[java] In Java, what is the best way to determine the size of an object?

I have an application that reads a CSV file with piles of data rows. I give the user a summary of the number of rows based on types of data, but I want to make sure that I don't read in too many rows of data and cause OutOfMemoryErrors. Each row translates into an object. Is there an easy way to find out the size of that object programmatically? Is there a reference that defines how large primitive types and object references are for a VM?

Right now, I have code that says read up to 32,000 rows, but I'd also like to have code that says read as many rows as possible until I've used 32MB of memory. Maybe that is a different question, but I'd still like to know.

This question is related to java memory

The answer is


You should use jol, a tool developed as part of the OpenJDK project.

JOL (Java Object Layout) is the tiny toolbox to analyze object layout schemes in JVMs. These tools are using Unsafe, JVMTI, and Serviceability Agent (SA) heavily to decoder the actual object layout, footprint, and references. This makes JOL much more accurate than other tools relying on heap dumps, specification assumptions, etc.

To get the sizes of primitives, references and array elements, use VMSupport.vmDetails(). On Oracle JDK 1.8.0_40 running on 64-bit Windows (used for all following examples), this method returns

Running 64-bit HotSpot VM.
Using compressed oop with 0-bit shift.
Using compressed klass with 3-bit shift.
Objects are 8 bytes aligned.
Field sizes by type: 4, 1, 1, 2, 2, 4, 4, 8, 8 [bytes]
Array element sizes: 4, 1, 1, 2, 2, 4, 4, 8, 8 [bytes]

You can get the shallow size of an object instance using ClassLayout.parseClass(Foo.class).toPrintable() (optionally passing an instance to toPrintable). This is only the space consumed by a single instance of that class; it does not include any other objects referenced by that class. It does include VM overhead for the object header, field alignment and padding. For java.util.regex.Pattern:

java.util.regex.Pattern object internals:
 OFFSET  SIZE        TYPE DESCRIPTION                    VALUE
      0     4             (object header)                01 00 00 00 (0000 0001 0000 0000 0000 0000 0000 0000)
      4     4             (object header)                00 00 00 00 (0000 0000 0000 0000 0000 0000 0000 0000)
      8     4             (object header)                cb cf 00 20 (1100 1011 1100 1111 0000 0000 0010 0000)
     12     4         int Pattern.flags                  0
     16     4         int Pattern.capturingGroupCount    1
     20     4         int Pattern.localCount             0
     24     4         int Pattern.cursor                 48
     28     4         int Pattern.patternLength          0
     32     1     boolean Pattern.compiled               true
     33     1     boolean Pattern.hasSupplementary       false
     34     2             (alignment/padding gap)        N/A
     36     4      String Pattern.pattern                (object)
     40     4      String Pattern.normalizedPattern      (object)
     44     4        Node Pattern.root                   (object)
     48     4        Node Pattern.matchRoot              (object)
     52     4       int[] Pattern.buffer                 null
     56     4         Map Pattern.namedGroups            null
     60     4 GroupHead[] Pattern.groupNodes             null
     64     4       int[] Pattern.temp                   null
     68     4             (loss due to the next object alignment)
Instance size: 72 bytes (reported by Instrumentation API)
Space losses: 2 bytes internal + 4 bytes external = 6 bytes total

You can get a summary view of the deep size of an object instance using GraphLayout.parseInstance(obj).toFootprint(). Of course, some objects in the footprint might be shared (also referenced from other objects), so it is an overapproximation of the space that could be reclaimed when that object is garbage collected. For the result of Pattern.compile("^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\\.[a-zA-Z0-9-.]+$") (taken from this answer), jol reports a total footprint of 1840 bytes, of which only 72 are the Pattern instance itself.

java.util.regex.Pattern instance footprint:
     COUNT       AVG       SUM   DESCRIPTION
         1       112       112   [C
         3       272       816   [Z
         1        24        24   java.lang.String
         1        72        72   java.util.regex.Pattern
         9        24       216   java.util.regex.Pattern$1
        13        24       312   java.util.regex.Pattern$5
         1        16        16   java.util.regex.Pattern$Begin
         3        24        72   java.util.regex.Pattern$BitClass
         3        32        96   java.util.regex.Pattern$Curly
         1        24        24   java.util.regex.Pattern$Dollar
         1        16        16   java.util.regex.Pattern$LastNode
         1        16        16   java.util.regex.Pattern$Node
         2        24        48   java.util.regex.Pattern$Single
        40                1840   (total)

If you instead use GraphLayout.parseInstance(obj).toPrintable(), jol will tell you the address, size, type, value and path of field dereferences to each referenced object, though that's usually too much detail to be useful. For the ongoing pattern example, you might get the following. (Addresses will likely change between runs.)

java.util.regex.Pattern object externals:
          ADDRESS       SIZE TYPE                             PATH                           VALUE
         d5e5f290         16 java.util.regex.Pattern$Node     .root.next.atom.next           (object)
         d5e5f2a0        120 (something else)                 (somewhere else)               (something else)
         d5e5f318         16 java.util.regex.Pattern$LastNode .root.next.next.next.next.next.next.next (object)
         d5e5f328      21664 (something else)                 (somewhere else)               (something else)
         d5e647c8         24 java.lang.String                 .pattern                       (object)
         d5e647e0        112 [C                               .pattern.value                 [^, [, a, -, z, A, -, Z, 0, -, 9, _, ., +, -, ], +, @, [, a, -, z, A, -, Z, 0, -, 9, -, ], +, \, ., [, a, -, z, A, -, Z, 0, -, 9, -, ., ], +, $]
         d5e64850        448 (something else)                 (somewhere else)               (something else)
         d5e64a10         72 java.util.regex.Pattern                                         (object)
         d5e64a58        416 (something else)                 (somewhere else)               (something else)
         d5e64bf8         16 java.util.regex.Pattern$Begin    .root                          (object)
         d5e64c08         24 java.util.regex.Pattern$BitClass .root.next.atom.val$rhs        (object)
         d5e64c20        272 [Z                               .root.next.atom.val$rhs.bits   [false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, true, false, true, true, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, true, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false]
         d5e64d30         24 java.util.regex.Pattern$1        .root.next.atom.val$lhs.val$lhs.val$lhs.val$lhs.val$lhs.val$lhs (object)
         d5e64d48         24 java.util.regex.Pattern$1        .root.next.atom.val$lhs.val$lhs.val$lhs.val$lhs.val$lhs.val$rhs (object)
         d5e64d60         24 java.util.regex.Pattern$5        .root.next.atom.val$lhs.val$lhs.val$lhs.val$lhs.val$lhs (object)
         d5e64d78         24 java.util.regex.Pattern$1        .root.next.atom.val$lhs.val$lhs.val$lhs.val$lhs.val$rhs (object)
         d5e64d90         24 java.util.regex.Pattern$5        .root.next.atom.val$lhs.val$lhs.val$lhs.val$lhs (object)
         d5e64da8         24 java.util.regex.Pattern$5        .root.next.atom.val$lhs.val$lhs.val$lhs (object)
         d5e64dc0         24 java.util.regex.Pattern$5        .root.next.atom.val$lhs.val$lhs (object)
         d5e64dd8         24 java.util.regex.Pattern$5        .root.next.atom.val$lhs        (object)
         d5e64df0         24 java.util.regex.Pattern$5        .root.next.atom                (object)
         d5e64e08         32 java.util.regex.Pattern$Curly    .root.next                     (object)
         d5e64e28         24 java.util.regex.Pattern$Single   .root.next.next                (object)
         d5e64e40         24 java.util.regex.Pattern$BitClass .root.next.next.next.atom.val$rhs (object)
         d5e64e58        272 [Z                               .root.next.next.next.atom.val$rhs.bits [false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, true, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false]
         d5e64f68         24 java.util.regex.Pattern$1        .root.next.next.next.atom.val$lhs.val$lhs.val$lhs (object)
         d5e64f80         24 java.util.regex.Pattern$1        .root.next.next.next.atom.val$lhs.val$lhs.val$rhs (object)
         d5e64f98         24 java.util.regex.Pattern$5        .root.next.next.next.atom.val$lhs.val$lhs (object)
         d5e64fb0         24 java.util.regex.Pattern$1        .root.next.next.next.atom.val$lhs.val$rhs (object)
         d5e64fc8         24 java.util.regex.Pattern$5        .root.next.next.next.atom.val$lhs (object)
         d5e64fe0         24 java.util.regex.Pattern$5        .root.next.next.next.atom      (object)
         d5e64ff8         32 java.util.regex.Pattern$Curly    .root.next.next.next           (object)
         d5e65018         24 java.util.regex.Pattern$Single   .root.next.next.next.next      (object)
         d5e65030         24 java.util.regex.Pattern$BitClass .root.next.next.next.next.next.atom.val$rhs (object)
         d5e65048        272 [Z                               .root.next.next.next.next.next.atom.val$rhs.bits [false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, true, true, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false]
         d5e65158         24 java.util.regex.Pattern$1        .root.next.next.next.next.next.atom.val$lhs.val$lhs.val$lhs.val$lhs (object)
         d5e65170         24 java.util.regex.Pattern$1        .root.next.next.next.next.next.atom.val$lhs.val$lhs.val$lhs.val$rhs (object)
         d5e65188         24 java.util.regex.Pattern$5        .root.next.next.next.next.next.atom.val$lhs.val$lhs.val$lhs (object)
         d5e651a0         24 java.util.regex.Pattern$1        .root.next.next.next.next.next.atom.val$lhs.val$lhs.val$rhs (object)
         d5e651b8         24 java.util.regex.Pattern$5        .root.next.next.next.next.next.atom.val$lhs.val$lhs (object)
         d5e651d0         24 java.util.regex.Pattern$5        .root.next.next.next.next.next.atom.val$lhs (object)
         d5e651e8         24 java.util.regex.Pattern$5        .root.next.next.next.next.next.atom (object)
         d5e65200         32 java.util.regex.Pattern$Curly    .root.next.next.next.next.next (object)
         d5e65220        120 (something else)                 (somewhere else)               (something else)
         d5e65298         24 java.util.regex.Pattern$Dollar   .root.next.next.next.next.next.next (object)

The "(something else)" entries describe other objects in the heap that are not part of this object graph.

The best jol documentation is the jol samples in the jol repository. The samples demonstrate common jol operations and show how you can use jol to analyze VM and garbage collector internals.


Here is a utility I made using some of the linked examples to handle 32-bit, 64-bit and 64-bit with compressed OOP. It uses sun.misc.Unsafe.

It uses Unsafe.addressSize() to get the size of a native pointer and Unsafe.arrayIndexScale( Object[].class ) for the size of a Java reference.

It uses the field offset of a known class to work out the base size of an object.

import java.lang.reflect.Array;
import java.lang.reflect.Field;
import java.lang.reflect.Modifier;
import java.util.IdentityHashMap;
import java.util.Stack;
import sun.misc.Unsafe;

/** Usage: 
 * MemoryUtil.sizeOf( object )
 * MemoryUtil.deepSizeOf( object )
 * MemoryUtil.ADDRESS_MODE
 */
public class MemoryUtil
{
    private MemoryUtil()
    {
    }

    public static enum AddressMode
    {
        /** Unknown address mode. Size calculations may be unreliable. */
        UNKNOWN,
        /** 32-bit address mode using 32-bit references. */
        MEM_32BIT,
        /** 64-bit address mode using 64-bit references. */
        MEM_64BIT,
        /** 64-bit address mode using 32-bit compressed references. */
        MEM_64BIT_COMPRESSED_OOPS
    }

    /** The detected runtime address mode. */
    public static final AddressMode ADDRESS_MODE;

    private static final Unsafe UNSAFE;

    private static final long ADDRESS_SIZE; // The size in bytes of a native pointer: 4 for 32 bit, 8 for 64 bit
    private static final long REFERENCE_SIZE; // The size of a Java reference: 4 for 32 bit, 4 for 64 bit compressed oops, 8 for 64 bit
    private static final long OBJECT_BASE_SIZE; // The minimum size of an Object: 8 for 32 bit, 12 for 64 bit compressed oops, 16 for 64 bit
    private static final long OBJECT_ALIGNMENT = 8;

    /** Use the offset of a known field to determine the minimum size of an object. */
    private static final Object HELPER_OBJECT = new Object() { byte b; };


    static
    {
        try
        {
            // Use reflection to get a reference to the 'Unsafe' object.
            Field f = Unsafe.class.getDeclaredField( "theUnsafe" );
            f.setAccessible( true );
            UNSAFE = (Unsafe) f.get( null );

            OBJECT_BASE_SIZE = UNSAFE.objectFieldOffset( HELPER_OBJECT.getClass().getDeclaredField( "b" ) );

            ADDRESS_SIZE = UNSAFE.addressSize();
            REFERENCE_SIZE = UNSAFE.arrayIndexScale( Object[].class );

            if( ADDRESS_SIZE == 4 )
            {
                ADDRESS_MODE = AddressMode.MEM_32BIT;
            }
            else if( ADDRESS_SIZE == 8 && REFERENCE_SIZE == 8 )
            {
                ADDRESS_MODE = AddressMode.MEM_64BIT;
            }
            else if( ADDRESS_SIZE == 8 && REFERENCE_SIZE == 4 )
            {
                ADDRESS_MODE = AddressMode.MEM_64BIT_COMPRESSED_OOPS;
            }
            else
            {
                ADDRESS_MODE = AddressMode.UNKNOWN;
            }
        }
        catch( Exception e )
        {
            throw new Error( e );
        }
    }


    /** Return the size of the object excluding any referenced objects. */
    public static long shallowSizeOf( final Object object )
    {
        Class<?> objectClass = object.getClass();
        if( objectClass.isArray() )
        {
            // Array size is base offset + length * element size
            long size = UNSAFE.arrayBaseOffset( objectClass )
                    + UNSAFE.arrayIndexScale( objectClass ) * Array.getLength( object );
            return padSize( size );
        }
        else
        {
            // Object size is the largest field offset padded out to 8 bytes
            long size = OBJECT_BASE_SIZE;
            do
            {
                for( Field field : objectClass.getDeclaredFields() )
                {
                    if( (field.getModifiers() & Modifier.STATIC) == 0 )
                    {
                        long offset = UNSAFE.objectFieldOffset( field );
                        if( offset >= size )
                        {
                            size = offset + 1; // Field size is between 1 and PAD_SIZE bytes. Padding will round up to padding size.
                        }
                    }
                }
                objectClass = objectClass.getSuperclass();
            }
            while( objectClass != null );

            return padSize( size );
        }
    }


    private static final long padSize( final long size )
    {
        return (size + (OBJECT_ALIGNMENT - 1)) & ~(OBJECT_ALIGNMENT - 1);
    }


    /** Return the size of the object including any referenced objects. */
    public static long deepSizeOf( final Object object )
    {
        IdentityHashMap<Object,Object> visited = new IdentityHashMap<Object,Object>();
        Stack<Object> stack = new Stack<Object>();
        if( object != null ) stack.push( object );

        long size = 0;
        while( !stack.isEmpty() )
        {
            size += internalSizeOf( stack.pop(), stack, visited );
        }
        return size;
    }


    private static long internalSizeOf( final Object object, final Stack<Object> stack, final IdentityHashMap<Object,Object> visited )
    {
        // Scan for object references and add to stack
        Class<?> c = object.getClass();
        if( c.isArray() && !c.getComponentType().isPrimitive() )
        {
            // Add unseen array elements to stack
            for( int i = Array.getLength( object ) - 1; i >= 0; i-- )
            {
                Object val = Array.get( object, i );
                if( val != null && visited.put( val, val ) == null )
                {
                    stack.add( val );
                }
            }
        }
        else
        {
            // Add unseen object references to the stack
            for( ; c != null; c = c.getSuperclass() )
            {
                for( Field field : c.getDeclaredFields() )
                {
                    if( (field.getModifiers() & Modifier.STATIC) == 0 
                            && !field.getType().isPrimitive() )
                    {
                        field.setAccessible( true );
                        try
                        {
                            Object val = field.get( object );
                            if( val != null && visited.put( val, val ) == null )
                            {
                                stack.add( val );
                            }
                        }
                        catch( IllegalArgumentException e )
                        {
                            throw new RuntimeException( e );
                        }
                        catch( IllegalAccessException e )
                        {
                            throw new RuntimeException( e );
                        }
                    }
                }
            }
        }

        return shallowSizeOf( object );
    }
}

This answer is not related to Object size, but when you are using array to accommodate the objects; how much memory size it will allocate for the object.

So arrays, list, or map all those collection won't be going to store objects really (only at the time of primitives, real object memory size is needed), it will store only references for those objects.

Now the Used heap memory = sizeOfObj + sizeOfRef (* 4 bytes) in collection

  • (4/8 bytes) depends on (32/64 bit) OS

PRIMITIVES

int   [] intArray    = new int   [1]; will require 4 bytes.
long  [] longArray   = new long  [1]; will require 8 bytes.

OBJECTS

Object[] objectArray = new Object[1]; will require 4 bytes. The object can be any user defined Object.
Long  [] longArray   = new Long  [1]; will require 4 bytes.

I mean to say all the object REFERENCE needs only 4 bytes of memory. It may be String reference OR Double object reference, But depends on object creation the memory needed will vary.

e.g) If i create object for the below class ReferenceMemoryTest then 4 + 4 + 4 = 12 bytes of memory will be created. The memory may differ when you are trying to initialize the references.

 class ReferenceMemoryTest {
    public String refStr;
    public Object refObj;
    public Double refDoub; 
}

So when are creating object/reference array, all its contents will be occupied with NULL references. And we know each reference requires 4 bytes.

And finally, memory allocation for the below code is 20 bytes.

ReferenceMemoryTest ref1 = new ReferenceMemoryTest(); ( 4(ref1) + 12 = 16 bytes) ReferenceMemoryTest ref2 = ref1; ( 4(ref2) + 16 = 20 bytes)


long heapSizeBefore = Runtime.getRuntime().totalMemory();

// Code for object construction
...
long heapSizeAfter = Runtime.getRuntime().totalMemory();
long size = heapSizeAfter - heapSizeBefore;

size gives you the increase in memory usage of the jvm due to object creation and that typically is the size of the object.


My answer is based on the code supplied by Nick. That code measures total amount of bytes which are occupied by the serialized object. So this actually measures serialization stuff + plain object memory footprint (just serialize for example int and you will see that total amount of serialized bytes is not 4). So if you want to get raw byte number used exactly for your object - you need to modify that code a bit. Like so:

import java.io.ByteArrayOutputStream;
import java.io.ObjectOutputStream;
import java.io.Serializable;

public class ObjectSizeCalculator {
    private Object getFirstObjectReference(Object o) {
        String objectType = o.getClass().getTypeName();

        if (objectType.substring(objectType.length()-2).equals("[]")) {
            try {
                if (objectType.equals("java.lang.Object[]"))
                    return ((Object[])o)[0];
                else if (objectType.equals("int[]"))
                    return ((int[])o)[0];
                else
                    throw new RuntimeException("Not Implemented !");
            } catch (IndexOutOfBoundsException e) {
                return null;
            }
        }

        return o;
    } 

    public int getObjectSizeInBytes(Object o) {
        final String STRING_JAVA_TYPE_NAME = "java.lang.String";

        if (o == null)
            return 0;

        String objectType = o.getClass().getTypeName();
        boolean isArray = objectType.substring(objectType.length()-2).equals("[]");

        Object objRef = getFirstObjectReference(o);
        if (objRef != null && !(objRef instanceof Serializable))
            throw new RuntimeException("Object must be serializable for measuring it's memory footprint using this method !");

        try {
            ByteArrayOutputStream baos = new ByteArrayOutputStream();
            ObjectOutputStream oos = new ObjectOutputStream(baos);
            oos.writeObject(o);
            oos.close();
            byte[] bytes = baos.toByteArray();

            for (int i = bytes.length - 1, j = 0; i != 0; i--, j++) {
                if (objectType != STRING_JAVA_TYPE_NAME) {
                    if (bytes[i] == 112)
                        if (isArray)
                            return j - 4;
                        else
                            return j;
                } else {
                    if (bytes[i] == 0)
                        return j - 1;
                }
            }
        } catch (Exception e) {
            return -1;
        }

        return -1;
    }    

}

I've tested this solution with primitive types, String, and on some trivial classes. There may be not covered cases also.


UPDATE: Example modified to support memory footprint calculation of array objects.


You have to walk the objects using reflection. Be careful as you do:

  • Just allocating an object has some overhead in the JVM. The amount varies by JVM so you might make this value a parameter. At least make it a constant (8 bytes?) and apply to anything allocated.
  • Just because byte is theoretically 1 byte doesn't mean it takes just one in memory.
  • There will be loops in object references, so you'll need to keep a HashMap or somesuch using object-equals as the comparator to eliminate infinite loops.

@jodonnell: I like the simplicity of your solution, but many objects aren't Serializable (so this would throw an exception), fields can be transient, and objects can override the standard methods.


I was looking for a runtime calculation of an object size that met the following requirements:

  • Available at runtime with no need to include instrumentation.
  • Works with Java 9+ without access to Unsafe.
  • Is based on the Class only. Not a deep sizeOf that takes into consideration string lengths, array lengths, etc.

The following is based on the core code of the original java specialists article (https://www.javaspecialists.eu/archive/Issue078.html) and a few bits from the Unsafe version in another answer to this question.

I hope someone finds it useful.

public class JavaSize {

private static final int NR_BITS = Integer.valueOf(System.getProperty("sun.arch.data.model"));
private static final int BYTE = 8;
private static final int WORD = NR_BITS / BYTE;
private static final int HEADER_SIZE = 8;

public static int sizeOf(Class<?> clazz) {
    int result = 0;

    while (clazz != null) {
        Field[] fields = clazz.getDeclaredFields();
        for (int i = 0; i < fields.length; i++) {
            if (!Modifier.isStatic(fields[i].getModifiers())) {
                if (fields[i].getType().isPrimitive()) {
                    Class<?> primitiveClass = fields[i].getType();
                    if (primitiveClass == boolean.class || primitiveClass == byte.class) {
                        result += 1;
                    } else if (primitiveClass == short.class) {
                        result += 2;
                    } else if (primitiveClass == int.class || primitiveClass == float.class) {
                        result += 4;
                    } else if (primitiveClass == double.class || primitiveClass == long.class) {
                        result += 8;
                    }

                } else {
                    // assume compressed references.
                    result += 4;
                }
            }
        }

        clazz = clazz.getSuperclass();

        // round up to the nearest WORD length.
        if ((result % WORD) != 0) {
            result += WORD - (result % WORD);
        }
    }

    result += HEADER_SIZE;

    return result;
}

}


Back when I worked at Twitter, I wrote a utility for calculating deep object size. It takes into account different memory models (32-bit, compressed oops, 64-bit), padding, subclass padding, works correctly on circular data structures and arrays. You can just compile this one .java file; it has no external dependencies:

https://github.com/twitter/commons/blob/master/src/java/com/twitter/common/objectsize/ObjectSizeCalculator.java


Suppose I declare a class named Complex like:

public class Complex {

    private final long real;
    private final long imaginary;

    // omitted
}

In order to see how much memory is allocated to live instances of this class:

$ jmap -histo:live <pid> | grep Complex

 num     #instances         #bytes  class name (module)
-------------------------------------------------------
 327:             1             32  Complex

There isn't a method call, if that's what you're asking for. With a little research, I suppose you could write your own. A particular instance has a fixed sized derived from the number of references and primitive values plus instance bookkeeping data. You would simply walk the object graph. The less varied the row types, the easier.

If that's too slow or just more trouble than it's worth, there's always good old-fashioned row counting rule-of-thumbs.


I wrote a quick test once to estimate on the fly:

public class Test1 {

    // non-static nested
    class Nested { }

    // static nested
    static class StaticNested { }

    static long getFreeMemory () {
        // waits for free memory measurement to stabilize
        long init = Runtime.getRuntime().freeMemory(), init2;
        int count = 0;
        do {
            System.out.println("waiting..." + init);
            System.gc();
            try { Thread.sleep(250); } catch (Exception x) { }
            init2 = init;
            init = Runtime.getRuntime().freeMemory();
            if (init == init2) ++ count; else count = 0;
        } while (count < 5);
        System.out.println("ok..." + init);
        return init;
    }

    Test1 () throws InterruptedException {

        Object[] s = new Object[10000];
        Object[] n = new Object[10000];
        Object[] t = new Object[10000];

        long init = getFreeMemory();

        //for (int j = 0; j < 10000; ++ j)
        //    s[j] = new Separate();

        long afters = getFreeMemory();

        for (int j = 0; j < 10000; ++ j)
            n[j] = new Nested();

        long aftersn = getFreeMemory();

        for (int j = 0; j < 10000; ++ j)
            t[j] = new StaticNested();

        long aftersnt = getFreeMemory();

        System.out.println("separate:      " + -(afters - init) + " each=" + -(afters - init) / 10000);
        System.out.println("nested:        " + -(aftersn - afters) + " each=" + -(aftersn - afters) / 10000);
        System.out.println("static nested: " + -(aftersnt - aftersn) + " each=" + -(aftersnt - aftersn) / 10000);

    }

    public static void main (String[] args) throws InterruptedException {
        new Test1();
    }

}

General concept is allocate objects and measure change in free heap space. The key being getFreeMemory(), which requests GC runs and waits for the reported free heap size to stabilize. The output of the above is:

nested:        160000 each=16
static nested: 160000 each=16

Which is what we expect, given alignment behavior and possible heap block header overhead.

The instrumentation method detailed in the accepted answer here the most accurate. The method I described is accurate but only under controlled conditions where no other threads are creating/discarding objects.


Without having to mess with instrumentation and so on, and if you don't need to know the byte-exact size of an object, you could go with the following approach:

System.gc();
Runtime.getRuntime().totalMemory() - Runtime.getRuntime().freeMemory();

do your job here

System.gc();
Runtime.getRuntime().totalMemory() - Runtime.getRuntime().freeMemory();

This way you read the used memory before and after, and calling the GC just before getting the used memory you lower the "noise" almost to 0.

For a more reliable result you can run your job n times, and then divide the used memory by n, obtaining how much memory one run takes. Even more, you can run the whole thing more times and make an average.


Just use java visual VM.

It has everything you need to profile and debug memory problems.

It also has a OQL (Object Query Language) console which allows you to do many useful things, one of which being sizeof(o)


I accidentally found a java class "jdk.nashorn.internal.ir.debug.ObjectSizeCalculator", already in jdk, which is easy to use and seems quite useful for determining the size of an object.

System.out.println(ObjectSizeCalculator.getObjectSize(new gnu.trove.map.hash.TObjectIntHashMap<String>(12000, 0.6f, -1)));
System.out.println(ObjectSizeCalculator.getObjectSize(new HashMap<String, Integer>(100000)));
System.out.println(ObjectSizeCalculator.getObjectSize(3));
System.out.println(ObjectSizeCalculator.getObjectSize(new int[]{1, 2, 3, 4, 5, 6, 7 }));
System.out.println(ObjectSizeCalculator.getObjectSize(new int[100]));

results:

164192
48
16
48
416

Firstly "the size of an object" isn't a well-defined concept in Java. You could mean the object itself, with just its members, the Object and all objects it refers to (the reference graph). You could mean the size in memory or the size on disk. And the JVM is allowed to optimise things like Strings.

So the only correct way is to ask the JVM, with a good profiler (I use YourKit), which probably isn't what you want.

However, from the description above it sounds like each row will be self-contained, and not have a big dependency tree, so the serialization method will probably be a good approximation on most JVMs. The easiest way to do this is as follows:

 Serializable ser;
 ByteArrayOutputStream baos = new ByteArrayOutputStream();
 ObjectOutputStream oos = new ObjectOutputStream(baos);
 oos.writeObject(ser);
 oos.close();
 return baos.size();

Remember that if you have objects with common references this will not give the correct result, and size of serialization will not always match size in memory, but it is a good approximation. The code will be a bit more efficient if you initialise the ByteArrayOutputStream size to a sensible value.


For JSONObject the below code can help you.

`JSONObject.toString().getBytes("UTF-8").length`

returns size in bytes

I checked it with my JSONArray object by writing it to a file. It is giving object size.


I doubt you want to do it programmatically unless you just want to do it once and store it for future use. It's a costly thing to do. There's no sizeof() operator in Java, and even if there was, it would only count the cost of the references to other objects and the size of the primitives.

One way you could do it is to serialize the thing to a File and look at the size of the file, like this:

Serializable myObject;
ObjectOutputStream oos = new ObjectOutputStream (new FileOutputStream ("obj.ser"));
oos.write (myObject);
oos.close ();

Of course, this assumes that each object is distinct and doesn't contain non-transient references to anything else.

Another strategy would be to take each object and examine its members by reflection and add up the sizes (boolean & byte = 1 byte, short & char = 2 bytes, etc.), working your way down the membership hierarchy. But that's tedious and expensive and ends up doing the same thing the serialization strategy would do.


The java.lang.instrument.Instrumentation class provides a nice way to get the size of a Java Object, but it requires you to define a premain and run your program with a java agent. This is very boring when you do not need any agent and then you have to provide a dummy Jar agent to your application.

So I got an alternative solution using the Unsafe class from the sun.misc. So, considering the objects heap alignment according to the processor architecture and calculating the maximum field offset, you can measure the size of a Java Object. In the example below I use an auxiliary class UtilUnsafe to get a reference to the sun.misc.Unsafe object.

private static final int NR_BITS = Integer.valueOf(System.getProperty("sun.arch.data.model"));
private static final int BYTE = 8;
private static final int WORD = NR_BITS/BYTE;
private static final int MIN_SIZE = 16; 

public static int sizeOf(Class src){
    //
    // Get the instance fields of src class
    // 
    List<Field> instanceFields = new LinkedList<Field>();
    do{
        if(src == Object.class) return MIN_SIZE;
        for (Field f : src.getDeclaredFields()) {
            if((f.getModifiers() & Modifier.STATIC) == 0){
                instanceFields.add(f);
            }
        }
        src = src.getSuperclass();
    }while(instanceFields.isEmpty());
    //
    // Get the field with the maximum offset
    //  
    long maxOffset = 0;
    for (Field f : instanceFields) {
        long offset = UtilUnsafe.UNSAFE.objectFieldOffset(f);
        if(offset > maxOffset) maxOffset = offset; 
    }
    return  (((int)maxOffset/WORD) + 1)*WORD; 
}
class UtilUnsafe {
    public static final sun.misc.Unsafe UNSAFE;

    static {
        Object theUnsafe = null;
        Exception exception = null;
        try {
            Class<?> uc = Class.forName("sun.misc.Unsafe");
            Field f = uc.getDeclaredField("theUnsafe");
            f.setAccessible(true);
            theUnsafe = f.get(uc);
        } catch (Exception e) { exception = e; }
        UNSAFE = (sun.misc.Unsafe) theUnsafe;
        if (UNSAFE == null) throw new Error("Could not obtain access to sun.misc.Unsafe", exception);
    }
    private UtilUnsafe() { }
}

When using JetBrains IntelliJ, first enable "Attach memory agent" in File | Settings | Build, Execution, Deployment | Debugger.

When debugging, right-click a variable of interest and choose "Calculate Retained Size": Calculate Retained Size


Some years back Javaworld had an article on determining the size of composite and potentially nested Java objects, they basically walk through creating a sizeof() implementation in Java. The approach basically builds on other work where people experimentally identified the size of primitives and typical Java objects and then apply that knowledge to a method that recursively walks an object graph to tally the total size.

It is always going to be somewhat less accurate than a native C implementation simply because of the things going on behind the scenes of a class but it should be a good indicator.

Alternatively a SourceForge project appropriately called sizeof that offers a Java5 library with a sizeof() implementation.

P.S. Do not use the serialization approach, there is no correlation between the size of a serialized object and the amount of memory it consumes when live.


There is also the Memory Measurer tool (formerly at Google Code, now on GitHub), which is simple and published under the commercial-friendly Apache 2.0 license, as discussed in a similar question.

It, too, requires a command-line argument to the java interpreter if you want to measure memory byte consumption, but otherwise seems to work just fine, at least in the scenarios I have used it.


If you would just like to know how much memory is being used in your JVM, and how much is free, you could try something like this:

// Get current size of heap in bytes
long heapSize = Runtime.getRuntime().totalMemory();

// Get maximum size of heap in bytes. The heap cannot grow beyond this size.
// Any attempt will result in an OutOfMemoryException.
long heapMaxSize = Runtime.getRuntime().maxMemory();

// Get amount of free memory within the heap in bytes. This size will increase
// after garbage collection and decrease as new objects are created.
long heapFreeSize = Runtime.getRuntime().freeMemory();

edit: I thought this might be helpful as the question author also stated he would like to have logic that handles "read as many rows as possible until I've used 32MB of memory."


Much of the other answers provide shallow sizes - e.g. the size of a HashMap without any of the keys or values, which isn't likely what you want.

The jamm project uses the java.lang.instrumentation package above but walks the tree and so can give you the deep memory use.

new MemoryMeter().measureDeep(myHashMap);

https://github.com/jbellis/jamm

To use MemoryMeter, start the JVM with "-javaagent:/jamm.jar"


You have to measure it with a tool, or estimate it by hand, and it depends on the JVM you are using.

There is some fixed overhead per object. It's JVM-specific, but I usually estimate 40 bytes. Then you have to look at the members of the class. Object references are 4 (8) bytes in a 32-bit (64-bit) JVM. Primitive types are:

  • boolean and byte: 1 byte
  • char and short: 2 bytes
  • int and float: 4 bytes
  • long and double: 8 bytes

Arrays follow the same rules; that is, it's an object reference so that takes 4 (or 8) bytes in your object, and then its length multiplied by the size of its element.

Trying to do it programmatically with calls to Runtime.freeMemory() just doesn't give you much accuracy, because of asynchronous calls to the garbage collector, etc. Profiling the heap with -Xrunhprof or other tools will give you the most accurate results.


You could generate a heap dump (with jmap, for example) and then analyze the output to find object sizes. This is an offline solution, but you can examine shallow and deep sizes, etc.