optimize sdot lsx kernel

Author: ErnstPeng

kernel/loongarch64/KERNEL.LA264

@@ -1,6 +1,6 @@
 ifndef NO_LSX
 
-SDOTKERNEL  = dot_lsx.S
+SDOTKERNEL  = sdot_lsx.S
 DSDOTKERNEL = dot_lsx.S
 DDOTKERNEL  = dot_lsx.S
 CDOTKERNEL  = cdot_lsx.S

kernel/loongarch64/sdot_lsx.S

@@ -0,0 +1,121 @@
+/***************************************************************************
+Copyright (c) 2023, 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.
+*****************************************************************************/
+
+#define ASSEMBLER
+
+#include "common.h"
+
+#define N      $r4
+#define X      $r5
+#define INCX   $r6
+#define Y      $r7
+#define INCY   $r8
+
+#define I      $r17
+#define TEMP   $r18
+
+/* Don't change following FR unless you know the effects. */
+#define s1     $f8
+#define s2     $f9
+#define a1     $f10
+#define b1     $f11
+
+    PROLOGUE
+
+    vxor.v    $vr8,   $vr8,    $vr8
+    vxor.v    $vr9,   $vr9,    $vr9
+    slli.d    INCX,   INCX,    BASE_SHIFT
+    li.d      TEMP,   SIZE
+    slli.d    INCY,   INCY,    BASE_SHIFT
+    bge       $r0,    N,       .L999
+    bne       INCX,   TEMP,    .L20   /* inc_x=1 */
+    bne       INCY,   TEMP,    .L20   /* inc_y=1 */
+
+    /* ((inc_x == 1) && (inc_y == 1)) */
+    srai.d    I,      N,       4
+    bge       $r0,    I,       .L12   /* FLOAT: <16 */
+.L11:
+    /* FLOAT: 16~ */
+    vld       $vr0,   X,       0
+    vld       $vr1,   X,       16
+    vld       $vr2,   X,       32
+    vld       $vr3,   X,       48
+    vld       $vr4,   Y,       0
+    vld       $vr5,   Y,       16
+    vld       $vr6,   Y,       32
+    vld       $vr7,   Y,       48
+    addi.d    I,      I,       -1
+    addi.d    X,      X,       64
+    addi.d    Y,      Y,       64
+
+    vfmadd.s    $vr8,   $vr0,    $vr4,   $vr8
+    vfmadd.s    $vr9,   $vr1,    $vr5,   $vr9
+    vfmadd.s    $vr8,   $vr2,    $vr6,   $vr8
+    vfmadd.s    $vr9,   $vr3,    $vr7,   $vr9
+
+    bnez      I,      .L11
+
+    vfadd.s   $vr8,   $vr8,    $vr9
+    vextrins.w $vr1,  $vr8,    0x01
+    vextrins.w $vr2,  $vr8,    0x02
+    vextrins.w $vr3,  $vr8,    0x03
+    fadd.s     $f8,   $f8,     $f1
+    fadd.s     $f8,   $f8,     $f2
+    fadd.s     $f8,   $f8,     $f3
+.L12:
+    andi      I,      N,       0xf
+    bge       $r0,    I,       .L999
+.L13:
+    addi.d    I,      I,       -1
+    fld.s     $f0,    X,       0
+    fld.s     $f4,    Y,       0
+    addi.d    X,      X,       4
+    addi.d    Y,      Y,       4
+    fmadd.s   $f8,   $f0,   $f4,  $f8
+    bnez      I,      .L13
+    b         .L999
+
+    /* !((inc_x == 1) && (inc_y == 1)) */
+.L20:
+    move      I,      N
+.L21:
+    addi.d    I,      I,       -1
+
+    fld.s     $f0,    X,    0
+    fld.s     $f4,    Y,    0
+    add.d     X,      X,    INCX
+    add.d     Y,      Y,    INCY
+    fmadd.s   $f8,    $f0,  $f4,   $f8
+
+    bnez      I,      .L21
+    b         .L999
+
+.L999:
+    fmov.s    $f0,    $f8
+    jirl      $r0,    $r1,     0x0
+
+    EPILOGUE