472 lines
15 KiB
C
472 lines
15 KiB
C
/*******************************************************************************
|
|
Copyright (c) 2016, The OpenBLAS Project
|
|
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.
|
|
3. Neither the name of the OpenBLAS project nor the names of
|
|
its contributors may be used to endorse or promote products
|
|
derived from this software without specific prior written permission.
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "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 OPENBLAS PROJECT OR CONTRIBUTORS 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.
|
|
*******************************************************************************/
|
|
|
|
#include "common.h"
|
|
#include "macros_msa.h"
|
|
|
|
#if !defined(CONJ)
|
|
#define OP0 +=
|
|
#define OP1 -=
|
|
#define OP2 +=
|
|
#else
|
|
#define OP0 -=
|
|
#define OP1 +=
|
|
#define OP2 -=
|
|
#endif
|
|
|
|
int CNAME(BLASLONG n, BLASLONG dummy0, BLASLONG dummy1, FLOAT da_r, FLOAT da_i,
|
|
FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y, FLOAT *dummy,
|
|
BLASLONG dummy2)
|
|
{
|
|
BLASLONG i, inc_x2, inc_y2;
|
|
FLOAT *py;
|
|
v4f32 x0, x1, x2, x3, x4, x5, x6, x7;
|
|
v4f32 y0, y1, y2, y3, y4, y5, y6, y7, dar_vec, dai_vec;
|
|
v4f32 x0r, x1r, x2r, x3r, x0i, x1i, x2i, x3i;
|
|
v4f32 y0r, y1r, y2r, y3r, y0i, y1i, y2i, y3i;
|
|
FLOAT xd0, xd1, xd2, xd3, xd4, xd5, xd6, xd7;
|
|
FLOAT yd0, yd1, yd2, yd3, yd4, yd5, yd6, yd7;
|
|
|
|
if (n < 0) return(0);
|
|
if ((da_r == 0.0) && (da_i == 0.0)) return(0);
|
|
|
|
py = y;
|
|
|
|
if ((1 == inc_x) && (1 == inc_y))
|
|
{
|
|
FLOAT *x_pref, *y_pref;
|
|
BLASLONG pref_offset;
|
|
|
|
pref_offset = (BLASLONG)x & (L1_DATA_LINESIZE - 1);
|
|
if (pref_offset > 0)
|
|
{
|
|
pref_offset = L1_DATA_LINESIZE - pref_offset;
|
|
pref_offset = pref_offset / sizeof(FLOAT);
|
|
}
|
|
x_pref = x + pref_offset + 64;
|
|
|
|
pref_offset = (BLASLONG)y & (L1_DATA_LINESIZE - 1);
|
|
if (pref_offset > 0)
|
|
{
|
|
pref_offset = L1_DATA_LINESIZE - pref_offset;
|
|
pref_offset = pref_offset / sizeof(FLOAT);
|
|
}
|
|
y_pref = y + pref_offset + 64;
|
|
|
|
dar_vec = COPY_FLOAT_TO_VECTOR(da_r);
|
|
dai_vec = COPY_FLOAT_TO_VECTOR(da_i);
|
|
|
|
for (i = (n >> 4); i--;)
|
|
{
|
|
PREF_OFFSET(x_pref, 0);
|
|
PREF_OFFSET(x_pref, 32);
|
|
PREF_OFFSET(x_pref, 64);
|
|
PREF_OFFSET(x_pref, 96);
|
|
PREF_OFFSET(y_pref, 0);
|
|
PREF_OFFSET(y_pref, 32);
|
|
PREF_OFFSET(y_pref, 64);
|
|
PREF_OFFSET(y_pref, 96);
|
|
x_pref += 32;
|
|
y_pref += 32;
|
|
|
|
LD_SP8_INC(x, 4, x0, x1, x2, x3, x4, x5, x6, x7);
|
|
LD_SP8_INC(py, 4, y0, y1, y2, y3, y4, y5, y6, y7);
|
|
PCKEVOD_W2_SP(x1, x0, x0r, x0i);
|
|
PCKEVOD_W2_SP(y1, y0, y0r, y0i);
|
|
PCKEVOD_W2_SP(x3, x2, x1r, x1i);
|
|
PCKEVOD_W2_SP(y3, y2, y1r, y1i);
|
|
PCKEVOD_W2_SP(x5, x4, x2r, x2i);
|
|
PCKEVOD_W2_SP(y5, y4, y2r, y2i);
|
|
PCKEVOD_W2_SP(x7, x6, x3r, x3i);
|
|
PCKEVOD_W2_SP(y7, y6, y3r, y3i);
|
|
|
|
FMADD4(x0r, x1r, x2r, x3r, dar_vec, y0r, y1r, y2r, y3r);
|
|
y0i OP0 dar_vec * x0i;
|
|
y1i OP0 dar_vec * x1i;
|
|
y2i OP0 dar_vec * x2i;
|
|
y3i OP0 dar_vec * x3i;
|
|
y0r OP1 dai_vec * x0i;
|
|
y1r OP1 dai_vec * x1i;
|
|
y2r OP1 dai_vec * x2i;
|
|
y3r OP1 dai_vec * x3i;
|
|
y0i OP2 dai_vec * x0r;
|
|
y1i OP2 dai_vec * x1r;
|
|
y2i OP2 dai_vec * x2r;
|
|
y3i OP2 dai_vec * x3r;
|
|
|
|
ILVRL_W2_SP(y0i, y0r, y0, y1);
|
|
ILVRL_W2_SP(y1i, y1r, y2, y3);
|
|
ILVRL_W2_SP(y2i, y2r, y4, y5);
|
|
ILVRL_W2_SP(y3i, y3r, y6, y7);
|
|
ST_SP8_INC(y0, y1, y2, y3, y4, y5, y6, y7, y, 4);
|
|
}
|
|
|
|
if (n & 15)
|
|
{
|
|
if (n & 8)
|
|
{
|
|
LD_SP4_INC(x, 4, x0, x1, x2, x3);
|
|
LD_SP4_INC(py, 4, y0, y1, y2, y3);
|
|
PCKEVOD_W2_SP(x1, x0, x0r, x0i);
|
|
PCKEVOD_W2_SP(y1, y0, y0r, y0i);
|
|
PCKEVOD_W2_SP(x3, x2, x1r, x1i);
|
|
PCKEVOD_W2_SP(y3, y2, y1r, y1i);
|
|
|
|
FMADD2(x0r, x1r, dar_vec, y0r, y1r);
|
|
y0i OP0 dar_vec * x0i;
|
|
y1i OP0 dar_vec * x1i;
|
|
y0r OP1 dai_vec * x0i;
|
|
y1r OP1 dai_vec * x1i;
|
|
y0i OP2 dai_vec * x0r;
|
|
y1i OP2 dai_vec * x1r;
|
|
|
|
ILVRL_W2_SP(y0i, y0r, y0, y1);
|
|
ILVRL_W2_SP(y1i, y1r, y2, y3);
|
|
ST_SP4_INC(y0, y1, y2, y3, y, 4);
|
|
}
|
|
|
|
if (n & 4)
|
|
{
|
|
LD_SP2_INC(x, 4, x0, x1);
|
|
LD_SP2_INC(py, 4, y0, y1);
|
|
PCKEVOD_W2_SP(x1, x0, x0r, x0i);
|
|
PCKEVOD_W2_SP(y1, y0, y0r, y0i);
|
|
|
|
y0r += dar_vec * x0r;
|
|
y0i OP0 dar_vec * x0i;
|
|
y0r OP1 dai_vec * x0i;
|
|
y0i OP2 dai_vec * x0r;
|
|
|
|
ILVRL_W2_SP(y0i, y0r, y0, y1);
|
|
ST_SP2_INC(y0, y1, y, 4);
|
|
}
|
|
|
|
if (n & 2)
|
|
{
|
|
LD_GP4_INC(x, 1, xd0, xd1, xd2, xd3);
|
|
LD_GP4_INC(py, 1, yd0, yd1, yd2, yd3);
|
|
|
|
FMADD2(xd0, xd2, da_r, yd0, yd2);
|
|
yd1 OP0 da_r * xd1;
|
|
yd3 OP0 da_r * xd3;
|
|
yd0 OP1 da_i * xd1;
|
|
yd2 OP1 da_i * xd3;
|
|
yd1 OP2 da_i * xd0;
|
|
yd3 OP2 da_i * xd2;
|
|
|
|
ST_GP4_INC(yd0, yd1, yd2, yd3, y, 1);
|
|
}
|
|
|
|
if (n & 1)
|
|
{
|
|
LD_GP2_INC(x, 1, xd0, xd1);
|
|
LD_GP2_INC(py, 1, yd0, yd1);
|
|
|
|
yd0 += da_r * xd0;
|
|
yd1 OP0 da_r * xd1;
|
|
yd0 OP1 da_i * xd1;
|
|
yd1 OP2 da_i * xd0;
|
|
|
|
ST_GP2_INC(yd0, yd1, y, 1);
|
|
}
|
|
}
|
|
}
|
|
else if (1 == inc_y)
|
|
{
|
|
FLOAT *y_pref;
|
|
BLASLONG pref_offset;
|
|
v4f32 x8, x9, x10, x11, x12, x13, x14;
|
|
|
|
pref_offset = (BLASLONG)y & (L1_DATA_LINESIZE - 1);
|
|
if (pref_offset > 0)
|
|
{
|
|
pref_offset = L1_DATA_LINESIZE - pref_offset;
|
|
pref_offset = pref_offset / sizeof(FLOAT);
|
|
}
|
|
y_pref = y + pref_offset + 64;
|
|
|
|
inc_x2 = 2 * inc_x;
|
|
|
|
dar_vec = COPY_FLOAT_TO_VECTOR(da_r);
|
|
dai_vec = COPY_FLOAT_TO_VECTOR(da_i);
|
|
|
|
for (i = (n >> 4); i--;)
|
|
{
|
|
PREF_OFFSET(y_pref, 0);
|
|
PREF_OFFSET(y_pref, 32);
|
|
PREF_OFFSET(y_pref, 64);
|
|
PREF_OFFSET(y_pref, 96);
|
|
y_pref += 32;
|
|
|
|
LD_SP8_INC(x, inc_x2, x0, x1, x2, x3, x4, x5, x6, x14);
|
|
LD_SP7_INC(x, inc_x2, x8, x9, x10, x11, x12, x13, x7);
|
|
PCKEV_D2_SP(x1, x0, x3, x2, x0, x1);
|
|
PCKEV_D2_SP(x5, x4, x14, x6, x2, x3);
|
|
PCKEV_D2_SP(x9, x8, x11, x10, x4, x5);
|
|
x6 = (v4f32) __msa_pckev_d((v2i64) x13, (v2i64) x12);
|
|
x7 = (v4f32) __msa_insert_w((v4i32) x7, 2, *((int *) x));
|
|
x7 = (v4f32) __msa_insert_w((v4i32) x7, 3, *((int *) (x + 1)));
|
|
x += inc_x2;
|
|
|
|
LD_SP8_INC(py, 4, y0, y1, y2, y3, y4, y5, y6, y7);
|
|
PCKEVOD_W2_SP(x1, x0, x0r, x0i);
|
|
PCKEVOD_W2_SP(y1, y0, y0r, y0i);
|
|
PCKEVOD_W2_SP(x3, x2, x1r, x1i);
|
|
PCKEVOD_W2_SP(y3, y2, y1r, y1i);
|
|
PCKEVOD_W2_SP(x5, x4, x2r, x2i);
|
|
PCKEVOD_W2_SP(y5, y4, y2r, y2i);
|
|
PCKEVOD_W2_SP(x7, x6, x3r, x3i);
|
|
PCKEVOD_W2_SP(y7, y6, y3r, y3i);
|
|
|
|
FMADD4(x0r, x1r, x2r, x3r, dar_vec, y0r, y1r, y2r, y3r);
|
|
y0i OP0 dar_vec * x0i;
|
|
y1i OP0 dar_vec * x1i;
|
|
y2i OP0 dar_vec * x2i;
|
|
y3i OP0 dar_vec * x3i;
|
|
y0r OP1 dai_vec * x0i;
|
|
y1r OP1 dai_vec * x1i;
|
|
y2r OP1 dai_vec * x2i;
|
|
y3r OP1 dai_vec * x3i;
|
|
y0i OP2 dai_vec * x0r;
|
|
y1i OP2 dai_vec * x1r;
|
|
y2i OP2 dai_vec * x2r;
|
|
y3i OP2 dai_vec * x3r;
|
|
|
|
ILVRL_W2_SP(y0i, y0r, y0, y1);
|
|
ILVRL_W2_SP(y1i, y1r, y2, y3);
|
|
ILVRL_W2_SP(y2i, y2r, y4, y5);
|
|
ILVRL_W2_SP(y3i, y3r, y6, y7);
|
|
|
|
ST_SP8_INC(y0, y1, y2, y3, y4, y5, y6, y7, y, 4);
|
|
}
|
|
|
|
if (n & 15)
|
|
{
|
|
if (n & 8)
|
|
{
|
|
LD_SP7_INC(x, inc_x2, x0, x1, x2, x6, x4, x5, x3);
|
|
PCKEV_D2_SP(x1, x0, x6, x2, x0, x1);
|
|
|
|
x2 = (v4f32) __msa_pckev_d((v2i64) x5, (v2i64) x4);
|
|
x3 = (v4f32) __msa_insert_w((v4i32) x3, 2, *((int *) x));
|
|
x3 = (v4f32) __msa_insert_w((v4i32) x3, 3, *((int *) (x + 1)));
|
|
x += inc_x2;
|
|
|
|
LD_SP4_INC(py, 4, y0, y1, y2, y3);
|
|
PCKEVOD_W2_SP(x1, x0, x0r, x0i);
|
|
PCKEVOD_W2_SP(y1, y0, y0r, y0i);
|
|
PCKEVOD_W2_SP(x3, x2, x1r, x1i);
|
|
PCKEVOD_W2_SP(y3, y2, y1r, y1i);
|
|
|
|
FMADD2(x0r, x1r, dar_vec, y0r, y1r);
|
|
y0i OP0 dar_vec * x0i;
|
|
y1i OP0 dar_vec * x1i;
|
|
y0r OP1 dai_vec * x0i;
|
|
y1r OP1 dai_vec * x1i;
|
|
y0i OP2 dai_vec * x0r;
|
|
y1i OP2 dai_vec * x1r;
|
|
|
|
ILVRL_W2_SP(y0i, y0r, y0, y1);
|
|
ILVRL_W2_SP(y1i, y1r, y2, y3);
|
|
ST_SP4_INC(y0, y1, y2, y3, y, 4);
|
|
}
|
|
|
|
if (n & 4)
|
|
{
|
|
LD_SP3_INC(x, inc_x2, x0, x2, x1);
|
|
|
|
x0 = (v4f32) __msa_pckev_d((v2i64) x2, (v2i64) x0);
|
|
x1 = (v4f32) __msa_insert_w((v4i32) x1, 2, *((int *) x));
|
|
x1 = (v4f32) __msa_insert_w((v4i32) x1, 3, *((int *) (x + 1)));
|
|
x += inc_x2;
|
|
|
|
LD_SP2_INC(py, 4, y0, y1);
|
|
PCKEVOD_W2_SP(x1, x0, x0r, x0i);
|
|
PCKEVOD_W2_SP(y1, y0, y0r, y0i);
|
|
|
|
y0r += dar_vec * x0r;
|
|
y0i OP0 dar_vec * x0i;
|
|
y0r OP1 dai_vec * x0i;
|
|
y0i OP2 dai_vec * x0r;
|
|
|
|
ILVRL_W2_SP(y0i, y0r, y0, y1);
|
|
ST_SP2_INC(y0, y1, y, 4);
|
|
}
|
|
|
|
if (n & 2)
|
|
{
|
|
xd0 = x[0];
|
|
xd1 = x[1];
|
|
x += inc_x2;
|
|
xd2 = x[0];
|
|
xd3 = x[1];
|
|
x += inc_x2;
|
|
|
|
LD_GP4_INC(py, 1, yd0, yd1, yd2, yd3);
|
|
|
|
FMADD2(xd0, xd2, da_r, yd0, yd2);
|
|
yd1 OP0 da_r * xd1;
|
|
yd3 OP0 da_r * xd3;
|
|
yd0 OP1 da_i * xd1;
|
|
yd2 OP1 da_i * xd3;
|
|
yd1 OP2 da_i * xd0;
|
|
yd3 OP2 da_i * xd2;
|
|
|
|
ST_GP4_INC(yd0, yd1, yd2, yd3, y, 1);
|
|
}
|
|
|
|
if (n & 1)
|
|
{
|
|
LD_GP2_INC(x, 1, xd0, xd1);
|
|
LD_GP2_INC(py, 1, yd0, yd1);
|
|
|
|
yd0 += da_r * xd0;
|
|
yd1 OP0 da_r * xd1;
|
|
yd0 OP1 da_i * xd1;
|
|
yd1 OP2 da_i * xd0;
|
|
|
|
ST_GP2_INC(yd0, yd1, y, 1);
|
|
}
|
|
}
|
|
}
|
|
else
|
|
{
|
|
inc_x2 = 2 * inc_x;
|
|
inc_y2 = 2 * inc_y;
|
|
|
|
for (i = (n >> 2); i--;)
|
|
{
|
|
xd0 = x[0];
|
|
xd1 = x[1];
|
|
x += inc_x2;
|
|
xd2 = x[0];
|
|
xd3 = x[1];
|
|
x += inc_x2;
|
|
xd4 = x[0];
|
|
xd5 = x[1];
|
|
x += inc_x2;
|
|
xd6 = x[0];
|
|
xd7 = x[1];
|
|
x += inc_x2;
|
|
|
|
yd0 = py[0];
|
|
yd1 = py[1];
|
|
py += inc_y2;
|
|
yd2 = py[0];
|
|
yd3 = py[1];
|
|
py += inc_y2;
|
|
yd4 = py[0];
|
|
yd5 = py[1];
|
|
py += inc_y2;
|
|
yd6 = py[0];
|
|
yd7 = py[1];
|
|
py += inc_y2;
|
|
|
|
FMADD4(xd0, xd2, xd4, xd6, da_r, yd0, yd2, yd4, yd6);
|
|
yd1 OP0 da_r * xd1;
|
|
yd3 OP0 da_r * xd3;
|
|
yd5 OP0 da_r * xd5;
|
|
yd7 OP0 da_r * xd7;
|
|
yd0 OP1 da_i * xd1;
|
|
yd2 OP1 da_i * xd3;
|
|
yd4 OP1 da_i * xd5;
|
|
yd6 OP1 da_i * xd7;
|
|
yd1 OP2 da_i * xd0;
|
|
yd3 OP2 da_i * xd2;
|
|
yd5 OP2 da_i * xd4;
|
|
yd7 OP2 da_i * xd6;
|
|
|
|
y[0] = yd0;
|
|
y[1] = yd1;
|
|
y += inc_y2;
|
|
y[0] = yd2;
|
|
y[1] = yd3;
|
|
y += inc_y2;
|
|
y[0] = yd4;
|
|
y[1] = yd5;
|
|
y += inc_y2;
|
|
y[0] = yd6;
|
|
y[1] = yd7;
|
|
y += inc_y2;
|
|
}
|
|
|
|
if (n & 3)
|
|
{
|
|
if (n & 2)
|
|
{
|
|
xd0 = x[0];
|
|
xd1 = x[1];
|
|
x += inc_x2;
|
|
xd2 = x[0];
|
|
xd3 = x[1];
|
|
x += inc_x2;
|
|
|
|
yd0 = py[0];
|
|
yd1 = py[1];
|
|
py += inc_y2;
|
|
yd2 = py[0];
|
|
yd3 = py[1];
|
|
py += inc_y2;
|
|
|
|
FMADD2(xd0, xd2, da_r, yd0, yd2);
|
|
yd1 OP0 da_r * xd1;
|
|
yd3 OP0 da_r * xd3;
|
|
yd0 OP1 da_i * xd1;
|
|
yd2 OP1 da_i * xd3;
|
|
yd1 OP2 da_i * xd0;
|
|
yd3 OP2 da_i * xd2;
|
|
|
|
y[0] = yd0;
|
|
y[1] = yd1;
|
|
y += inc_y2;
|
|
y[0] = yd2;
|
|
y[1] = yd3;
|
|
y += inc_y2;
|
|
}
|
|
|
|
if (n & 1)
|
|
{
|
|
xd0 = x[0];
|
|
xd1 = x[1];
|
|
|
|
yd0 = y[0];
|
|
yd1 = y[1];
|
|
|
|
yd0 += da_r * xd0;
|
|
yd1 OP0 da_r * xd1;
|
|
yd0 OP1 da_i * xd1;
|
|
yd1 OP2 da_i * xd0;
|
|
|
|
y[0] = yd0;
|
|
y[1] = yd1;
|
|
}
|
|
}
|
|
}
|
|
|
|
return (0);
|
|
}
|