#!/usr/bin/perl
#
print <<EOF
/* This file is autogenerated. Do not edit. */
/*
* LIBOIL - Library of Optimized Inner Loops
* Copyright (c) 2005 David A. Schleef <ds@schleef.org>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include <math.h>
#include <liboil/liboil.h>
#include <liboil/liboilclasses.h>
EOF
;
sub binary_pointer
{
my $kernel = shift;
my $precision = shift;
my $type = "oil_type_$precision";
my $operator = shift;
print <<EOF
static void
${kernel}_${precision}_pointer (${type} *dest, ${type} *src1, ${type} *src2, int n)
{
while (n) {
*dest = *src1 ${operator} *src2;
dest++;
src1++;
src2++;
n--;
}
}
OIL_DEFINE_IMPL (${kernel}_${precision}_pointer, ${kernel}_${precision});
EOF
;
}
sub binary_unroll2
{
my $kernel = shift;
my $precision = shift;
my $type = "oil_type_$precision";
my $operator = shift;
print <<EOF
static void
${kernel}_${precision}_unroll2 (${type} *dest, ${type} *src1, ${type} *src2, int n)
{
int i;
if (n & 1) {
dest[0] = src1[0] ${operator} src2[0];
dest++;
src1++;
src2++;
n--;
}
for(i=0;i<n;i+=2){
dest[i] = src1[i] ${operator} src2[i];
dest[i+1] = src1[i+1] ${operator} src2[i+1];
}
}
OIL_DEFINE_IMPL (${kernel}_${precision}_unroll2, ${kernel}_${precision});
EOF
;
}
sub binary_unroll4a
{
my $kernel = shift;
my $precision = shift;
my $type = "oil_type_$precision";
my $operator = shift;
print <<EOF
static void
${kernel}_${precision}_unroll4a (${type} *dest, ${type} *src1, ${type} *src2, int n)
{
int i;
while (n & 3) {
dest[0] = src1[0] ${operator} src2[0];
dest++;
src1++;
src2++;
n--;
}
for(i=0;i<n;i+=4){
dest[i] = src1[i] ${operator} src2[i];
dest[i+1] = src1[i+1] ${operator} src2[i+1];
dest[i+2] = src1[i+2] ${operator} src2[i+2];
dest[i+3] = src1[i+3] ${operator} src2[i+3];
}
}
OIL_DEFINE_IMPL (${kernel}_${precision}_unroll4a, ${kernel}_${precision});
EOF
;
}
sub binary_unroll4b
{
my $kernel = shift;
my $precision = shift;
my $type = "oil_type_$precision";
my $operator = shift;
print <<EOF
static void
${kernel}_${precision}_unroll4b (${type} *dest, ${type} *src1, ${type} *src2, int n)
{
int i;
for(i=0;i<(n&(~0x3));i+=4){
dest[i+0] = src1[i+0] ${operator} src2[i+0];
dest[i+1] = src1[i+1] ${operator} src2[i+1];
dest[i+2] = src1[i+2] ${operator} src2[i+2];
dest[i+3] = src1[i+3] ${operator} src2[i+3];
}
for(;i<n;i++){
dest[i] = src1[i] ${operator} src2[i];
}
}
OIL_DEFINE_IMPL (${kernel}_${precision}_unroll4b, ${kernel}_${precision});
EOF
;
}
sub binary_unroll4c
{
my $kernel = shift;
my $precision = shift;
my $type = "oil_type_$precision";
my $operator = shift;
print <<EOF
static void
${kernel}_${precision}_unroll4c (${type} *dest, ${type} *src1, ${type} *src2, int n)
{
int i;
for(i=0;i<(n&(~0x3));i+=4){
*dest++ = *src1++ ${operator} *src2++;
*dest++ = *src1++ ${operator} *src2++;
*dest++ = *src1++ ${operator} *src2++;
*dest++ = *src1++ ${operator} *src2++;
}
for(;i<n;i++){
*dest++ = *src1++ ${operator} *src2++;
}
}
OIL_DEFINE_IMPL (${kernel}_${precision}_unroll4c, ${kernel}_${precision});
EOF
;
}
my %binary_operators = (
"add" => "+",
"subtract" => "-",
"multiply" => "*",
"divide" => "/"
);
my @types = ( "f32", "f64" );
while ( ($name, $op) = each %binary_operators ) {
foreach $prec (@types) {
binary_pointer($name, $prec, $op);
binary_unroll2($name, $prec, $op);
binary_unroll4a($name, $prec, $op);
binary_unroll4b($name, $prec, $op);
binary_unroll4c($name, $prec, $op);
}
}
exit 0;
binary_pointer("subtract", "f32", "-");
binary_unroll2("subtract", "f32", "-");
binary_unroll4a("subtract", "f32", "-");
binary_unroll4b("subtract", "f32", "-");
binary_unroll4c("subtract", "f32", "-");
binary_pointer("add", "f32", "+");
binary_unroll2("add", "f32", "+");
binary_unroll4a("add", "f32", "+");
binary_unroll4b("add", "f32", "+");
binary_unroll4c("add", "f32", "+");
binary_pointer("multiply", "f32", "*");
binary_unroll2("multiply", "f32", "*");
binary_unroll4a("multiply", "f32", "*");
binary_unroll4b("multiply", "f32", "*");
binary_unroll4c("multiply", "f32", "*");
binary_pointer("divide", "f32", "/");
binary_unroll2("divide", "f32", "/");
binary_unroll4a("divide", "f32", "/");
binary_unroll4b("divide", "f32", "/");
binary_unroll4c("divide", "f32", "/");
binary_pointer("subtract", "f64", "-");
binary_unroll2("subtract", "f64", "-");
binary_unroll4a("subtract", "f64", "-");
binary_unroll4b("subtract", "f64", "-");
binary_unroll4c("subtract", "f64", "-");
binary_pointer("add", "f64", "+");
binary_unroll2("add", "f64", "+");
binary_unroll4a("add", "f64", "+");
binary_unroll4b("add", "f64", "+");
binary_unroll4c("add", "f64", "+");
binary_pointer("multiply", "f64", "*");
binary_unroll2("multiply", "f64", "*");
binary_unroll4a("multiply", "f64", "*");
binary_unroll4b("multiply", "f64", "*");
binary_unroll4c("multiply", "f64", "*");
binary_pointer("divide", "f64", "/");
binary_unroll2("divide", "f64", "/");
binary_unroll4a("divide", "f64", "/");
binary_unroll4b("divide", "f64", "/");
binary_unroll4c("divide", "f64", "/");
$blah = "
static void
subtract_f32_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = src1[i] - src2[i];
}
}
OIL_DEFINE_IMPL (subtract_f32_ref, subtract_f32);
static void
multiply_f32_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = src1[i] * src2[i];
}
}
OIL_DEFINE_IMPL (multiply_f32_ref, multiply_f32);
static void
divide_f32_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = src1[i] / src2[i];
}
}
OIL_DEFINE_IMPL_REF (divide_f32_ref, divide_f32);
static void
minimum_f32_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = (src1[i] < src2[i]) ? src1[i] : src2[i];
}
}
OIL_DEFINE_IMPL_REF (minimum_f32_ref, minimum_f32);
static void
maximum_f32_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = (src1[i] > src2[i]) ? src1[i] : src2[i];
}
}
OIL_DEFINE_IMPL_REF (maximum_f32_ref, maximum_f32);
static void
negative_f32_ref (float *dest, float *src1, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = -src1[i];
}
}
OIL_DEFINE_IMPL_REF (negative_f32_ref, negative_f32);
static void
inverse_f32_ref (float *dest, float *src1, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = 1.0/src1[i];
}
}
OIL_DEFINE_IMPL_REF (inverse_f32_ref, inverse_f32);
static void
sign_f32_ref (float *dest, float *src1, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = (src1[i] < 0) ? -src1[i] : src1[i];
}
}
OIL_DEFINE_IMPL_REF (sign_f32_ref, sign_f32);
static void
floor_f32_ref (float *dest, float *src1, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = floor(src1[i]);
}
}
OIL_DEFINE_IMPL_REF (floor_f32_ref, floor_f32);
static void
scalaradd_f32_ns_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = src1[i] + src2[0];
}
}
OIL_DEFINE_IMPL_REF (scalaradd_f32_ns_ref, scalaradd_f32_ns);
static void
scalarmultiply_f32_ns_ref (float *dest, float *src1, float *src2, int n)
{
int i;
for(i=0;i<n;i++){
dest[i] = src1[i] * src2[0];
}
}
OIL_DEFINE_IMPL_REF (scalarmultiply_f32_ns_ref, scalarmultiply_f32_ns);
";