Software at OSC
Software by Field
• Biosciences
• Chemistry
• Structural Mechanics
• Fluid Dynamics
• Programming
• Visualization
Software by System
• BALE
• Glenn
Related Links
• Supercomputing
• Support
• Get a New OSC Account
• Available Hardware
• Training
• Accounts
• Statewide Software
• Manuals
• Consult
• Notices
|
ACML - AMD Core Math Library
Introduction
The AMD Core Math Library (ACML) is a set of numerical routines tuned specifically for AMD64 platform processors (including OpteronTM and AthlonTM64 ). The routines, which are available via both FORTRAN 77 and C interfaces, include BLAS, LAPACK, FFT and RND.
Version
The versions currently available at OSC are:
| Version |
Command to load software |
| 3.6.0 |
module switch pgi pgi-3.6.0 |
| 4.0.1 (default) |
(automatically loaded at login) |
| 4.3.0 |
module switch pgi pgi-4.3.0 |
Availability
ACML is available on the Glenn Cluster and the BALE Cluster.
Usage
To use the ACML libraries type the commands below, based on the compiler you are using:
| Compiler |
Module Command |
| Portland Group Compilers |
module load acml-pgi |
| Intel Compilers |
module load acml-intel |
| GNU Compilers |
module load acml-gnu |
| |
module load acml-gfortran |
Use module show acml-<compiler> to view the environment varialbles set by the acml module.
Examples
In the C programming style, here is an example of the dot product of two complex vectors:
/* cdotu Example Program Text */
/*
* ACML version 1.0 Copyright AMD,NAG 2003
*/
#include <acml.h>
#include <stdio.h>
int main(void)
{
complex r;
complex x[3], y[3];
int n = 3;
int i;
printf("ACML example: dot product of two complex vectors using cdotu\n");
printf("------------------------------------------------------------\n");
printf("\n");
x[0] = compose_complex(1.0, 2.0);
x[1] = compose_complex(2.0, 1.0);
x[2] = compose_complex(1.0, 3.0);
y[0] = compose_complex(3.0, 1.0);
y[1] = compose_complex(1.0, 4.0);
y[2] = compose_complex(1.0, 2.0);
printf("Vector x: ");
for (i = 0; i < n; i++)
printf(" (%7.4f,%7.4f)\n", x[i].real, x[i].imag);
printf("Vector y: ");
for (i = 0; i < n; i++)
printf(" (%7.4f,%7.4f)\n", y[i].real, y[i].imag);
r = cdotu(n, x, 1, y, 1);
printf("r = x.y = (%12.3f,%12.3f)\n", r.real, r.imag);
return 0;
} |
Using the Portland Group compiler, pgcc, the command sequence would be:
>> module load acml-pgi
>> pgcc $ACML_CFLAGS $ACML -lm -lpgftnrtl –lrt cdotu_c_example.c
>> ./a.out
ACML example: dot product of two complex vectors using cdotu
------------------------------------------------------------
Vector x: ( 1.0000, 2.0000)
(2.0000, 1.0000)
(1.0000, 3.0000)
Vector y: ( 3.0000, 1.0000)
(1.0000, 4.0000)
(1.0000, 2.0000)
r = x.y = ( -6.000, 21.000)
>>
An example of initializing and generating random numbers, in the Fortran Programming Language:
C Simple DRANDINITIALIZE and DRANDUNIFORM example program
C .. Parameters ..
INTEGER MSTATE, MSEED, MN
PARAMETER (MSTATE=20,MSEED=10,MN=12)
C .. Local Scalars ..
DOUBLE PRECISION A, B
INTEGER GENID, I, INFO, LSEED, LSTATE, N, SUBID
C .. Local Arrays ..
DOUBLE PRECISION X(MN)
INTEGER SEED(MSEED), STATE(MSTATE)
C .. External Subroutines ..
EXTERNAL DRANDINITIALIZE, DRANDUNIFORM
C .. Executable Statements ..
CONTINUE
WRITE (*,FMT=*)
WRITE (*,FMT=99999)
* 'ACML example: DRANDINITIALIZE and DRANDUNIFORM'
WRITE (*,FMT=99999)
* '----------------------------------------------'
WRITE (*,FMT=*)
C Initialize the number of variates required
N = MN
C Use the basic generator as the base generator
GENID = 1
SUBID = 1
C Populate the SEED, basic generator needs one seed, and a
C STATE array of length 16
LSTATE = 16
LSEED = 1
SEED(1) = 122421
C Initialize the base generator
CALL DRANDINITIALIZE(GENID,SUBID,SEED,LSEED,STATE,LSTATE,INFO)
C Generate a sequence from a uniform U(0,1) distribution
A = 0.0D0
B = 1.0D0
CALL DRANDUNIFORM(N,A,B,STATE,X,INFO)
C Print the sequence
WRITE (*,FMT=99999) 'Numbers from a uniform U(0,1) distribution:'
WRITE (*,FMT=99998) X
C
99999 FORMAT (A)
99998 FORMAT (4F10.4)
END |
Using the Portland Group compiler, pgf77, the command sequence would be:
>> module load acml-pgi
>> pgf77 dranduniform_example.f $ACML
>> ./a.out
ACML example: DRANDINITIALIZE and DRANDUNIFORM
----------------------------------------------
Numbers from a uniform U(0,1) distribution:
0.6416 0.5516 0.2852 0.3752
0.2618 0.5148 0.5048 0.6239
0.6155 0.5263 0.4899 0.2464
>>
Documentation
Documentation is available at: http://developer.amd.com/assets/acml_userguide.pdf
|
