I install app PerfMon to check how much cores its uses when playing games.
I am noob and I dont know what is the reason of that, could anyone explain it?
So this is my result:
I was playing Wingsuit
{
"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"
}
This is in Antutu:
All cores uses 100% but they are rated at 1300 Mhz constantly
Again only one core:
In OpenGL ES 2.0 again only one core:
I will post more screenshots, but you can add it too in comments.
I hope this will see any of developers. :fingers-crossed:
on default kernel, when phone is using all four cores, they don't run over 1.3 ghz (probably due to temp limitation). so it's not just you. as for why it runs only one core, i can't really answer that; all i know is that i have same issue on 20a & CM, and i don't have that issue on 20b HK. hope this helps.
i have CM 20/04 nightly. i'll install the apps, and test what are you telling us
---------- Post added at 03:56 PM ---------- Previous post was at 03:43 PM ----------
yes... in CM we have the same issue, but in antutu, i was playing CRS racing, and all cores are actives most of the time, but, when i have the cpu limited to 1000Mhz, maybe is something related with cpu temperature when al cores are running, when i set cpu to 1500Mhz, when all cores are actives it never runs over 1300Mhz
Same to me..only one core (sometimes second core) while playing MC4..in antutu all cores max to 1300 mhz.. strange
Sent from my LG-P880 using xda app-developers app
Pretty sure that's the way it's supposed to work. Most non-tegra games rely on one processor and a gpu. I don't think there's a game out there that's actually optimized for, say, four cores and a gpu. ..that's why several newer chipsets go back to dual-core arm5 designs, and add more "gpu cores", for example..
Anyway. So one core active, max is 1500Mhz, four cores active 1300Mhz. Same as on any other processor - to stay within the thermal dispersion limit, the max clock is reduced when all cores are active.
It's also unlikely that 1300Mhz cores will cause any slowdown. Lots and lots of phones running on 1200Mhz, that these games will have to be compatible with. And one core for core logic/main thread typically is enough anyway.. What could happen is that the gpu cores are activated, the main thread response drops/clock-speed turned down, rinse, repeat. I.e., could be that the chipset is set up to throttle max speed on any activity on the other cores.
So if you play games that initially and always have full load on all cores won't change clock-speed, and therefore don't have the microstutter..?
Related
After quick view from some website , I found that the MSM7200 which is assembled in Kaiser is faster than Samsung 400Mhz and 500Mhz Processor.
Samsung 500Mhz(P3600i), Samsung 400Mhz(Dopod D810),TYTN II(MSM7200)
{
"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"
}
---------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------------------
Below is my summary of the test statistic above.
CPU Index
SAMSUNG 400Mh - 13XX
SAMSUNG 500Mhz - 14XX
MSM7200 400Mhz - 15XX
Graphics Index
SAMSUNG 400Mh - 9XX-12XX
SAMSUNG 500Mhz - 12XX
MSM7200 400Mhz - 3XXX
According to the statistic, seem the MSM7200 is faster than the Samsung 400 and 500Mhz CPU £¬ especially Graphics Index¡£ But in real world , how come it is underperformance and slower than 838Pro ? especially gaming , video and Emulator . Is the software optimization issue? direct draw issue or the test results are fake?
belon_cfy said:
According to the statistic, seem the MSM7200 is faster than the Samsung 400 and 500Mhz CPU £¬ especially Graphics Index¡£ But in real world , how come it is underperformance and slower than 838Pro ? especially gaming , video and Emulator . Is the software optimization issue? direct draw issue or the test results are fake?
Click to expand...
Click to collapse
Who knows what exactly these tests are actually testing. Everything, but the msm7200 devices, are certainly using their ATI chips.
Didn't MSM7200 had like a duo core type of working???
And cause not working 100% is cause of lack of drivers. But will be fixed end March.
xmoo said:
Didn't MSM7200 had like a duo core type of working???
And cause not working 100% is cause of lack of drivers. But will be fixed end March.
Click to expand...
Click to collapse
The driver is not a driver to enable ATI imagion.
Hopefully the new "Driver" can much improve 2D rendering performance for gaming , video and WM GUI.
Hello,
I'm curious as to the stock voltage of the captivate's CPU at 1Ghz. I've done some searching, and can't find it anywhere (maybe I'm looking in all the wrong places)
Because my 24/7 setup is:
Cyanogenmod 7
Launcher7
Glitch v13.1 kernel
1.4Ghz @ 1.325v
1.3Ghz @ 1.275v
1.2Ghz @ 1.25
1Ghz @ 1.2v
stock down the board from here
I/O scheduler: noop
Conservative governor
I'm just curious as to how much damage I'm doing to the chip, I'm not sure if 1.275v is the stock 1Ghz, just came here to check.
Thanks
(With the phone at 1.4 and running cyanogenmod, it absolutely screams, scores >2700 points in quadrant. It's no sgs2, but it holds its own)
1ghz stock is 1275mV
Here you go with all the defaults
{
"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"
}
Thanks guys
I remember reading somewhere that 1300mv was the max safe voltage.
mrhaley30705 said:
I remember reading somewhere that 1300mv was the max safe voltage.
Click to expand...
Click to collapse
Could you source that for me? I normally run 1.325 for 1.4
Sent from my SGH-I897 using XDA App
Fear of Oneself said:
Could you source that for me? I normally run 1.325 for 1.4
Sent from my SGH-I897 using XDA App
Click to expand...
Click to collapse
I heard from someone else that told me where he heard it. I don't know where you might be able to find official documentation on it. You could always email Samsung.
Technically anything past 1275mV and you are shortening the life expectancy. Don't really need to know much more than that. Simple rule about OC any device. More voltage and/or heat the more you kill the life of it. Since no one knows "that" number per device it comes down to how much you are willing to gamble on it.
prbassplayer said:
Technically anything past 1275mV and you are shortening the life expectancy. Don't really need to know much more than that. Simple rule about OC any device. More voltage and/or heat the more you kill the life of it. Since no one knows "that" number per device it comes down to how much you are willing to gamble on it.
Click to expand...
Click to collapse
This is true. Electromigration would take years on the 40 (or 45, I forget) nanometer chip, especially if it does not maintain that clock constantly. And temperature remains at hand because it only peaks at 1.4 in short bursts.
Thanks for the reply, I think I'll just keep it where I have it and not go any higher.
Sent from my SGH-I897 using XDA App
Ive just done a dirty flash of the 5.01 update -
Through fastboot I flashed system, boot loader and used the decrypted boot image.
Just run antutu & geekbench 3 and found the final score is pretty naff. Anyone else had similar or know why it might be?
{
"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"
}
Looked at better battery stats and noticed it was stuck at 1.7ghz. Does the nexus automatically limit the cpu when the battery is low?? I didn't enable battery saving mode
Amos91 said:
Ive just done a dirty flash of the 5.01 update -
Through fastboot I flashed system, boot loader and used the decrypted boot image.
Just run antutu & geekbench 3 and found the final score is pretty naff. Anyone else had similar or know why it might be?
Click to expand...
Click to collapse
i guaranty that your issue is your own fault. first off, i bet you just left the cpu speed as is(stock) then ran a benchmark test. well, guess what? that is the wrong way to benchmark. what cpu speed did the benchmarks run at? ill tell you, you have no idea. thats because itll vary however it feels like in between the max and minimum cpu speed tested. so, you couldve been testing a run at 1.5ghz instead of the 2.6 you thought you were using.
to run a benchmark properly, you have to set the max cpu and the min cpu to the max speed you want to test. or youll end up testing a lower speed. and if you are running stock, thatll be very hard for you since stock uses mpdecision. mpdecision will change your cpu speeds, even if you set them.
---------- Post added at 06:31 PM ---------- Previous post was at 06:30 PM ----------
Amos91 said:
Looked at better battery stats and noticed it was stuck at 1.7ghz. Does the nexus automatically limit the cpu when the battery is low?? I didn't enable battery saving mode
Click to expand...
Click to collapse
yes, at 20% or lower it reduces cpu speeds and amount of cores used.
My stupidity then. For some reason I didn't think it limited unless manually set to saver mode.
Amos91 said:
My stupidity then. For some reason I didn't think it limited unless manually set to saver mode.
Click to expand...
Click to collapse
yup. and at the moment, it automatically does this on custom roms as well.
Hello Xda :
I am currently running Franco_Kernel , stock Rom my CPU Voltage vs Freq Table is in the attachment and here :
{
"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"
}
https://goo.gl/photos/aNL4HjZ6VC9XU2Dg9
My CPU PVS_BIN IS 8 and i was on the prevois setting with max CPU speed 1.9Ghz and Min 0.7 GHz with liquid smooth_device
i deiced why i can't go further by under_volting more for each freq so 1.9 was on 930 mV so i put it on 910 to test if the devices will be stable or not and the device is stable as rock but i don't know why i feel slower performance , i know it's weird but does the under_volting_affect the performance even the cpu is stable or not and what i got are illusion ?
Thanks you in advance
muhamadamrueldan said:
Hello Xda :
I am currently running Franco_Kernel , stock Rom my CPU Voltage vs Freq Table is in the attachment and here :
https://goo.gl/photos/aNL4HjZ6VC9XU2Dg9
My CPU PVS_BIN IS 8 and i was on the prevois setting with max CPU speed 1.9Ghz and Min 0.7 GHz with liquid smooth_device
i deiced why i can't go further by under_volting more for each freq so 1.9 was on 930 mV so i put it on 910 to test if the devices will be stable or not and the device is stable as rock but i don't know why i feel slower performance , i know it's weird but does the under_volting_affect the performance even the cpu is stable or not and what i got are illusion ?
Thanks you in advance
Click to expand...
Click to collapse
yes, it can. i will never undervolte. most ppl undervolte because they think it saves battery. turning your brightness down saves you much more battery.
Shouldn't do really. If a cpu is running stable at a frequency, that frequency will always produce the same performance. It's pretty much impossible not too. Cpu frequency is the amount of clock cycles per second a cpu can run. Let's make this simple and use ridiculously low values that aren't possible, for the sake of easy maths....
Let's say your cpu is running at 10hz. That's 10 clock cycles per second. If opening your email client takes 10 clock cycles, it will take 1 second to open it. If it takes longer, the cpu isn't running at 10hz.
A clock cycle is the same.
Now, its possible that it isn't stable even though you think it is. The cpu either has enough power to run at that frequency or it doesn't. If it doesn't, it can become unstable. Now that would affect performance.
So if performance is affected, its unstable, if its stable, performance isn't affected.
Hello everyone,
I am running the latest stable firmware (10.2.24.0) and have noticed a weird behaviour regarding this device's minimum CPU frequency.
After starting up the device fresh, the minimum CPU States available are 576MHz for the small cores and 710 MHz for the big cores.(Screenshot)
After a day or two of use, the minimum CPU freq changes to 1036 MHz for the small cores and 1056 MHz for the big cores resulting to increased battery consumption. I cannot find what triggers this behaviour and the only way to make lower freqs available again is to reboot the device....
I believe there is something wrong with the device kernel, possibly something regarding the way the governor works.
I suggest we make some poll to see how many of you face this problem and maybe we can address it to Xiaomi.
I use cpu-z for accessing the CPU states. Can you please check and give me a feedback of the minimum CPU frequency you witness;
Thank you!
{
"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 MI 9 using Tapatalk
There are widespread complaints about the poor battery life and high idle drain on the MI 9. It's probably caused by the Snapdragon 855 since it's a new CPU but Qualcomm makes the firmware for their chips so a fix likely needs to come from them. Most people think the problems are with network searching which would indicate that there's a problem with the modem that hopefully can be fixed in a firmware update. You're the first person I know of that has suggested an issue with how the CPU cores are utilized. I think there are a couple of custom kernels in the dev thread if you think that's the problem.
I have also the same firmware and I can confirm the same problem. A reboot fixes the problem, which appears without apparent reason.
MysteriousStranger said:
I have also the same firmware and I can confirm the same problem. A reboot fixes the problem, which appears without apparent reason.
Click to expand...
Click to collapse
I have the same problem, but my Mi9 is chinese version, 10.2.25.0
mangaba said:
I have the same problem, but my Mi9 is chinese version, 10.2.25.0
Click to expand...
Click to collapse
Is there a post on Xiaomi relevant forum on this? This problem is very common but no one notices I guess
On 10.2.14.0 Global same issue after reboot cpu 0-3 @ 576mhz and cpu 4-6 @710mhz and cpu [email protected]
otherwise idle 0-3 @1036 and [email protected]
This is a long standing issue and xiaomi is known to stray away from fixing such issues. I wouldn't expect them to fix it voluntarily. If we're lucky, android 10+miui 11 will be free of this by virtue of rewritten code because they won't fix it just because 20 people are complaining. The impact to standby consumption is next to 0(at least as I analyzed on mine) so It's more of an OCD thing than anything else.