#include "config.h"
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include "libgfortran.h"
#if defined (HAVE_GFC_INTEGER_1)
extern void matmul_i1 (gfc_array_i1 * const restrict retarray,
gfc_array_i1 * const restrict a, gfc_array_i1 * const restrict b);
export_proto(matmul_i1);
void
matmul_i1 (gfc_array_i1 * const restrict retarray,
gfc_array_i1 * const restrict a, gfc_array_i1 * const restrict b)
{
const GFC_INTEGER_1 * restrict abase;
const GFC_INTEGER_1 * restrict bbase;
GFC_INTEGER_1 * restrict dest;
index_type rxstride, rystride, axstride, aystride, bxstride, bystride;
index_type x, y, n, count, xcount, ycount;
assert (GFC_DESCRIPTOR_RANK (a) == 2
|| GFC_DESCRIPTOR_RANK (b) == 2);
if (retarray->data == NULL)
{
if (GFC_DESCRIPTOR_RANK (a) == 1)
{
retarray->dim[0].lbound = 0;
retarray->dim[0].ubound = b->dim[1].ubound - b->dim[1].lbound;
retarray->dim[0].stride = 1;
}
else if (GFC_DESCRIPTOR_RANK (b) == 1)
{
retarray->dim[0].lbound = 0;
retarray->dim[0].ubound = a->dim[0].ubound - a->dim[0].lbound;
retarray->dim[0].stride = 1;
}
else
{
retarray->dim[0].lbound = 0;
retarray->dim[0].ubound = a->dim[0].ubound - a->dim[0].lbound;
retarray->dim[0].stride = 1;
retarray->dim[1].lbound = 0;
retarray->dim[1].ubound = b->dim[1].ubound - b->dim[1].lbound;
retarray->dim[1].stride = retarray->dim[0].ubound+1;
}
retarray->data
= internal_malloc_size (sizeof (GFC_INTEGER_1) * size0 ((array_t *) retarray));
retarray->offset = 0;
}
if (GFC_DESCRIPTOR_RANK (retarray) == 1)
{
rxstride = rystride = retarray->dim[0].stride;
}
else
{
rxstride = retarray->dim[0].stride;
rystride = retarray->dim[1].stride;
}
if (GFC_DESCRIPTOR_RANK (a) == 1)
{
axstride = a->dim[0].stride;
aystride = 1;
xcount = 1;
count = a->dim[0].ubound + 1 - a->dim[0].lbound;
}
else
{
axstride = a->dim[0].stride;
aystride = a->dim[1].stride;
count = a->dim[1].ubound + 1 - a->dim[1].lbound;
xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
}
assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
if (GFC_DESCRIPTOR_RANK (b) == 1)
{
bxstride = b->dim[0].stride;
bystride = 0xDEADBEEF;
ycount = 1;
}
else
{
bxstride = b->dim[0].stride;
bystride = b->dim[1].stride;
ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
}
abase = a->data;
bbase = b->data;
dest = retarray->data;
if (rxstride == 1 && axstride == 1 && bxstride == 1)
{
const GFC_INTEGER_1 * restrict bbase_y;
GFC_INTEGER_1 * restrict dest_y;
const GFC_INTEGER_1 * restrict abase_n;
GFC_INTEGER_1 bbase_yn;
if (rystride == xcount)
memset (dest, 0, (sizeof (GFC_INTEGER_1) * xcount * ycount));
else
{
for (y = 0; y < ycount; y++)
for (x = 0; x < xcount; x++)
dest[x + y*rystride] = (GFC_INTEGER_1)0;
}
for (y = 0; y < ycount; y++)
{
bbase_y = bbase + y*bystride;
dest_y = dest + y*rystride;
for (n = 0; n < count; n++)
{
abase_n = abase + n*aystride;
bbase_yn = bbase_y[n];
for (x = 0; x < xcount; x++)
{
dest_y[x] += abase_n[x] * bbase_yn;
}
}
}
}
else if (rxstride == 1 && aystride == 1 && bxstride == 1)
{
if (GFC_DESCRIPTOR_RANK (a) != 1)
{
const GFC_INTEGER_1 *restrict abase_x;
const GFC_INTEGER_1 *restrict bbase_y;
GFC_INTEGER_1 *restrict dest_y;
GFC_INTEGER_1 s;
for (y = 0; y < ycount; y++)
{
bbase_y = &bbase[y*bystride];
dest_y = &dest[y*rystride];
for (x = 0; x < xcount; x++)
{
abase_x = &abase[x*axstride];
s = (GFC_INTEGER_1) 0;
for (n = 0; n < count; n++)
s += abase_x[n] * bbase_y[n];
dest_y[x] = s;
}
}
}
else
{
const GFC_INTEGER_1 *restrict bbase_y;
GFC_INTEGER_1 s;
for (y = 0; y < ycount; y++)
{
bbase_y = &bbase[y*bystride];
s = (GFC_INTEGER_1) 0;
for (n = 0; n < count; n++)
s += abase[n*axstride] * bbase_y[n];
dest[y*rystride] = s;
}
}
}
else if (axstride < aystride)
{
for (y = 0; y < ycount; y++)
for (x = 0; x < xcount; x++)
dest[x*rxstride + y*rystride] = (GFC_INTEGER_1)0;
for (y = 0; y < ycount; y++)
for (n = 0; n < count; n++)
for (x = 0; x < xcount; x++)
dest[x*rxstride + y*rystride] += abase[x*axstride + n*aystride] * bbase[n*bxstride + y*bystride];
}
else if (GFC_DESCRIPTOR_RANK (a) == 1)
{
const GFC_INTEGER_1 *restrict bbase_y;
GFC_INTEGER_1 s;
for (y = 0; y < ycount; y++)
{
bbase_y = &bbase[y*bystride];
s = (GFC_INTEGER_1) 0;
for (n = 0; n < count; n++)
s += abase[n*axstride] * bbase_y[n*bxstride];
dest[y*rxstride] = s;
}
}
else
{
const GFC_INTEGER_1 *restrict abase_x;
const GFC_INTEGER_1 *restrict bbase_y;
GFC_INTEGER_1 *restrict dest_y;
GFC_INTEGER_1 s;
for (y = 0; y < ycount; y++)
{
bbase_y = &bbase[y*bystride];
dest_y = &dest[y*rystride];
for (x = 0; x < xcount; x++)
{
abase_x = &abase[x*axstride];
s = (GFC_INTEGER_1) 0;
for (n = 0; n < count; n++)
s += abase_x[n*aystride] * bbase_y[n*bxstride];
dest_y[x*rxstride] = s;
}
}
}
}
#endif