you are using OnDemand, have you tried swithching to Interactive ? In my experience, it would solve most of my stuttering launcher issues.
If you want better scores, you might wanna try Performance, but I wouldn't recommend it for everyday use.
Click to expand...
Click to collapse
Yes, with interactive i didn't notice better performance. In previous tests interactive actually always scored lower.
Sure - the performance governor can get better scores, but i'm not just after scores and ondemand or interactive should manage to score almost equal to performance when benching (within a 1000 point error of margin at least).