Why does the new linpack update run slower.... when i click run benchmark it takes like2 and a half seconds...and benchmarks slower like 32-35 instead of almost 40. Wtf
Instead of doing a single run, it now does multiple and averages them giving a more accurate score.
Related
Has anyone noticed deteriorated display performance after DI08? I see this most-notably when switching applications, opening the notification drawer, and sliding between TW home screens. It didn't do this nearly as badly, at least according to my perception, before DI08. I just opened Quadrant, and the highest I've ever seen "Current freq" is 800MHz - I can't seem to get it to stick at 1GHz no matter how much I load the CPU. That being said, Quadrant and Linpack scores are high as ever...
I also noticed that it drops down to 400MHz very quickly. So, is it possible that the display sluggishness is linked to a more aggressive clock throttling to save battery? Or am I just seeing things?
Does anyone know if the desire hd has any sort of processor throttling enabled on it? Am running Linpack benchmarks on my phone and am getting some very weird results.
My linpack benchmark changes between ~35Mflops and ~62Mflops seemingly at random, this is a fresh install of cyanogen 7.0 with LorDMod V4.0 kernel. Also have the google apps installed (are any of these known to cause high cpu-usage?) I have checked the background CPU usage using SystemPanel and there appears to be no other processes using the CPU.
Phone is running at 1.996ghz at 1300mv, have run mutiple stress tests over several hours and its rock solid in terms of stability, the max temperature i've seen is 37 degrees which also seems perfectly fine (I've heard from other threads that the processor temperature is normally around 5 degrees higher than the battery, so this doesn't seem to be an issue)
Have also tried increasing the voltage at 1.996ghz all the way up to 1475mv (incase of strange stability issues) however, as expected this has no impact on the linpack readings.
Basically i'm at a complete loss to explain this myself and therefore am wondering if any of you guys can come up with any other ideas?
Thanks.
What CPU govenor are you running the benchmarks with? You should use Performance, as it will lock your CPU at the maximum allowed frequency
I hope your not running your processor at that speed all day
Sent from my Desire HD using XDA App
Wait wait wait...You are running 2GHz @1300mV?What the F*CK?Mine needs 1450mV to run @1.8GHz,maximum is a really unstable 1.9GHz with 1500mV,and this only for some seconds!
Anyway...You could try killing some apps.It's normal that tests aren't consistent however.
This thread is Meant to display my results from "battery" testing "performance" benchmarks etc. to show its fullest potential.
I've ran "quadrent, linepack and some other benchmark" results vary.
QUADRENT I usually get depending on how many widgets, background task are open about 1900-2300 and Used "SET CPU" To performance mode "926 MHZ - 1000 MHZ
Eagerly waiting on a OC For the tegra 2 chipset on the iconia "xoom" has already achieved great speeds of 1.5 (1500) GHZ
I made a short brief 6 minute video of me running a few benchmarks, to show the speeds it achieves. "video proccessing will be watchable in about 5 minutes"
http://www.youtube.com/watch?v=6CN2bZisCvg
Next test i will be running on this device is battery life, something lots of people are converned about should have results in qbout 48 Hours, will keep update will run many test on this device, from straight playing games battery life to just movies web browsing etc. the whole 9 yard.
Will continuasly post updates on new things i found out from overclocking testing everything you can imagine.
I also ask that we all as a community can share our expieriences with the device from batterly life to performance to enchancements.
"l3giticonia out"
Wow I wish you had kept this thred up I would have loved to see this before I purchased.
hmmmm
Correct me if I am wrong, please: but by using setCPU to set a "performance" option and setting the speed lower then stock clock speed (926 mhz?) you have now disabled the systems ability to lower the clock speed and are underclocking the system.
Am I mistaken in assuming that you are going to get rather useless data on the overall performance of the system by setting these options? By underclocking, you have installed a custom kernel on the device and any results you get are going to be subject to the performance of that kernel, which is not what the a500 stock is running.
Why 926 mhz? Stock is 1.0 ghz, I generally run at 1.4ghz+. Am I missing something? Again, by setting SetCPU to "performance" any data you come up with is going to be on a system that cannot lower clock speed during low use periods, so your battery life stats will be lower then a normal users experience.
ARM based chips are built to scale the speed depending on load. By setting performance mode, you disable the chipsets ability to scale, right?
Complete benchmark completed.
How did I do this test? I runned each benchmark 3 time's and calculated the average score. (test1+test2+test3÷3= End result)
Energy saving was not enabled.
Wi-fi was enabled
Datatrafik was enabled (3G)
Auto adjust screen tone was not enabled.
Monjori: 33FPS
Electopia: 43FPS
nenamark2: 58FPS
M3D: Does not start up. (not compitable?)
Basemark ES 2.0 Taiji: 42fps
BenchmarkPi: 327
Linpack: 65MFLOPS
CF-Bench: 15180
Vellamo: HTML5: 1814 + Metal: 619 = 2433
Sunspider (stock webbrowser): 1017 +/- 0.5%
Rightware BrowserMark: 174584
AndroBench storage benchmar: Sequential read speed: 37.85 MB/s, Sequential write speed 9.74 MB/s.
Phew... some test's take a hell load of time to do.
Sent from my GT-N7100
Nice!! Btw, I think you gotta try the the Linpack in multi-core test, too. It should show score of more than 150+ MFLOPS, may reach even 250!!!
CrimzonEyed said:
Complete benchmark completed.
How did I do this test? I runned each benchmark 3 time's and calculated the average score. (test1+test2+test3÷3= End result)
Sent from my GT-N7100
Click to expand...
Click to collapse
That's not something I would do.
say (3+3+3/3)=7 and that's not average.
it should be ([test1+test2+test3]÷3= End result)
You mean do the "+" math and "=" before using "÷"?
Because thats how I did.
Sent from my GT-N7100
(((("The devices range seems to be huge. Not being interested into scores that much I had not tried Antutu on my S4 to date but did a few minutes ago:
On my rooted stock Vodafone Rom the Antutu score with Faux 3m stock settings and stock CPU clocks is 24.413.
Lowering the CPU clocks to between 324-1674 and changing to mpdecision off and intellidemand it is 24.221...
No problems with Antutu - both scores fine...
Edit: taking a closer look it seems to me as if Antutu overwrites the upper CPU clock limit somehow. The 1674 before running Antutu popped back to 1892 when checked after Antutu had finished. The lower limit remains untouched though - strange... At least the resulting scores are consistent - 2nd run with the second settings setup ended with equal 24.370.)))
How are you able to do the test???
As soon as I start Antutu, It bounces out from antutu and no mather if I lower the frequency ,still same result.
i think it is an antutu problem my galaxy s3 was getting lower scores while i was overclocking :/ maybe when the app gets updated it will be fixed
ilcabrone said:
(((("The devices range seems to be huge. Not being interested into scores that much I had not tried Antutu on my S4 to date but did a few minutes ago:
On my rooted stock Vodafone Rom the Antutu score with Faux 3m stock settings and stock CPU clocks is 24.413.
Lowering the CPU clocks to between 324-1674 and changing to mpdecision off and intellidemand it is 24.221...
No problems with Antutu - both scores fine...
Edit: taking a closer look it seems to me as if Antutu overwrites the upper CPU clock limit somehow. The 1674 before running Antutu popped back to 1892 when checked after Antutu had finished. The lower limit remains untouched though - strange... At least the resulting scores are consistent - 2nd run with the second settings setup ended with equal 24.370.)))
How are you able to do the test???
As soon as I start Antutu, It bounces out from antutu and no mather if I lower the frequency ,still same result.
Click to expand...
Click to collapse
Yea it's a problem with AnTuTu, after you play with your phone's frequencies, it force closes. Sometimes with Quadrant too. It's weird because benchmarking apps are almost ment for testing your phone with overclocking, you (we) should contact the dev!
Thanks for you reply.
Yes,Antutu seem to be the culprit.
Funny though some are able to do the Benchmarking without a problem.
I do enjoy the smoothness and stability I find in FAUX Kernel for my Galaxy s4 9505,dont really need a benchmark to tell me if its good or not.
Anyway would be fun for comparison.
Regards,