Databases
StackOverflow2013
Postsdbo
PO
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PO
    primarykey
    data
    text
    <p>There are only 4 billion possible <code>int</code> values so the most efficient way of doing this, is to count the number of occurrences of any value. You can use a memory MappedByteBuffer so you don't have to have 16 GB of memory. Once you have counted all the occurrences the counts will naturally be in order, so no further sorting is required. The time complexity is O(n) instead of O(n * log n) line merge sort or quick sort.</p> <hr> <pre><code>import sun.nio.ch.DirectBuffer; import java.io.File; import java.io.IOException; import java.io.RandomAccessFile; import java.nio.MappedByteBuffer; import java.nio.channels.FileChannel; public class Sort10Billion { public static void main(String... args) throws IOException { Runtime runtime = Runtime.getRuntime(); long used1 = runtime.totalMemory() - runtime.freeMemory(); MassiveCounterStore mcs = new MassiveCounterStore(); long start = System.nanoTime(); long count = 10 * 1000 * 1000 * 1000L; for (long i = count; i &gt; 0; i--) mcs.incrementIndex((int) (i / 1019)); mcs.iterator(new NumberCountFunction() { @Override public void counted(int n, long count) { // System.out.println(n + ": " + count); } }); long time = System.nanoTime() - start; long used2 = runtime.totalMemory() - runtime.freeMemory(); System.out.printf("Took %.1f seconds to sort %,d numbers, using %.3f MB%n", time / 1e9, count, (used2-used1)/1e6); mcs.close(); } } interface NumberCountFunction { public void counted(int n, long count); } class MassiveCounterStore { public static final int PARTITION_BITS = 26; static final int PARTITIONS = (1 &lt;&lt; (34 - PARTITION_BITS)); // 32-bit * 4 bytes. final MappedByteBuffer[] buffers = new MappedByteBuffer[PARTITIONS]; final FileChannel channel; int smallest = PARTITIONS; int largest = 0; public MassiveCounterStore() throws IOException { File tmpStore = File.createTempFile("counter", "dat"); tmpStore.deleteOnExit(); channel = new RandomAccessFile(tmpStore, "rw").getChannel(); for (int i = 0; i &lt; PARTITIONS; i++) buffers[i] = channel.map(FileChannel.MapMode.READ_WRITE, (long) i &lt;&lt; PARTITION_BITS, 1 &lt;&lt; PARTITION_BITS); } public void incrementIndex(int n) { long l = (n + Integer.MIN_VALUE) &amp; 0xFFFFFFFFL; int partition = (int) (l &gt;&gt; (PARTITION_BITS - 2)); // 4 bytes each. int index = (int) ((l &lt;&lt; 2) &amp; ((1 &lt;&lt; PARTITION_BITS) - 1)); MappedByteBuffer buffer = buffers[partition]; int count = buffer.getInt(index); buffer.putInt(index, count + 1); if (smallest &gt; partition) smallest = partition; if (largest &lt; partition) largest = partition; } public void iterator(NumberCountFunction nfc) { int n = (smallest &lt;&lt; (PARTITION_BITS -2)) + Integer.MIN_VALUE; for (int p = smallest; p &lt;= largest; p++) { MappedByteBuffer buffer = buffers[p]; for (int i = 0; i &lt; 1 &lt;&lt; PARTITION_BITS; i += 4) { int count = buffer.getInt(i); if (count != 0) nfc.counted(n, count &amp; 0xFFFFFFFFL); n++; } } assert n == Integer.MIN_VALUE; } public void close() { try { channel.close(); } catch (IOException ignored) { } for (MappedByteBuffer buffer : buffers) { ((DirectBuffer) buffer).cleaner().clean(); } } } </code></pre> <p>prints when run with -XX:-UseTLAB (which gives you more accurate memory usage)</p> <pre><code>Took 150.7 seconds to sort 10,000,000,000 numbers, using 0.202 MB </code></pre> <p>I think using 202 KB is pretty good. ;)</p> <p>Note: your performance is heavily dependant on the distribution of values as this impacts the efficiency of the cache.</p>
    singulars
    1. This table or related slice is empty.
    1. POWrite a Java class to sort 10 billion integers
      singulars
      1. PTQuestion
    1. PTAnswer
    1. USPeter Lawrey
    1. USPeter Lawrey
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. POWrite a Java class to sort 10 billion integers
      singulars
      1. PTQuestion
    1. This table or related slice is empty.
    1. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    3. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTAcceptedByOriginator
    1. COoffbeat but cool.
      singulars
      1. PO
      1. USStanleyZ
    2. COi have sorted the millions of integer records into 10MB sorted files. now i need to merge them and make a big sorted file. But problem is that i am using scanner to read files, now i can't read the files parallel and also like we read an array. i keep the index and so on. So due to this problem i am stuck, any idea how can i do this ???
      singulars
      1. PO
      1. USuser1430003
    3. COIf you don't want to use Scanner, use a binary format. You can read the files in different threads, but if you use a binary format the CPU cost will be much lower and you won't have a problem. Not sure what the rest of your problems are... Have you considered doing what I suggested? Are your numbers 32-bit or 64-bit etc?
      singulars
      1. PO
      1. USPeter Lawrey
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload