starpu-max/doc/vector_scal_cpu.texi

@c StarPU --- Runtime system for heterogeneous multicore architectures.
@c
@c Copyright (C) 2009-2011  Université de Bordeaux 1
@c Copyright (C) 2010, 2011  Centre National de la Recherche Scientifique
@c
@c StarPU is free software; you can redistribute it and/or modify
@c it under the terms of the GNU Lesser General Public License as published by
@c the Free Software Foundation; either version 2.1 of the License, or (at
@c your option) any later version.
@c
@c StarPU is distributed in the hope that it will be useful, but
@c WITHOUT ANY WARRANTY; without even the implied warranty of
@c MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
@c
@c See the GNU Lesser General Public License in COPYING.LGPL for more details.

#include <starpu.h>
#include <xmmintrin.h>

/* This kernel takes a buffer and scales it by a constant factor */
void scal_cpu_func(void *buffers[], void *cl_arg)
@{
    unsigned i;
    float *factor = cl_arg;

    /* 
     * The "buffers" array matches the task->buffers array: for instance
     * task->buffers[0].handle is a handle that corresponds to a data with
     * vector "interface", so that the first entry of the array in the
     * codelet  is a pointer to a structure describing such a vector (ie.
     * struct starpu_vector_interface_s *). Here, we therefore manipulate
     * the buffers[0] element as a vector: nx gives the number of elements
     * in the array, ptr gives the location of the array (that was possibly
     * migrated/replicated), and elemsize gives the size of each elements.
     */
    starpu_vector_interface_t *vector = buffers[0];

    /* length of the vector */
    unsigned n = STARPU_VECTOR_GET_NX(vector);

    /* get a pointer to the local copy of the vector : note that we have to
     * cast it in (float *) since a vector could contain any type of
     * elements so that the .ptr field is actually a uintptr_t */
    float *val = (float *)STARPU_VECTOR_GET_PTR(vector);

    /* scale the vector */
    for (i = 0; i < n; i++)
        val[i] *= *factor;
@}

void scal_sse_func(void *buffers[], void *cl_arg)
@{
	float *vector = (float *) STARPU_VECTOR_GET_PTR(buffers[0]);
	unsigned int n = STARPU_VECTOR_GET_NX(buffers[0]);
	unsigned int n_iterations = n/4;
	if (n % 4 != 0)
		n_iterations++;

	__m128 *VECTOR = (__m128*) vector;
	__m128 factor __attribute__((aligned(16)));
	factor = _mm_set1_ps(*(float *) cl_arg);

	unsigned int i;	
	for (i = 0; i < n_iterations; i++)
		VECTOR[i] = _mm_mul_ps(factor, VECTOR[i]);
@}