A generic and efficient SIMD implementation of MSB Radix Sort with separate key and payload datastreams that supports arbitrary key and payload data types written in C++ accompanied by a bachelor's thesis.
This library is the result of the bachelor's thesis "SIMD Implementation of MSB Radix Sort with Separate Key and Payload Datastreams" at the Bielefeld University.
Requires at least C++20.
The file radixSort.hpp
provides a single header version of the radix sort library that has been stripped of classes and functions that were only needed for the benchmarking and testing.
For compilation with gcc, compile with the following flags:
-mavx512f -mavx512bw -mavx512dq -mavx512vl -mavx512vbmi -mavx512vbmi2
To use the single header version of the library, simply include the file radixSort.hpp
in your project.
To sort a dataset, call the sort function like the following:
simd_sort::radix_sort::sort(num, keyArray, payloadArrays...);
where num
is the number of elements in the dataset, keyArray
is a pointer
to the key array, and payloadArrays
are the pointers to the payload arrays.
Or, if key and payload datastreams are combined into a single array, call the sort function like the following:
simd_sort::radix_sort::sort(num, (simd_sort::DataElement<K, Ps...> *)combinedArray);
where num
is the number of elements in the dataset, combinedArray
is a
pointer to the combined array and K
and Ps...
are the types of the key and
the payloads.
By default the sort function sorts in ascending order. To sort in descending
order, call the sort function with the template parameter Up
set to false:
simd_sort::radix_sort::sort<false>(num, keyArray, payloadArrays...);
or
simd_sort::radix_sort::sort<false>(num, (simd_sort::DataElement<K, Ps...> *) combinedArray);
The file test.cpp
provides a simple test program for the library. It can be compiled with:
$ make test
and run with:
$ ./build/test
The file perf-thesis.cpp
provides the program that was used to generate the data for the experiments of the bachelor's thesis.
It can be compiled with:
$ make perf-thesis
and run with:
$ ./build/perf-thesis
Note that to be able to run all tests of the perf-thesis.cpp
program, the following libraries need to be installed:
- Intel IPP library, the environment variable
IPPROOT
needs to be set. - The generic SIMD implementation of MSB Radix Sort by Prof. Dr.-Ing. Ralf Möller (the file
SIMDRadixSortGeneric.H
) included in the T-SIMD library must be present in themoeller/
directory.
If either of these libraries is not present, the perf-thesis.cpp
program will still be able to be compiled and run, but the tests that require them will not be executed.
There have been works to publish the findings of the bachelor's thesis in a research paper. As part of this, the radix sort implementation was modified to turn it into a quicksort implementation.
However, the quicksort implementation turned out to have a very similar performance to the radix sort implementation. Additionally, there are already efficient and sufficiently generic SIMD sorting algorithms (for example x86-simd-sort by Intel) so it was decided that there weren't enough new findings to justify a research paper.
The results of this work can however still be found in the file quick_sort.hpp
and the corresponding benchmarks in the file perf-paper.cpp
.