GtkSort

GtkSort is a multiprocessor external disk sorter and data manipulator for systems that support GTK+. Its source code has been ported and tested to produce valid results in Linux/x86, Linux/amd64, Linux/alpha, HP-UX/hp-pa 11.11i, Tru64 5.1B, Solaris 10/x86 and Win32 NT Class operating systems.

GtkSort processes files using multi threading in order to implement parallel algorithms. It overlaps disk I/O with sorting and reduces I/O waits. By using only sequential files it utilizes the most of the disk's I/O bandwidth. By giving exclusive read or write permissions for each processed file to only one thread, it minimizes the conflicting I/O requests that reduce the hard disk's efficiency. In order to sort records, and depending on the data type of the sort keys, GtkSort uses the standard library quick sort or its own implementation of Most Significant Digit (MSD) radix sort. GtkSort uses cache efficient algorithms keeping the cache misses at a very low rate (0.4% in version 0.2.0).

GtkSort integrates a Graphical User Interface (GUI) based on GTK+ in order to increase its friendliness against the end user. It also integrates a Command Line Interface (CLI) so it can be used in shell scripts.

GtkSort is free and open source software distributed under the terms of the
GNU Public License version 2.

What GtkSort Can do:

* Sort text or binary files.

* Sort large data sets much faster than the standard sort utility of the operating system.

* Sort by binary data or text data keys.

* Sort on a limited number of twelve keys.

* Support Fixed Length Record (FLR) text or binary files with fixed size keys.

* Support Variable Length Record (VLR) text files (lines of text) with delimited or fixed size keys.

* Use multiple processors and disks in parallel.

* Perform ascending or descending sort of each key.

* Preview the keys of the unsorted input and the sorted output.

* Use memory dynamically according to the limitations set by the user.

* Exploit CPU and file system specific characteristics such as the L2 cache size, the D cache size and the disk I/O block size.

The Win32 installation package needs GTK+ for Win32 runtime and the Visual C++ redistributable package.

11-Feb-2008: New release
Version 0.3.0 has been released. This version includes:

* gtksortgen utility introduced. gtksortgen is used in the test suite instead of Microsoft's SortGen.
* A 64 bit portability issue solved
* Minor cosmetic changes

08-Aug-2007: New release
Version 0.2.3 has been released. This version includes:

* A serious bug fix for win32. GtkSort was crashing when the input file had size greater than two gigabytes.
* Minor performance improvements for win32.
* Minor cosmetic fixes for win32.

28-Jun-2007: New release
Version 0.2.2 has been released. This version includes:

* A test suite introduced. Serious bugs found and fixed. The test suite will
be extended in the future.
* Now the temporary directory can be defined by the user through the
properties.
* New key data types added for 32 and 64 bit floating point.

2-Mar-2007: New web page and documentation release

* First release of the documentation along with the new web page.

25-Feb-2007: New release
Version 0.2.1 has been released.

* GtkSort could not sort correctly continuous text keys of the same order. This release fixes the bug.

13-Feb-2007: New release
Version 0.2.0 has been released. This version includes:

* New key data types added for 16, 32 and 64 bit integers.
* Command Line Interface implemented. Now you can use GtkSort in your scripts. Type gtksort -h for more information.
* Minor bug fixes and GUI changes.

17-Jan-2007: GtkSort released
Version 0.1.0 has been released.

* This is the initial release of the source code.