https://github.com/lanl/perfecthash

A perfect hash code for CPUs and GPUs using OpenCL
https://github.com/lanl/perfecthash

Last synced: 8 months ago
JSON representation

A perfect hash code for CPUs and GPUs using OpenCL

• Host: GitHub
• URL: https://github.com/lanl/perfecthash
• Owner: lanl
• License: other
• Created: 2014-02-15T15:29:15.000Z (over 12 years ago)
• Default Branch: master
• Last Pushed: 2020-10-24T14:54:44.000Z (over 5 years ago)
• Last Synced: 2025-04-19T16:06:24.437Z (about 1 year ago)
• Language: C
• Size: 743 KB
• Stars: 16
• Watchers: 8
• Forks: 6
• Open Issues: 1
• Metadata Files:
  • Readme: README
  • License: LICENSE

Awesome Lists containing this project

README

          
This code is a set of hash functions to support the paper "Hash-based Algorithms

for Discretized Data" to be published in the SIAM Journal of Scientific 

Computing". The publication details are below and in the Papers.bib file in bibtex

format. 

   Robey,R.N., Nicholaeff,D., and Robey,R.W. "Hash-Based Algorithms for Discretized Data", 

   SIAM Journal of Scientific Computing, July 2013, Volume 35, Number 4, C346--C368

   The pre-publication version has the LANL report number LA-UR-12-01566.

This code has been released under an open-source Apache 2 license to

encourage further development of hashing methods. See the LICENSE file for more

information about the license and the use of this code.

Through a web search we hope to gather statistics on the use of the method

and its improvements and help to encourage more open technology transfer by LANL

and other government research organizations. This can be thought of as analagous

to journal article citations, but within software products. If code is not

distributed with a software product, a reference should be provided in a text

file so that attribution can be determined.

Under this license, it is required to include a reference to this work. We

request that each derivative work contain a reference to LANL Copyright 

Disclosure C13002/LA-CC-12-022 so that this work’s impact can be roughly

measured. In addition, it is requested that a modifier is included as in

the following example:

// LANL Copyright Disclosure C13002/LA-CC-12-022

This is LANL Copyright Disclosure C13002/LA-CC-12-022

Authors: Bob Robey       XCP-2   brobey@lanl.gov

         David Nicholaeff        dnic@lanl.gov, mtrxknight@aol.com

         Rachel Robey            rnrobey@gmail.com

This code uses cmake for builds. To build the code:

cmake .

make

There will be several executables built -- sort, sort2d, neigh, neigh2d, remap, remap2d, table

Each executable runs all of the methods for the mesh operation. There is a controlling

loop at the top of the main routine that users may want to modify for the algorithms

to be run. Also, there is a random number seed for varying the problem setups that is

normally off, but users may want to turn on for some investigations. Results do vary for

different problems, but the pattern does not change significantly. Also, the OpenCL library

looks for a GPU to use for the OpenCL code. This may need to be modified for your particular

hardware.

Output should look something like the following:

    Sorting Performance Results

Size,   	Qsort,    	Heapsort, 	Mergesort, 	Hash CPU, 	Hash GPU

Max diff is 1 times min_diff

1024,     	0.000105,	0.000168,	0.000123,	0.000010,	0.000294,

2048,     	0.000223,	0.000353,	0.000262,	0.000020,	0.000330,

4096,     	0.000493,	0.000781,	0.000573,	0.000040,	0.000400,

8192,     	0.001036,	0.001683,	0.001135,	0.000079,	0.000610,

16384,     	0.002242,	0.003470,	0.002613,	0.000161,	0.000934,

32768,     	0.005360,	0.007348,	0.005469,	0.000415,	0.001535,

65536,     	0.011800,	0.015796,	0.012560,	0.000800,	0.002629,

131072,     	0.020707,	0.036958,	0.024513,	0.001561,	0.005294,

262144,     	0.042710,	0.075466,	0.052864,	0.003751,	0.011301,

524288,     	0.089662,	0.167398,	0.113505,	0.011855,	0.024291,

1048576,     	0.185135,	0.407333,	0.249813,	0.026447,	0.018890,

2097152,     	0.385789,	1.018970,	0.498638,	0.063970,	0.114473,

4194304,     	0.818016,	2.475966,	1.052133,	0.141997,	0.246392,

8388608,     	1.696586,	5.843146,	2.184568,	0.307373,	0.503904,

Max diff is 2 times min_diff

1024,     	0.000102,	0.000168,	0.000121,	0.000015,	0.000303,

2048,     	0.000224,	0.000332,	0.000240,	0.000026,	0.000339,

   ...

   ...

   ...