CODEWITHSUNDEEP
android
Vivek Warde
Vivek Warde

Reputation: 1514

How to get the number of cores of an android device

I was looking to determine(or count) the number of cores in the embedded processor of android device.

I tried using /proc/cpuinfo, but it is not returning the number of cores of device !

I don't found the code anywhere on the Internet. Does anyone here know how can I determine this, then please answer. Thanks in advance

UPDATE:

The answers on this question How can you detect a dual-core cpu on an Android device from code? doesn't run well in some devices. I tested them is dual core & quad core processors, they returned 2 & 4 respectively, fine !

But, On Octa Core processor like in Samsung Note 3 it returned 4. (Perhaps in note 3 there are 2 sets of quad core processors running individually )

I was looking to solve this problem.

UPDATE

The app CPU-Z is returning the correct core count in my device Samsung note 3 enter image description here Here it seems that there exists a possible solution...

Upvotes: 44

Views: 40407

Answers (9)

Ghosthunter
Ghosthunter

Reputation: 1

Did you try cat /proc/cpuinfo

It works for me.

Upvotes: 0

user14555681
user14555681

Reputation:

Here you go (Java):

    try {
        int ch, processorCount = 0;
        Process process = Runtime.getRuntime().exec("cat /proc/cpuinfo");
        StringBuilder sb = new StringBuilder();
        while ((ch = process.getInputStream().read()) != -1)
            sb.append((char) ch);
        Pattern p = Pattern.compile("processor");
        Matcher m = p.matcher(sb.toString());
        while (m.find())
            processorCount++;
        System.out.println(processorCount);
    } catch (IOException e) {
        e.printStackTrace();
    }

Upvotes: 1

G00fY
G00fY

Reputation: 5357

Nice and simple solution in kotlin:

fun getCPUCoreNum(): Int {
  val pattern = Pattern.compile("cpu[0-9]+")
  return Math.max(
    File("/sys/devices/system/cpu/")
      .walk()
      .maxDepth(1)
      .count { pattern.matcher(it.name).matches() },
    Runtime.getRuntime().availableProcessors()
  )
}

Upvotes: 1

Oleg Bezverhiy
Oleg Bezverhiy

Reputation: 31

good thing is to use JNI, if possible in project

std::thread::hardware_concurrency()

Upvotes: 0

Saurabh Ande
Saurabh Ande

Reputation: 427

You can try this method to get number of core. 

 private int getNumOfCores()
      {
        try
        {
          int i = new File("/sys/devices/system/cpu/").listFiles(new FileFilter()
          {
            public boolean accept(File params)
            {
              return Pattern.matches("cpu[0-9]", params.getName());
            }
          }).length;
          return i;
        }
        catch (Exception e)
        {
              e.printstacktrace();
        }
        return 1;
      }

Upvotes: 4

Anthony
Anthony

Reputation: 3074

Firstly, it seems like this is not possible, here is some information on this:

We will use the Samsung Exynos 5 Octa CPU as an example for this thread, but generally any ARM processor will work the same.

The Samsung Exynos 5 Octa CPU has 8 CPU cores. but in reality it has two processes with 4 cores each. This is called big.LITTLE. You have 1 fast processor and 1 power efficient processor.

Within the Exynos 5 CPU it has two different CPUs in built, a Cortex a15 and a Cortex a7...

Within big.LITTLE both CPUs cannot run at the same time, only one CPU can be active at any given time...

But there is also something called "big.LITTLE mp" in which both CPU's can be active at the same time, but here is the catch (again!) Only FOUR cores can be active to the software at any given time. Here is an image to explain this a bit better:

big.LITTLE mp

Now as you can see, a single CPU core is used from the more powerful processor here and then the other four cores are active from the more energy efficient a7 CPU cluster.

You can read more about the big.LITTLE architecture here : http://www.arm.com/products/processors/technologies/biglittleprocessing.php

You can read more about the big.LITTLE mp architecture here: http://www.arm.com/products/processors/technologies/biglittleprocessing.php

What you want to find out now, is if it is possible to know if the CPU architecture is Big.LITTLE or BIG.LITTLE mp. Then see if you can directly find out the CPU count from each individual CPU, but I cannot find any relevant code. There is a lot of documentation on ARMs website, but I am not too sure if it is possible to get this information. Either way though, only 4 CPU cores can be used therefor the code you have is technically correct as that is the number of useable cores.

I hope this helped, if you have any questions or want anything clearing up then let me know. There is a lot of information out there

Upvotes: 11

Anirudh Sharma
Anirudh Sharma

Reputation: 7974

Use this to get no. of cores.Do read this link very carefully and also see what the author is trying to tell between virtual device and a real device.

The code is referred from THIS SITE

/**
 * Gets the number of cores available in this device, across all processors.
 * Requires: Ability to peruse the filesystem at "/sys/devices/system/cpu"
 * @return The number of cores, or 1 if failed to get result
 */
private int getNumCores() {
    //Private Class to display only CPU devices in the directory listing
    class CpuFilter implements FileFilter {
        @Override
        public boolean accept(File pathname) {
            //Check if filename is "cpu", followed by a single digit number
            if(Pattern.matches("cpu[0-9]+", pathname.getName())) {
                return true;
            }
            return false;
        }      
    }

    try {
        //Get directory containing CPU info
        File dir = new File("/sys/devices/system/cpu/");
        //Filter to only list the devices we care about
        File[] files = dir.listFiles(new CpuFilter());
        Log.d(TAG, "CPU Count: "+files.length);
        //Return the number of cores (virtual CPU devices)
        return files.length;
    } catch(Exception e) {
        //Print exception
        Log.d(TAG, "CPU Count: Failed.");
        e.printStackTrace();
        //Default to return 1 core
        return 1;
    }
}

Upvotes: 0

Bojan Kseneman
Bojan Kseneman

Reputation: 15668

You may use a combination of above answer with this one. This way it will perform faster on devices running API 17+ as this method is much faster than filtering out files.

private int getNumberOfCores() {
    if(Build.VERSION.SDK_INT >= 17) {
        return Runtime.getRuntime().availableProcessors()
    }
    else {
       // Use saurabh64's answer
       return getNumCoresOldPhones();
    }
}

/**
 * Gets the number of cores available in this device, across all processors.
 * Requires: Ability to peruse the filesystem at "/sys/devices/system/cpu"
 * @return The number of cores, or 1 if failed to get result
 */
private int getNumCoresOldPhones() {
    //Private Class to display only CPU devices in the directory listing
    class CpuFilter implements FileFilter {
        @Override
        public boolean accept(File pathname) {
            //Check if filename is "cpu", followed by a single digit number
            if(Pattern.matches("cpu[0-9]+", pathname.getName())) {
                return true;
            }
            return false;
        }      
    }

    try {
        //Get directory containing CPU info
        File dir = new File("/sys/devices/system/cpu/");
        //Filter to only list the devices we care about
        File[] files = dir.listFiles(new CpuFilter());
        //Return the number of cores (virtual CPU devices)
        return files.length;
    } catch(Exception e) {
        //Default to return 1 core
        return 1;
    }
}

public int availableProcessors ()

Added in API level 1 Returns the number of processor cores available to the VM, at least 1. Traditionally this returned the number currently online, but many mobile devices are able to take unused cores offline to save power, so releases newer than Android 4.2 (Jelly Bean) return the maximum number of cores that could be made available if there were no power or heat constraints.

Also there is information about number of cores inside a file located in /sys/devices/system/cpu/present It reports the number of available CPUs in the following format:

  • 0 -> single CPU/core
  • 0-1 -> two CPUs/cores
  • 0-3 -> four CPUs/cores
  • 0-7 -> eight CPUs/cores

etc.

Also please check what is inside /sys/devices/system/cpu/possible on a note 3

Both files have r--r--r-- or 444 file permissions set on them, so you should be able to read them without a rooted device in code.

EDIT: Posting code to help you out

 private void printNumberOfCores() {
        printFile("/sys/devices/system/cpu/present");
        printFile("/sys/devices/system/cpu/possible");

    }

    private void printFile(String path) {
        InputStream inputStream = null;
        try {
            inputStream = new FileInputStream(path);
            if (inputStream != null) {

                BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(inputStream));
                String line;

                do {
                    line = bufferedReader.readLine();
                    Log.d(path, line);
                } while (line != null);
            }
        } catch (Exception e) {

        } finally {
            if (inputStream != null) {
                try {
                    inputStream.close();
                } catch (IOException e) {
                }
            }
        }
    }

With the result

D//sys/devices/system/cpu/present﹕ 0-3
D//sys/devices/system/cpu/possible﹕ 0-3

The test was run on OnePlus One running BlissPop ROM with Android v5.1.1 and it prints as it should. Please try on your Samsung

Upvotes: 57

SilentKnight
SilentKnight

Reputation: 14021

Try this:

Runtime.getRuntime().availableProcessors();

This returns the number of CPU's available for THIS specific virtual machine, as I experienced. That may not be what you want, still, for a few purposes this is very handy. You can test this really easily: Kill all apps, run the above code. Open 10 very intensive apps, and then run the test again. Sure, this will only work on a multi cpu device, I guess.

Upvotes: 7

Related Questions