[Q] Benchmark question - HTC EVO 3D

Now I know benchmarks are not that accurate but I figured I would ask about this anyway and see what everyone else thinks...
First, whenever I run linpack I only score anywhere between 12 and 16 overclocked @ 1.8...even my og Evo got about 40 so just wanted some input on that...
And second, when I run quadrant the first 12 tests go real slow but them I seem to still score from 2500-3000 but when I looked at the system info it said my phone was only a single core...kinda thought it was strange...here's the screen shot so you can see what I'm talking about
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Just a couple questions I had but not really urgent...still, any input is much appreciated
Sent from my MIUI'd 3VO (Finally switched)

I am guessing that you have the CPU overclocked too much and it is choking on the benchmarks. Try lowering your clock to 1500 and see if you get the same results.
Sent from my PG86100 using xda premium

htc phones have always fallen short on benchmarks dont worry it doesnt rlly say anything about real life performance. yeh it says one core you have 2 is it using both for the test? the world may never know

Yea looks like it isn't picking up second core for testing, hut the second core shouldn't be on all the time anyways, look for a dual core supported benchmark,
I believe cf bench is good for dual core,
And yea linpack borks when oc to far,
~Shot from my shooter in 3D~
~injected with cleanrom~

not to hijack the thread, but i didn't want to start a new post, and can't post in the developers section yet.
I just flashed myns warmtwopointthree and when running nenamark 2 i get about 20fps, I got about 26 on stock before root. I haven't messed with the cpu yet as i just flashed it, but i was wondering if anyone knows whats up with it.
I just don't want to sift through 2,500+ replies to find out. i read in the latest release notes that the gpu was overclocked but i'm getting worse performance. I wanted higher performance so i chose that rom. Anyone know why?

Unfortunately most apps are not multithreaded which means they're only going to use one core anyway. So if you want to see how your average app will perform, Quadrant is actually pretty good because it is written like most apps: for a single core.
If you want a benchmark more indicative of what the hardware is actually capable of, a multi-core test like CF Bench or SmartBench 2011 is in order. Unfortunately, I've found that the governors on most ROM's are not up to the task and they actually slow things down quite a bit so if you want max performance (numbers in the tests), you need to use SetCPU or a similar tool and choose the "performance" governor otherwise the phone will be jumping around at various clock speeds for all the tests. I've tried the "lag free", "smartass" and all the governors and none of them (on any ROM I've tested) truly know how to keep the CPU at your max frequency even when doing the demanding tasks that the benchmarks throw at it.
Mike

Related

[Q] Only one core going upto 1.512GHz

ok i've done everything apart from muffle system logging and max blocker (both fail to install for some reason =S ) , BUT in Govnah the 1st core shows the speed of going upto 1.512GHz, but the 2nd core doesn't go higher than 192MHz...why is that??
ace9988 said:
ok i've done everything apart from muffle system logging and max blocker (both fail to install for some reason =S ) , BUT in Govnah the 1st core shows the speed of going upto 1.512GHz, but the 2nd core doesn't go higher than 192MHz...why is that??
Click to expand...
Click to collapse
It only goes up if it needs it to perform some task. If you're sitting there watching it, it won't go up as it is not needed.
That is a power saving feature of dual cores.
yh but the graph that shows it hasn't gone higher than 192MHz for a while...wheras the 1st core keeps cycling at many different speeds, and i've run everything to get that second core going...but it still won't
What kernel? This has been a known issue with f15. Change the govenor to "ondemand" instead of "ctlondemand". I switched to warthog and dont have the issue anymore.
BTW....not saying its better. Just works for me
using the F4 kernel, both cores will run at the max speed you put, and when it's idling and not doing anything, will run at the lowest speed you put. I totally recmomend it. make sure you leave what ever governer is on the f4 kernel, if you don't, then the second core will stay at the min speed you put, thus not doing much.
thebadfrog said:
What kernel? This has been a known issue with f15. Change the govenor to "ondemand" instead of "ctlondemand". I switched to warthog and dont have the issue anymore.
BTW....not saying its better. Just works for me
Click to expand...
Click to collapse
i'm using uberkernel 22- on demandtcl 1512
go play a 720p YouTube video if fits smooth then everything's fine. I don't get peoples obsession with needing it to go up and down.
durps said:
using the F4 kernel, both cores will run at the max speed you put, and when it's idling and not doing anything, will run at the lowest speed you put. I totally recmomend it. make sure you leave what ever governer is on the f4 kernel, if you don't, then the second core will stay at the min speed you put, thus not doing much.
Click to expand...
Click to collapse
f4 is working really nicely over 2 weeks now. 1.8 both cores. set the min the lowest. still all day battery life.
3 words "Lithium ENYO BenchMark"
test and report your 2nd core activity via govnah... (running in background during test)
edit: "Lithium ENYO BenchMark" eh.... its ok at getting a result but not as much as some more mainstream activities and i now question its "bench marking validity". but the 2nd core is obviously active
look closely as the graph moves quicky depending on your settings
---------- Post added at 04:36 AM ---------- Previous post was at 04:21 AM ----------
i get mixed results but i always see activity from core 2
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
the 2nd is actually while playing a 720 vid
both using "uber kernel" ondemand not the ondemandctl

benchmark scores antutu

What's everyone getting on which ROM, kernel and CPU oc/uv program?
(not sure if this belongs in Q/A or not )
I just got 3799 using the set up below and CPU tuner + virtuos overclock , I got 5000 odd yesterday ?
I still ain't getting any more than 3900 , even uninstalling virtuos overclock controller for sleep n wake, and setting my pc to full power , with my brightness down and all tasks killed in taskmanager.
Try backing down the frequency you set your processor to. I get crappy scores at 1.86 but over 6000 at 1.78. Every chip is different.
What is your setup?ROM,Kernel and mods?Im running Eternity 1.4 w/ziggy's latest kernel and no other mods. Ive gotten [email protected] (reboots at 1.8ghz) and just tested now, [email protected] Try playing around with the governor, Im using setcpu with lagfree.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
1.78ghz stock de-sensed
Infected ROM eternity 1.4.1.
Ziggy dual core kernel @
1.7
Sent from my PG86100 using XDA App
This is with CleanROM 2.7 (included kernel, no tweaks) not even breaking a sweat. Governor was set to conservative and I'm running "only" 1.62 GHZ OC. I don't like to run it at max. This ROM is a great combo of speed and reliability.
Mike
justj said:
1.78ghz stock de-sensed
Click to expand...
Click to collapse
what launcher are you using ??? do you have many widgets ? many apps installed ? have your wifi enabled ? does this matter do you think ?
I'm one who wants to always get the most out of his phone, but you guys do realize that with numbers like these, it is almost impossible to find any app that is going to have lag, so why even bother to go faster? At these speeds, you're now limited by things like your (data) connection speed and SD card read/write speed. So if we make these things faster, we're never going to see a difference anyway.
Just a does of reality. I don't think many of us use our phones to do number crunching, calculating pi to a billion digits, or use them to process SETI data to search for alien transmissions. Playing Angry Birds at 1.8 Ghz with both dual cores forced online isn't going to look any different than it does stock... except maybe your left hand gets warmer while holding the phone.
Mike
gav-collins1983 said:
what launcher are you using ??? do you have many widgets ? many apps installed ? have your wifi enabled ? does this matter do you think ?
Click to expand...
Click to collapse
I have launcher Pro with many widgets and apps installed. wifi, date everything on. As I found turnning airplane mode on doesn't really matters.
mikeyxda said:
Just a does of reality. I don't think many of us use our phones to do number crunching, calculating pi to a billion digits, or use them to process SETI data to search for alien transmissions. Playing Angry Birds at 1.8 Ghz with both dual cores forced online isn't going to look any different than it does stock... except maybe your left hand gets warmer while holding the phone.
Mike
Click to expand...
Click to collapse
I don't think anyone will run their cpu at max all the time. Not with stock battery anyway. My daily setup is 1.02GHz single core UV @ 950mV. I play games like 9mm, ShadowGun etc and never had any problem with that speed.

Raise your benchmark scores

The other day I achieved the highest benchmark I've ever gotten on this beast of a phone with the help of rom toolbox. Here's how.
Go to performance in rom toolbox.
CPU slider set to 1.45 GHz (unified kernel)
Memory preset to agressive
SD boost set to 4096
Sysctl set to 4096, 50, 10, 1
Try these out and let me know if it helps. Post anything you changed to help others.
If your phone blows up or bricks (neither are likely to happen but) I'm not responsible,
Sent from my MB860 using Tapatalk
How useful is sysctl and do you know any other tips about it?
Sent from my MB860 using Tapatalk
Have no clue what it is or what it does......just know putting those numbers in there makes it faster lol......glad I could help
Sent from my A500 using Tapatalk
hatefuel19 said:
Have no clue what it is or what it does......just know putting those numbers in there makes it faster lol......glad I could help
Sent from my A500 using Tapatalk
Click to expand...
Click to collapse
You expect people to follow your advice when you self admittedly have absolutely no idea what you are doing???
Thanks but no thanks!
CaelanT said:
You expect people to follow your advice when you self admittedly have absolutely no idea what you are doing???
Thanks but no thanks!
Click to expand...
Click to collapse
I expect nothing. I know that by messing around I found a combo that raised antutu from the mid 5000 to a 6926. It also took cf-bench from 10000 range native to a 13700 native. Follow what I put or don't. I don't care. I don't know what I'm doing but I learn by trial and error. If you want to know what these values mean, Google it. If you want a quick and easy way to improve performance, give my settings a try.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Sent from my MB860 using Tapatalk
try antutu with default settings. use faux 1.45ghz kernel.
antutu usually gives bm score around 6900 for atrix with faux 1.45 ghz kernel.....
---------- Post added at 01:52 AM ---------- Previous post was at 01:46 AM ----------
you guys ever noticed the person who is at the top of the antutu atrix benchmarkk?? hes got a score of 7500 with the 1.45ghz. i always wonder how he made that score. i have tried all performance upgrades. my device usually comes with a score of 6900.
Newly flashed reloaded rom @ 1.45 gave me antutu score in the 5000 range. It was only after changing these values I could get into 6000s and blew me away when I got a 6926. I'm still tweaking the SD boost setting in rom toolbox. I could break 7000 if I could get just a little more out of it but I don't want to pay for a higher class card. I think mine is class 4.
Sent from my MB860 using Tapatalk

[Q]High quadrant scores?

So I hate benchmarks but I got bored so I ran a quadrant on my tab...I got 4221...is that a mistake?
wow. seems great to me. I was getting about 2500. I switch my governor to performance and it jumped up to over 3500 though. I've got ally of apps though so I figure not bad for me either. What's you governor set at getting over 4000?
If you see this it was sent using my GT-P7510.
Well guyz You should try a new rom to get High Scores on your quardrent tests
Well Here is a few roms that i used to get high quardrent scores
Paris rom http://forum.xda-developers.com/showthread.php?t=1284355 - Personally I love this rom because it overclockes my Tab and give me 3000++ scores
Starbust Rom http://forum.xda-developers.com/showthread.php?t=1197776 -This one also overclockes your processor to 1.4Ghz and improves I/O and other graphic values
I think This post would helpful for you If so Please Click on Thanks Button
KaFrEEkAbOom said:
Well guyz You should try a new rom to get High Scores on your quardrent tests
Well Here is a few roms that i used to get high quardrent scores
Paris rom http://forum.xda-developers.com/showthread.php?t=1284355 - Personally I love this rom because it overclockes my Tab and give me 3000++ scores
Starbust Rom http://forum.xda-developers.com/showthread.php?t=1197776 -This one also overclockes your processor to 1.4Ghz and improves I/O and other graphic values
I think This post would helpful for you If so Please Click on Thanks Button
Click to expand...
Click to collapse
ok...not to be mean...don't tell me what is helpful or not...
btw: i dont CARE about benchmark scores...AT ALL!!!!!
i just wondered why it was so high since engadget and other review sites say its like 1000.
@benzoman - just stock settings on cm9 preview (rootzwiki)
Running Task 14 overclocked to 1400 here. Couldn't get me to change ROMs for almost anything. Unless it was to an updated Task ROM. I'm running some of pershoots kernel stuff too.
Don't really care about benchmarks myself either. Just thought I'd run the quadrant to see what I got compared to what you got.
As long as it's stable and doesn't lag that's all I care.
Task has always taken VERY good care of me in all aspects of what I want, and need. So that's why I'm so partial to his stuff and suggestions of what to run with his ROMs.
Thanks mate for something to do at least for a few minutes.
EDIT: Just noticed both suggested ROMs are old Task ROMs too. LMAO
Yeah I'm running to change to an older ROM so my scores go up. NOT. :what:
If you see this it was sent using my GT-P7510.
tried many different roms and quadrant scores were different. never felt any difference on daily use of my tab though. Let's face it, we all are affected by that guy on that galaxytabhacks site
davsarus said:
tried many different roms and quadrant scores were different. never felt any difference on daily use of my tab though. Let's face it, we all are affected by that guy on that galaxytabhacks site
Click to expand...
Click to collapse
i know!!!!!
all he does...."alright lets just open up quadrant here and see what we get...2 mins later....alright thats the rom..it only got 2000..screw that..."
lol!
ezcats said:
i know!!!!!
all he does...."alright lets just open up quadrant here and see what we get...2 mins later....alright thats the rom..it only got 2000..screw that..."
lol!
Click to expand...
Click to collapse
Lol glad that im not alone
Mine on ICS for the first time its benchmark is nearly 4000 too, my expression was like this:
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
well never mind, i can return to human form at anytime so benchmark is not really a big deal, just for showing off, i thought
what is actually your best score? i have about 4000 with cm9 from pershoot and 1.7.4 aacount-kernel.
Ran this on my tab with AOKP 4.1.2 and A1 Kernell 1.8.2 running at 1.6Ghz and only got 3435
My Tab is currently running Fourth Bar Revolution ICS 4.0.4 ROM with default ROM settings and Yoda Kernel 2.4.
Quadrant @ 1.2GHz (Interactive) = 2915
Quadrant @ 1.4GHz (Interactive) = 3317
Quadrant @ 1.6GHz (Interactive) = 3558
AnTuTu @ 1.2GHz (Interactive) = 6060
AnTuTu @ 1.4GHz (Interactive) = 6793
AnTuTu @ 1.6GHz (Interactive) = 7409

Kernel Benchmarks inconsistent

Ive been trying to find the best performance/reliability from some of our kernels and have a question.
HTC stock kernel quadrant 2500
HTC stock kernel antutu 5200-5400
DSB's kernel quadrant 1700
DSB's kernel antutu 5600
SO my question is, how can stock be that much faster in quadrant? I know benchmarkas generally dont matter but thats an 800 point diffrence. DO you guys feel that stock is faster? And how come antutu doesnt reflect that big of a diffrence?
mforminio7 said:
Ive been trying to find the best performance/reliability from some of our kernels and have a question.
HTC stock kernel quadrant 2500
HTC stock kernel antutu 5200-5400
DSB's kernel quadrant 1700
DSB's kernel antutu 5600
SO my question is, how can stock be that much faster in quadrant? I know benchmarkas generally dont matter but thats an 800 point diffrence. DO you guys real stock is faster? And how come antutu doesnt reflect that big of a diffrence?
Click to expand...
Click to collapse
You'll learn very quick here that many believe quadrant to be insanely inaccurate. Only look at antutu for results.
Tho benchmarks vary, it all matters how fluid the phone feels while using it.
Sent from my ADR6425LVW using XDA App
and everyone knows that most all benchmarks are hit or miss... it all about how it feels to you
After 11 hours of use, with widgets & apps running on the latest version of quadrant I got the following on Scott's ics 3.1
Seems good to me
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Khayos said:
After 11 hours of use, with widgets & apps running on the latest version of quadrant I got the following on Scott's ics 3.1
Seems good to me
Click to expand...
Click to collapse
I'm talking about gingerbread stock vs non stock kernels not ics. I know ics scores higher in benchmarks.
I get 2550 using quadrant and its pretty much the same every time and I'm stock but I take these app with a grain of salt.
Sent from my GT-P7510 using xda premium
Lol I got 1730 with DSBs on 1.8ghz, obviously thats waayy wrong. I got higher than that with my old Desire at 1.15ghz, my old galaxy got over 3k a lot. Lol.

Categories

Resources