Java：读取序列化文件时堆空间不足

Question

In one method I create a serialized file from an adjacency list which I'm trying to read in another method as follows:

Map<Integer, List<Integer>> adj;
        int[] count;
        int N;
        ArrayList<Double> weights;

            try (
                    ObjectInputStream input1 = new ObjectInputStream(new FileInputStream("GIRGnew5adj.ser"));
                ){


        N = (int)input2.readObject();     
        adj = (Map<Integer, List<Integer>>)input1.readObject();

Now, while the file itself is around 500MB large I get an OOM:Java Heap Space error even when I increase my heapsize to 2000MB but then again it works for a heap of 6000MB.

I wonder, how come does the same information take so much more space in memory than on hard disk?

Answer 1

This is a very rough path of what happens. Just so you have an idea what the computer does when load the file. This will not be 100% accurate and actual process might differ, be more efficient, this is just for puposes as giving an isight why the computer needs so much ram.

• The file is loaded into memory: 500mb.
• The file is parsed, temp variables are allocated during parsing, key value pairs are built up with type descriptions and verified, around 1000-1500mb
• An object is instantiated. Default variables are loaded. however much that takes. lets say 0.01mb.
• Variables are assigned to object, values are cast to actual types. another 500mb in temp variables.
• Variables are in object instance, let's say 400mb.

So a total tally of rougly 3650.01mb.

Then the garbage collector comes by at some point and cleans up the mess and you're left with about 400mb.