OpenBLAS/kernel/alpha/gemm_kernel_4x4.S

/*********************************************************************/
/* Copyright 2009, 2010 The University of Texas at Austin.           */
/* 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  UNIVERSITY OF  TEXAS AT    */
/*    AUSTIN  ``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 UNIVERSITY  OF TEXAS AT    */
/*    AUSTIN 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.                                    */
/*                                                                   */
/* The views and conclusions contained in the software and           */
/* documentation are those of the authors and should not be          */
/* interpreted as representing official policies, either expressed   */
/* or implied, of The University of Texas at Austin.                 */
/*********************************************************************/

#define ASSEMBLER
#include "common.h"
#include "version.h"

#if !defined(EV4) && !defined(EV5) && !defined(EV6)
#error "Architecture is not specified."
#endif

#ifdef EV6
#define PREFETCHSIZE 56
#define UNOP unop
#endif

#ifdef EV5
#define PREFETCHSIZE 56
#define UNOP
#endif

#ifdef EV4
#define UNOP
#endif

#define STACKSIZE 80

#define M	$16
#define N	$17
#define K	$18
#define A	$20
#define	B	$21
#define C	$22
#define	LDC	$23

#define C1	$19
#define C2	$24
#define	C3	$25
#define C4	$27

#define AO	$at
#define	BO	$5
#define I	$6
#define J	$7
#define L	$8

#define a1	$f16
#define a2	$f17
#define a3	$f18
#define a4	$f19

#define b1	$f20
#define b2	$f21
#define b3	$f22
#define b4	$f23

#define t1	$f24
#define t2	$f25
#define t3	$f26
#define t4	$f27

#define a5	$f28
#define a6	$f30
#define b5	$f29

#define alpha	$f30

#define c01	$f0
#define c02	$f1
#define c03	$f2
#define c04	$f3

#define c05	$f4
#define c06	$f5
#define c07	$f6
#define c08	$f7

#define c09	$f8
#define c10	$f9
#define c11	$f10
#define c12	$f11

#define c13	$f12
#define c14	$f13
#define c15	$f14
#define c16	$f15

#define TMP1	$0
#define TMP2	$1
#define KK	$2
#define	BB	$3
#define OFFSET  $4

#define ALPHA	64($sp)

	PROLOGUE
	PROFCODE
	.frame	$sp, STACKSIZE, $26, 0

	lda	$sp, -STACKSIZE($sp)

	ldq	C,        0 + STACKSIZE($sp)
	ldq	LDC,      8 + STACKSIZE($sp)
#ifdef TRMMKERNEL
	ldq	OFFSET,  16 + STACKSIZE($sp)
#endif

	SXADDQ	LDC, 0, LDC

	stt	$f2,   0($sp)
	stt	$f3,   8($sp)
	stt	$f4,  16($sp)
	stt	$f5,  24($sp)
	stt	$f6,  32($sp)
	stt	$f7,  40($sp)
	stt	$f8,  48($sp)
	stt	$f9,  56($sp)
	stt	$f19, ALPHA

	cmple	M, 0, $0
	cmple	N, 0, $1
	cmple	K, 0, $2

	or	$0, $1, $0
	or	$0, $2, $0
	bne	$0, $L999

#if defined(TRMMKERNEL) && !defined(LEFT)
	subq	$31, OFFSET, KK
#endif

	sra	N, 2, J
	ble	J, $L40
	.align 4

$L01:
	mov	C,  C1
	addq	C,  LDC, C2
	mov	A, AO
	s4addq	K, 0, BB

#if defined(TRMMKERNEL) &&  defined(LEFT)
	mov	OFFSET, KK
#endif

	addq	C2, LDC, C3
	s4addq	LDC, C, C

	SXADDQ	BB, B, BB
	fclr	t1
	addq	C3, LDC, C4
	fclr	t2

	sra	M,  2, I
	fclr	t3
	fclr	t4
	ble	I, $L20
	.align 4

$L11:
#if defined(EV5) || defined(EV6)
	ldl	$31,   0 * SIZE(BB)
	ldl	$31,   8 * SIZE(BB)
	unop
	lda	BB,   16 * SIZE(BB)
#endif

#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 4, TMP1
#else
	addq	KK, 4, TMP1
#endif
#endif

	LD	a1,    0 * SIZE(AO)
	fclr	c11
	LD	a2,    1 * SIZE(AO)
	fclr	c12

	LD	a3,    2 * SIZE(AO)
	fclr	c16
	LD	a4,    3 * SIZE(AO)
	fclr	c15

	LD	b1,    0 * SIZE(B)
	fclr	c01
	LD	b2,    1 * SIZE(B)
	fclr	c02

	LD	b3,    2 * SIZE(B)
	fclr	c06
	LD	b4,    3 * SIZE(B)
	fclr	c05

 	lds	$f31,  4 * SIZE(C1)
	fclr	c03
#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif
	fclr	c04

	lds	$f31,  7 * SIZE(C2)
	fclr	c08
	lda	BO,    4 * SIZE(B)
	fclr	c13

 	lds	$f31,  4 * SIZE(C3)
 	fclr	c09
	lda	AO,    4 * SIZE(AO)
	fclr	c10

#else
	sll	KK, BASE_SHIFT + 2, TMP1
	addq	AO, TMP1, AO
	addq	B,  TMP1, BO
	subq	K, KK, TMP1

	LD	a1,    0 * SIZE(AO)
	fclr	c11
	LD	a2,    1 * SIZE(AO)
	fclr	c12

	LD	a3,    2 * SIZE(AO)
	fclr	c16
	LD	a4,    3 * SIZE(AO)
	fclr	c15

	LD	b1,    0 * SIZE(BO)
	fclr	c01
	LD	b2,    1 * SIZE(BO)
	fclr	c02

	LD	b3,    2 * SIZE(BO)
	fclr	c06
	LD	b4,    3 * SIZE(BO)
	fclr	c05

 	lds	$f31,  4 * SIZE(C1)
	fclr	c03
	lda	L,        -2(TMP1)
	fclr	c04

	lds	$f31,  7 * SIZE(C2)
	fclr	c08
	lda	BO,    4 * SIZE(BO)
	fclr	c13

 	lds	$f31,  4 * SIZE(C3)
 	fclr	c09
	lda	AO,    4 * SIZE(AO)
	fclr	c10
#endif

	lds	$f31,  7 * SIZE(C4)
	fclr	c14
	fclr	c07
	ble	L, $L15
	.align	5

$L12:
/*  1 */
	ADD	c11,  t1, c11
#ifndef EV4
	ldq	$31,   PREFETCHSIZE * SIZE(AO)
#else
	unop
#endif
	MUL	b1, a1, t1
#ifndef EV4
	ldl	$31,   PREFETCHSIZE * SIZE(BO)
#else
	unop
#endif

	ADD	c12,  t2, c12
	unop
	MUL	b1, a2, t2
	unop

	ADD	c16,  t3, c16
	unop
	MUL	b2, a2, t3
	LD	a5,   0 * SIZE(AO)

	ADD	c15, t4, c15
	unop
	MUL	b2, a1, t4
	LD	b5,   0 * SIZE(BO)

/*  2 */
	ADD	c01, t1, c01
	UNOP
	MUL	b1, a3, t1
	UNOP

	ADD	c02, t2, c02
	UNOP
	MUL	b1, a4, t2
	UNOP

	ADD	c06,  t3, c06
	unop
	MUL	b2, a4, t3
	unop

	ADD	c05, t4, c05
	unop
	MUL	b4, a1, t4
	unop

/*  3 */
	ADD	c03, t1, c03
	unop
	MUL	b3, a1, t1
	unop

	ADD	c04, t2, c04
	unop
	MUL	b3, a2, t2
	unop

	ADD	c08,  t3, c08
	unop
	MUL	b4, a2, t3
	LD	a2,  1 * SIZE(AO)

	ADD	c13, t4, c13
	unop
	MUL	b2, a3, t4
	LD	b2,  1 * SIZE(BO)

/*  4 */
	ADD	c09,  t1, c09
	unop
	MUL	b3, a3, t1
	LD	a6,  2 * SIZE(AO)

	ADD	c10,  t2, c10
	unop
	MUL	b3, a4, t2
	LD	b3,  2 * SIZE(BO)

	ADD	c14, t3, c14
	unop
	MUL	b4, a4, t3
	LD	a4,  3 * SIZE(AO)

	ADD	c07,  t4, c07
	unop
	MUL	b4, a3, t4
	LD	b4,  3 * SIZE(BO)

/*  5 */
	ADD	c11,  t1, c11
	unop
	MUL	b5,  a5,  t1
	LD	a1,  4 * SIZE(AO)

	ADD	c12,  t2, c12
	lda	L,        -2(L)
	MUL	b5,  a2, t2
	LD	b1,  4 * SIZE(BO)

	ADD	c16,  t3, c16
	unop
	MUL	b2, a2, t3
	unop

	ADD	c15, t4, c15
	unop
	MUL	b2, a5,  t4
	unop

/*  6 */
	ADD	c01, t1, c01
	unop
	MUL	b5,  a6, t1
	unop

	ADD	c02, t2, c02
	unop
	MUL	b5,  a4, t2
	unop

	ADD	c06,  t3, c06
	unop
	MUL	b2, a4, t3
	unop

	ADD	c05, t4, c05
	unop
	MUL	b4, a5,  t4
	unop

/*  7 */
	ADD	c03, t1, c03
	lda	AO,    8 * SIZE(AO)
	MUL	b3, a5,  t1
	unop

	ADD	c04, t2, c04
	lda	BO,    8 * SIZE(BO)
	MUL	b3, a2, t2
	unop

	ADD	c08,  t3, c08
	unop
	MUL	b4, a2, t3
	LD	a2, -3 * SIZE(AO)

	ADD	c13, t4, c13
	unop
	MUL	b2, a6, t4
	LD	b2, -3 * SIZE(BO)

/*  8 */
	ADD	c09,  t1, c09
	unop
	MUL	b3, a6, t1
	LD	a3, -2 * SIZE(AO)

	ADD	c10,  t2, c10
	unop
	MUL	b3, a4, t2
	LD	b3, -2 * SIZE(BO)

	ADD	c14, t3, c14
	unop
	MUL	b4, a4, t3
	LD	a4, -1 * SIZE(AO)

	ADD	c07,  t4, c07
	MUL	b4, a6, t4
	LD	b4, -1 * SIZE(BO)
	bgt	L,  $L12
	.align 4

$L15:
	ADD	c11,  t1, c11
	ldt	alpha, ALPHA
	MUL	b1, a1, t1
#ifndef TRMMKERNEL
	blbs	K, $L18
#else
	blbs	TMP1, $L18
#endif
	.align 4

	ADD	c12,  t2, c12
	MUL	b1, a2, t2
	ADD	c16,  t3, c16
	MUL	b2, a2, t3

	ADD	c15, t4, c15
	MUL	b2, a1, t4
	ADD	c01, t1, c01
	MUL	b1, a3, t1

	ADD	c02, t2, c02
	unop
	MUL	b1, a4, t2
	LD	b1,  0 * SIZE(BO)

	ADD	c06,  t3, c06
	MUL	b2, a4, t3
	ADD	c05, t4, c05
	MUL	b4, a1, t4

	ADD	c03, t1, c03
	unop
	MUL	b3, a1, t1
	LD	a1,  0 * SIZE(AO)

	ADD	c04, t2, c04
	unop
	MUL	b3, a2, t2
	unop

	ADD	c08,  t3, c08
	unop
	MUL	b4, a2, t3
	LD	a2,  1 * SIZE(AO)

	ADD	c13, t4, c13
	unop
	MUL	b2, a3, t4
	LD	b2,  1 * SIZE(BO)

	ADD	c09,  t1, c09
	unop
	MUL	b3, a3, t1
	lda	AO,  4 * SIZE(AO)

	ADD	c10,  t2, c10
	unop
	MUL	b3, a4, t2
 	LD	b3,  2 * SIZE(BO)

	ADD	c14, t3, c14
	unop
	MUL	b4, a4, t3
	LD	a4, -1 * SIZE(AO)

	ADD	c07,  t4, c07
	unop
	MUL	b4, a3, t4
	LD	a3, -2 * SIZE(AO)

	ADD	c11,  t1, c11
	LD	b4,  3 * SIZE(BO)
	MUL	b1, a1, t1
	lda	BO,  4 * SIZE(BO)
	.align 4

$L18:
	ADD	c12,  t2, c12
	unop
	MUL	b1, a2, t2
#ifndef TRMMKERNEL
	LD	a5,  0 * SIZE(C1)
#else
	unop
#endif

	ADD	c16,  t3, c16
	unop
	MUL	b2, a2, t3
	unop

	ADD	c15, t4, c15
	unop
	MUL	b2, a1, t4
#ifndef TRMMKERNEL
	LD	b5,  1 * SIZE(C1)
#else
	unop
#endif

	ADD	c01, t1, c01
	unop
	MUL	b1, a3, t1
	unop

	ADD	c02, t2, c02
	unop
	MUL	b1, a4, t2
#ifndef TRMMKERNEL
	LD	b1,  0 * SIZE(C2)
#else
	unop
#endif

	ADD	c06,  t3, c06
	unop
	MUL	b2, a4, t3
	unop

	ADD	c05, t4, c05
	unop
	MUL	b4, a1, t4
	unop

	ADD	c03, t1, c03
	unop
	MUL	b3, a1, t1
	unop

	ADD	c04, t2, c04
	unop
	MUL	b3, a2, t2
#ifndef TRMMKERNEL
	LD	a1,  0 * SIZE(C3)
#else
	unop
#endif

	ADD	c08,  t3, c08
	unop
	MUL	b4, a2, t3
#ifndef TRMMKERNEL
	LD	a2,  2 * SIZE(C1)
#else
	unop
#endif

	ADD	c13, t4, c13
	unop
	MUL	b2, a3, t4
#ifndef TRMMKERNEL
	LD	b2,  3 * SIZE(C1)
#else
	unop
#endif

	ADD	c09,  t1, c09
	lda	I,        -1(I)
	MUL	b3, a3, t1
	unop

	ADD	c10,  t2, c10
	unop
	MUL	b3, a4, t2
#ifndef TRMMKERNEL
	LD	b3,  0 * SIZE(C4)
#else
	unop
#endif

	ADD	c14, t3, c14
	unop
	MUL	b4, a4, t3
#ifndef TRMMKERNEL
	LD	a4,  1 * SIZE(C2)
#else
	unop
#endif

	ADD	c07,  t4, c07
	unop
	MUL	b4, a3, t4
#ifndef TRMMKERNEL
  	LD	a3,  2 * SIZE(C2)
#else
	unop
#endif

	ADD	c11,  t1, c11
	unop
	MUL	alpha, c01, c01
#ifndef TRMMKERNEL
	LD	b4,  3 * SIZE(C2)
#else
	unop
#endif

	ADD	c12,  t2, c12
	unop
	MUL	alpha, c02, c02
#ifndef TRMMKERNEL
	LD	t1,  1 * SIZE(C3)
#else
	unop
#endif

	ADD	c16,  t3, c16
	unop
  	MUL	alpha, c03, c03
#ifndef TRMMKERNEL
	LD	t2,  2 * SIZE(C3)
#else
	unop
#endif

	ADD	c15, t4, c15
	unop
	MUL	alpha, c04, c04
#ifndef TRMMKERNEL
	LD	t3,  3 * SIZE(C3)
#else
	unop
#endif

	MUL	alpha, c05, c05
	unop
#ifndef TRMMKERNEL
	ADD	c01,  a5, c01
	LD	t4,  1 * SIZE(C4)
#else
	unop
	unop
#endif

	MUL	alpha, c06, c06
#ifndef TRMMKERNEL
	unop
	ADD	c02,  b5, c02
	LD	a5,  2 * SIZE(C4)
#endif

	MUL	alpha, c07, c07
#ifndef TRMMKERNEL
	unop
	ADD	c03,  a2, c03
	LD	b5,  3 * SIZE(C4)
#endif

	MUL	alpha, c08, c08
#ifndef TRMMKERNEL
	unop
	ADD	c04,  b2, c04
	unop
#endif

	MUL	alpha, c09, c09
	ST	c01,  0 * SIZE(C1)
#ifndef TRMMKERNEL
	ADD	c05,  b1, c05
	unop
#endif

	MUL	alpha, c10, c10
	ST	c02,  1 * SIZE(C1)
#ifndef TRMMKERNEL
	ADD	c06,  a4, c06
	unop
#endif

	MUL	alpha, c11, c11
	ST	c03,  2 * SIZE(C1)
#ifndef TRMMKERNEL
	ADD	c07,  a3, c07
	unop
#endif

	MUL	alpha, c12, c12
	ST	c04,  3 * SIZE(C1)
#ifndef TRMMKERNEL
	ADD	c08,  b4, c08
#else
	unop
#endif
	lda	C1,   4 * SIZE(C1)

	MUL	alpha, c13, c13
	ST	c05,  0 * SIZE(C2)
#ifndef TRMMKERNEL
	ADD	c09,  a1, c09
	unop
#endif

	MUL	alpha, c14, c14
 	ST	c06,  1 * SIZE(C2)
#ifndef TRMMKERNEL
	ADD	c10,  t1, c10
	unop
#endif

	MUL	alpha, c15, c15
	ST	c07,  2 * SIZE(C2)
#ifndef TRMMKERNEL
	ADD	c11,  t2, c11
	unop
#endif

	MUL	alpha, c16, c16
	ST	c08,  3 * SIZE(C2)
#ifndef TRMMKERNEL
	ADD	c12,  t3, c12
#else
	unop
#endif
	lda	C2,   4 * SIZE(C2)

#ifndef TRMMKERNEL
	ADD	c13,  b3, c13
#else
	unop
#endif
	ST	c09,  0 * SIZE(C3)
	fclr	t1
	lda	C4,   4 * SIZE(C4)

#ifndef TRMMKERNEL
	ADD	c14,  t4, c14
#else
	unop
#endif
	ST	c10,  1 * SIZE(C3)
	fclr	t2
	unop

#ifndef TRMMKERNEL
	ADD	c15,  a5, c15
#else
	unop
#endif
	ST	c11,  2 * SIZE(C3)
	fclr	t3
	unop

#ifndef TRMMKERNEL
	ADD	c16,  b5, c16
#else
	unop
#endif
	ST	c12,  3 * SIZE(C3)
	fclr	t4
	lda	C3,   4 * SIZE(C3)

	ST	c13, -4 * SIZE(C4)
	ST	c14, -3 * SIZE(C4)
	ST	c15, -2 * SIZE(C4)
	ST	c16, -1 * SIZE(C4)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 4, TMP1
#else
	subq	TMP1, 4, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 2, TMP1
	addq	AO, TMP1, AO
	addq	BO, TMP1, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 4, KK
#endif

	bgt	I, $L11
	.align 4

$L20:
	and	M,  2, I
	ble	I, $L30

#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 2, TMP1
#else
	addq	KK, 4, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
 	fclr	c09
	LD	a2,  1 * SIZE(AO)
	fclr	c13

	LD	a3,  2 * SIZE(AO)
	fclr	c10
	LD	a4,  3 * SIZE(AO)
	fclr	c14

	LD	b1,  0 * SIZE(B)
#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif
	LD	b2,  1 * SIZE(B)
	lda	AO,  2 * SIZE(AO)

	LD	b3,  2 * SIZE(B)
	fclr	c01
	LD	b4,  3 * SIZE(B)
	fclr	c05

	lda	BO,  4 * SIZE(B)
	fclr	c02
	fclr	c06
	ble	L, $L25

#else
	sll	KK, BASE_SHIFT + 1, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 2, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
 	fclr	c09
	LD	a2,  1 * SIZE(AO)
	fclr	c13

	LD	a3,  2 * SIZE(AO)
	fclr	c10
	LD	a4,  3 * SIZE(AO)
	fclr	c14

	LD	b1,  0 * SIZE(BO)
	lda	L,        -2(TMP1)
	LD	b2,  1 * SIZE(BO)
	lda	AO,  2 * SIZE(AO)

	LD	b3,  2 * SIZE(BO)
	fclr	c01
	LD	b4,  3 * SIZE(BO)
	fclr	c05

	lda	BO,  4 * SIZE(BO)
	fclr	c02
	fclr	c06
	ble	L, $L25
#endif
	.align	4

$L22:
	ADD	c09, t1, c09
	unop
	MUL	a1, b1, t1
	unop

	ADD	c10, t2, c10
	unop
	MUL	a2, b1, t2
	LD	b1,  0 * SIZE(BO)

	ADD	c13, t3, c13
	unop
	MUL	a1, b2, t3
	lda	BO,    8 * SIZE(BO)

	ADD	c14, t4, c14
	unop
	MUL	a2, b2, t4
	LD	b2, -7 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b3, t1
	unop

	ADD	c02, t2, c02
	unop
	MUL	a2, b3, t2
	LD	b3, -6 * SIZE(BO)

	ADD	c05, t3, c05
	unop
	MUL	a1, b4, t3
	LD	a1,  2 * SIZE(AO)

	ADD	c06, t4, c06
	MUL	a2, b4, t4
	LD	b5, -5 * SIZE(BO)

	ADD	c09, t1, c09
	unop
	MUL	a3, b1, t1
	LD	a2,  3 * SIZE(AO)

	ADD	c10, t2, c10
	unop
	MUL	a4, b1, t2
	LD	b1, -4 * SIZE(BO)

	ADD	c13, t3, c13
	unop
	MUL	a3, b2, t3
	lda	AO,    4 * SIZE(AO)

	ADD	c14, t4, c14
	MUL	a4, b2, t4
	LD	b2, -3 * SIZE(BO)

	ADD	c01, t1, c01
	lda	L,        -2(L)
	MUL	a3, b3, t1
	LD	b4, -1 * SIZE(BO)

	ADD	c02, t2, c02
	unop
	MUL	a4, b3, t2
	LD	b3, -2 * SIZE(BO)

	ADD	c05, t3, c05
	unop
	MUL	a3, b5, t3
	LD	a3,  0 * SIZE(AO)

	ADD	c06, t4, c06
	MUL	a4, b5, t4
	LD	a4,  1 * SIZE(AO)
	bgt	L,  $L22
	.align 4

$L25:
	ADD	c09, t1, c09
	ldt	alpha, ALPHA
	MUL	a1, b1, t1
#ifndef TRMMKERNEL
	blbs	K, $L28
#else
	blbs	TMP1, $L28
#endif

	ADD	c10, t2, c10
	unop
	MUL	a2, b1, t2
	LD	b1,  0 * SIZE(BO)

	ADD	c13, t3, c13
	unop
	MUL	a1, b2, t3
	unop

	ADD	c14, t4, c14
	unop
	MUL	a2, b2, t4
	LD	b2,  1 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b3, t1
	lda	AO,  2 * SIZE(AO)

	ADD	c02, t2, c02
	unop
	MUL	a2, b3, t2
	LD	b3,  2 * SIZE(BO)

	ADD	c05, t3, c05
	unop
	MUL	a1, b4, t3
	LD	a1, -2 * SIZE(AO)

	ADD	c06, t4, c06
	unop
	MUL	a2, b4, t4
	LD	a2, -1 * SIZE(AO)

	ADD	c09, t1, c09
	LD	b4,  3 * SIZE(BO)
	MUL	a1, b1, t1
	lda	BO,  4 * SIZE(BO)
	.align 4

$L28:
	ADD	c10, t2, c10
	unop
	MUL	a2, b1, t2
#ifndef TRMMKERNEL
	LD	a3,  0 * SIZE(C1)
#else
	unop
#endif

	ADD	c13, t3, c13
	unop
	MUL	a1, b2, t3
#ifndef TRMMKERNEL
	LD	a4,  1 * SIZE(C1)
#else
	unop
#endif

	ADD	c14, t4, c14
	unop
	MUL	a2, b2, t4
#ifndef TRMMKERNEL
	LD	a5,  0 * SIZE(C2)
#else
	unop
#endif

	ADD	c01, t1, c01
	unop
	MUL	a1, b3, t1
#ifndef TRMMKERNEL
	LD	b5,  1 * SIZE(C2)
#else
	unop
#endif

	ADD	c02, t2, c02
	unop
	MUL	a2, b3, t2
#ifndef TRMMKERNEL
	LD	b1,  0 * SIZE(C3)
#else
	unop
#endif

	ADD	c05, t3, c05
	unop
	MUL	a1, b4, t3
#ifndef TRMMKERNEL
	LD	b2,  1 * SIZE(C3)
#else
	unop
#endif

	ADD	c06, t4, c06
	unop
	MUL	a2, b4, t4
#ifndef TRMMKERNEL
	LD	b3,  0 * SIZE(C4)
#else
	unop
#endif

	ADD	c09, t1, c09
	unop
	MUL	alpha, c01, c01
#ifndef TRMMKERNEL
	LD	b4,  1 * SIZE(C4)
#else
	unop
#endif

	ADD	c10, t2, c10
	unop
	MUL	alpha, c02, c02
	unop

	ADD	c13, t3, c13
	MUL	alpha, c05, c05
	ADD	c14, t4, c14
	MUL	alpha, c06, c06

	MUL	alpha, c09, c09
#ifndef TRMMKERNEL
	ADD	c01,  a3, c01
#endif
	MUL	alpha, c10, c10
#ifndef TRMMKERNEL
	ADD	c02,  a4, c02
#endif

	MUL	alpha, c13, c13
#ifndef TRMMKERNEL
	ADD	c05,  a5, c05
#endif
	MUL	alpha, c14, c14
#ifndef TRMMKERNEL
	ADD	c06,  b5, c06
#endif

#ifndef TRMMKERNEL
	ADD	c09,  b1, c09
	unop
#endif
	ST	c01,  0 * SIZE(C1)
	fclr	t1

#ifndef TRMMKERNEL
	ADD	c10,  b2, c10
	unop
#endif
	ST	c02,  1 * SIZE(C1)
	fclr	t2

#ifndef TRMMKERNEL
	ADD	c13,  b3, c13
	unop
#endif
	ST	c05,  0 * SIZE(C2)
	fclr	t3

#ifndef TRMMKERNEL
	ADD	c14,  b4, c14
	unop
#endif
	ST	c06,  1 * SIZE(C2)
	fclr	t4

	ST	c09,  0 * SIZE(C3)
	lda	C1,   2 * SIZE(C1)
	ST	c10,  1 * SIZE(C3)
	lda	C2,   2 * SIZE(C2)

	ST	c13,  0 * SIZE(C4)
	lda	C3,   2 * SIZE(C3)
	ST	c14,  1 * SIZE(C4)
	lda	C4,   2 * SIZE(C4)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 2, TMP1
#else
	subq	TMP1, 4, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 1, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 2, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 2, KK
#endif
	.align 4

$L30:
	and	M,  1, I
	ble	I, $L39

#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 1, TMP1
#else
	addq	KK, 4, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	c01
	LD	a2,  1 * SIZE(AO)
	fclr	c05

	LD	b1,  0 * SIZE(B)
#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif
	LD	b2,  1 * SIZE(B)
	lda	AO,  1 * SIZE(AO)

	LD	b3,  2 * SIZE(B)
 	fclr	c09
	LD	b4,  3 * SIZE(B)
	fclr	c13

	lda	BO,  4 * SIZE(B)
	ble	L, $L35
#else
	sll	KK, BASE_SHIFT + 0, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 2, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	c01
	LD	a2,  1 * SIZE(AO)
	fclr	c05

	LD	b1,  0 * SIZE(BO)
	lda	L,        -2(TMP1)
	LD	b2,  1 * SIZE(BO)
	lda	AO,  1 * SIZE(AO)

	LD	b3,  2 * SIZE(BO)
 	fclr	c09
	LD	b4,  3 * SIZE(BO)
	fclr	c13

	lda	BO,  4 * SIZE(BO)
	ble	L, $L35
#endif
	.align	4

$L32:
	ADD	c01, t1, c01
	lda	L,        -2(L)
	MUL	a1, b1, t1
	LD	b1,  0 * SIZE(BO)

	ADD	c05, t2, c05
	lda	AO,    2 * SIZE(AO)
	MUL	a1, b2, t2
	LD	b2,  1 * SIZE(BO)

	ADD	c09, t3, c09
	LD	b5,  3 * SIZE(BO)
	MUL	a1, b3, t3
	LD	b3,  2 * SIZE(BO)

	ADD	c13, t4, c13
	MUL	a1, b4, t4
	LD	a1, -1 * SIZE(AO)

	ADD	c01, t1, c01
	MUL	a2, b1, t1
	LD	b1,  4 * SIZE(BO)
	lda	BO,    8 * SIZE(BO)

	ADD	c05, t2, c05
	MUL	a2, b2, t2
	LD	b2, -3 * SIZE(BO)

	ADD	c09, t3, c09
	LD	b4, -1 * SIZE(BO)
	MUL	a2, b3, t3
	LD	b3, -2 * SIZE(BO)

	ADD	c13, t4, c13
	MUL	a2, b5, t4
	LD	a2,  0 * SIZE(AO)
	bgt	L,  $L32
	.align 4

$L35:
	ADD	c01, t1, c01
	ldt	alpha, ALPHA
	MUL	a1, b1, t1
#ifndef TRMMKERNEL
	blbs	K, $L38
#else
	blbs	TMP1, $L38
#endif
	.align 4

	ADD	c05, t2, c05
	LD	b1,  0 * SIZE(BO)
	MUL	a1, b2, t2
	LD	b2,  1 * SIZE(BO)

	ADD	c09, t3, c09
	MUL	a1, b3, t3
	LD	b3,  2 * SIZE(BO)

	ADD	c13, t4, c13
	MUL	a1, b4, t4
	LD	a1,  0 * SIZE(AO)
	lda	AO,  1 * SIZE(AO)

	ADD	c01, t1, c01
	LD	b4,  3 * SIZE(BO)
	MUL	a1, b1, t1
	lda	BO,  4 * SIZE(BO)
	.align 4

$L38:
	ADD	c05, t2, c05
	unop
	MUL	a1, b2, t2
#ifndef TRMMKERNEL
	LD	a5,  0 * SIZE(C1)
#else
	unop
#endif

	ADD	c09, t3, c09
	unop
	MUL	a1, b3, t3
#ifndef TRMMKERNEL
	LD	b5,  0 * SIZE(C2)
#else
	unop
#endif

	ADD	c13, t4, c13
	unop
	MUL	a1, b4, t4
#ifndef TRMMKERNEL
	LD	a2,  0 * SIZE(C3)
#else
	unop
#endif

	ADD	c01, t1, c01
	unop
	MUL	alpha, c01, c01
#ifndef TRMMKERNEL
	LD	a3,  0 * SIZE(C4)
#else
	unop
#endif

	ADD	c05, t2, c05
	unop
	MUL	alpha, c05, c05
	unop

	ADD	c09, t3, c09
	MUL	alpha, c09, c09
	ADD	c13, t4, c13
	MUL	alpha, c13, c13

#ifndef TRMMKERNEL
	ADD	c01,  a5, c01
	ADD	c05,  b5, c05
	ADD	c09,  a2, c09
	ADD	c13,  a3, c13
#endif

	ST	c01,  0 * SIZE(C1)
	ST	c05,  0 * SIZE(C2)
	ST	c09,  0 * SIZE(C3)
	ST	c13,  0 * SIZE(C4)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 1, TMP1
#else
	subq	TMP1, 4, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 0, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 2, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 1, KK
#endif
	.align 4

$L39:
 	mov	BO, B
	lda	J,        -1(J)
#if defined(TRMMKERNEL) && !defined(LEFT)
	addq	KK, 4, KK
#else
	unop
#endif
	bgt	J, $L01
	.align 4

$L40:
	and	N, 2, J
	ble	J, $L80

	mov	C,  C1
	addq	C,  LDC, C2
	mov	A, AO
	fclr	t1
	addq	C2, LDC, C
	fclr	t2

#if defined(TRMMKERNEL) &&  defined(LEFT)
	mov	OFFSET, KK
#endif

	sra	M,  2, I
	fclr	t3
	fclr	t4
	ble	I, $L60
	.align 4

$L51:
#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 4, TMP1
#else
	addq	KK, 2, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	c03
	LD	a2,  1 * SIZE(AO)
	fclr	c07
	LD	a3,  2 * SIZE(AO)
	fclr	c04
	LD	a4,  3 * SIZE(AO)
	fclr	c08

	LD	b1,  0 * SIZE(B)
	fclr	c01
	LD	b2,  1 * SIZE(B)
	fclr	c05
	LD	b3,  2 * SIZE(B)
 	fclr	c02
	LD	b4,  3 * SIZE(B)
	fclr	c06

#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif
	lda	BO,  2 * SIZE(B)
	lda	AO,  4 * SIZE(AO)
	ble	L, $L55
#else
	sll	KK, BASE_SHIFT + 2, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 1, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	c03
	LD	a2,  1 * SIZE(AO)
	fclr	c07
	LD	a3,  2 * SIZE(AO)
	fclr	c04
	LD	a4,  3 * SIZE(AO)
	fclr	c08

	LD	b1,  0 * SIZE(BO)
	fclr	c01
	LD	b2,  1 * SIZE(BO)
	fclr	c05
	LD	b3,  2 * SIZE(BO)
 	fclr	c02
	LD	b4,  3 * SIZE(BO)
	fclr	c06

	lda	L,        -2(TMP1)
	lda	BO,  2 * SIZE(BO)
	lda	AO,  4 * SIZE(AO)
	ble	L, $L55
#endif
	.align	4

$L52:
	ADD	c05, t1, c05
	unop
	MUL	a1, b1, t1
	unop

	ADD	c06, t2, c06
	lda	L,   -2(L)
	MUL	a2, b1, t2
	unop

	ADD	c07, t3, c07
	unop
	MUL	a3, b1, t3
	unop

	ADD	c08, t4, c08
	unop
	MUL	a4, b1, t4
	LD	b1,  2 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b2, t1
	LD	a1,  0 * SIZE(AO)

	ADD	c02, t2, c02
	lda	BO,  4 * SIZE(BO)
	MUL	a2, b2, t2
	LD	a2,  1 * SIZE(AO)

	ADD	c03, t3, c03
	unop
	MUL	a3, b2, t3
	LD	a3,  2 * SIZE(AO)

	ADD	c04, t4, c04
	unop
	MUL	a4, b2, t4
	LD	a5,  3 * SIZE(AO)

	ADD	c05, t1, c05
	unop
	MUL	a1, b3, t1
	LD	b2, -1 * SIZE(BO)

	ADD	c06, t2, c06
	unop
	MUL	a2, b3, t2
	unop

	ADD	c07, t3, c07
	unop
	MUL	a3, b3, t3
	lda	AO,  8 * SIZE(AO)

	ADD	c08, t4, c08
	unop
	MUL	a5, b3, t4
	LD	b3,  0 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b4, t1
	LD	a1, -4 * SIZE(AO)

	ADD	c02, t2, c02
	unop
	MUL	a2, b4, t2
	LD	a2, -3 * SIZE(AO)

	ADD	c03, t3, c03
	LD	a4, -1 * SIZE(AO)
	MUL	a3, b4, t3
	LD	a3, -2 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a5, b4, t4
	LD	b4,  1 * SIZE(BO)
	bgt	L,  $L52
	.align 4

$L55:
	ADD	c05, t1, c05
	ldt	alpha, ALPHA
	MUL	a1, b1, t1
#ifndef TRMMKERNEL
	blbs	K, $L58
#else
	blbs	TMP1, $L58
#endif
	.align 4

	ADD	c06, t2, c06
	MUL	a2, b1, t2
	ADD	c07, t3, c07
	MUL	a3, b1, t3

	ADD	c08, t4, c08
	unop
	MUL	a4, b1, t4
	LD	b1,  0 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b2, t1
	LD	a1,  0 * SIZE(AO)

	ADD	c02, t2, c02
	unop
	MUL	a2, b2, t2
	LD	a2,  1 * SIZE(AO)

	ADD	c03, t3, c03
	unop
	MUL	a3, b2, t3
	LD	a3,  2 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a4, b2, t4
	LD	a4,  3 * SIZE(AO)
	lda	AO,  4 * SIZE(AO)

	ADD	c05, t1, c05
	LD	b2,  1 * SIZE(BO)
	MUL	a1, b1, t1
	lda	BO,  2 * SIZE(BO)
	.align 4

$L58:
	ADD	c06, t2, c06
	unop
	MUL	a2, b1, t2
#ifndef TRMMKERNEL
	LD	c09,  0 * SIZE(C1)
#else
	unop
#endif

	ADD	c07, t3, c07
	unop
	MUL	a3, b1, t3
#ifndef TRMMKERNEL
	LD	c10,  1 * SIZE(C1)
#else
	unop
#endif

	ADD	c08, t4, c08
	unop
	MUL	a4, b1, t4
#ifndef TRMMKERNEL
	LD	c11,  2 * SIZE(C1)
#else
	unop
#endif

	ADD	c01, t1, c01
	unop
	MUL	a1, b2, t1
#ifndef TRMMKERNEL
	LD	c12,  3 * SIZE(C1)
#else
	unop
#endif

	ADD	c02, t2, c02
	unop
	MUL	a2, b2, t2
#ifndef TRMMKERNEL
	LD	c13,  0 * SIZE(C2)
	unop
#endif

	ADD	c03, t3, c03
	unop
	MUL	a3, b2, t3
#ifndef TRMMKERNEL
	LD	c14,  1 * SIZE(C2)
#else
	unop
#endif

	ADD	c04, t4, c04
	unop
	MUL	a4, b2, t4
#ifndef TRMMKERNEL
	LD	c15,  2 * SIZE(C2)
#else
	unop
#endif

	ADD	c05, t1, c05
	unop
	MUL	alpha, c01, c01
#ifndef TRMMKERNEL
	LD	c16,  3 * SIZE(C2)
#else
	unop
#endif

	ADD	c06, t2, c06
	lda	I,        -1(I)
	MUL	alpha, c02, c02
	unop

	ADD	c07, t3, c07
	MUL	alpha, c03, c03
	ADD	c08, t4, c08
	MUL	alpha, c04, c04

	MUL	alpha, c05, c05
#ifndef TRMMKERNEL
	ADD	c01,  c09, c01
#endif
	MUL	alpha, c06, c06
#ifndef TRMMKERNEL
	ADD	c02,  c10, c02
#endif

	MUL	alpha, c07, c07
#ifndef TRMMKERNEL
	ADD	c03,  c11, c03
#endif
	MUL	alpha, c08, c08
#ifndef TRMMKERNEL
	ADD	c04,  c12, c04
#endif

#ifndef TRMMKERNEL
	ADD	c05,  c13, c05
#endif
	ST	c01,  0 * SIZE(C1)
#ifndef TRMMKERNEL
	ADD	c06,  c14, c06
#endif
	ST	c02,  1 * SIZE(C1)

#ifndef TRMMKERNEL
	ADD	c07,  c15, c07
#endif
	ST	c03,  2 * SIZE(C1)
#ifndef TRMMKERNEL
	ADD	c08,  c16, c08
#endif
	ST	c04,  3 * SIZE(C1)

	ST	c05,  0 * SIZE(C2)
	fclr	t1
	ST	c06,  1 * SIZE(C2)
	fclr	t2
	ST	c07,  2 * SIZE(C2)
	fclr	t3
	ST	c08,  3 * SIZE(C2)
	fclr	t4

	lda	C1,   4 * SIZE(C1)
	lda	C2,   4 * SIZE(C2)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 4, TMP1
#else
	subq	TMP1, 2, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 2, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 1, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 4, KK
#endif
	bgt	I, $L51
	.align 4

$L60:
	and	M,  2, I
	ble	I, $L70

#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 2, TMP1
#else
	addq	KK, 2, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	c01
	LD	a2,  1 * SIZE(AO)
	fclr	c05
	LD	a3,  2 * SIZE(AO)
 	fclr	c02
	LD	a4,  3 * SIZE(AO)
	fclr	c06

	LD	b1,  0 * SIZE(B)
#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif
	LD	b2,  1 * SIZE(B)
	lda	AO,  2 * SIZE(AO)

	LD	b3,  2 * SIZE(B)
	LD	b4,  3 * SIZE(B)
	lda	BO,  2 * SIZE(B)
	ble	L, $L65
#else
	sll	KK, BASE_SHIFT + 1, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 1, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	c01
	LD	a2,  1 * SIZE(AO)
	fclr	c05
	LD	a3,  2 * SIZE(AO)
 	fclr	c02
	LD	a4,  3 * SIZE(AO)
	fclr	c06

	LD	b1,  0 * SIZE(BO)
	lda	L,        -2(TMP1)
	LD	b2,  1 * SIZE(BO)
	lda	AO,  2 * SIZE(AO)

	LD	b3,  2 * SIZE(BO)
	LD	b4,  3 * SIZE(BO)
	lda	BO,  2 * SIZE(BO)
	ble	L, $L65
#endif
	.align	4

$L62:
	ADD	c01, t1, c01
	unop
	MUL	a1, b1, t1
	unop

	ADD	c02, t2, c02
	lda	AO,    4 * SIZE(AO)
	MUL	a2, b1, t2
	LD	b1,  2 * SIZE(BO)

	ADD	c05, t3, c05
	lda	L,        -2(L)
	MUL	a1, b2, t3
	LD	a1, -2 * SIZE(AO)

	ADD	c06, t4, c06
	unop
	MUL	a2, b2, t4
	LD	a2, -1 * SIZE(AO)

	ADD	c01, t1, c01
	LD	b2,  3 * SIZE(BO)
	MUL	a3, b3, t1
	lda	BO,    4 * SIZE(BO)

	ADD	c02, t2, c02
	unop
	MUL	a4, b3, t2
	LD	b3,  0 * SIZE(BO)

	ADD	c05, t3, c05
	unop
	MUL	a3, b4, t3
	LD	a3,  0 * SIZE(AO)

	ADD	c06, t4, c06
	MUL	a4, b4, t4
	LD	b4,  1 * SIZE(BO)
	unop

	LD	a4,  1 * SIZE(AO)
	unop
	unop
	bgt	L,  $L62
	.align 4

$L65:
	ADD	c01, t1, c01
	ldt	alpha, ALPHA
	MUL	a1, b1, t1
#ifndef TRMMKERNEL
	blbs	K, $L68
#else
	blbs	TMP1, $L68
#endif
	.align 4

	ADD	c02, t2, c02
	unop
	MUL	a2, b1, t2
	LD	b1,  0 * SIZE(BO)

	ADD	c05, t3, c05
	lda	BO,  2 * SIZE(BO)
	MUL	a1, b2, t3
	LD	a1,  0 * SIZE(AO)

	ADD	c06, t4, c06
	unop
	MUL	a2, b2, t4
	LD	a2,  1 * SIZE(AO)

	ADD	c01, t1, c01
	LD	b2, -1 * SIZE(BO)
	MUL	a1, b1, t1
	lda	AO,  2 * SIZE(AO)
	.align 4

$L68:
	ADD	c02, t2, c02
	unop
	MUL	a2, b1, t2
#ifndef TRMMKERNEL
	LD	c09,  0 * SIZE(C1)
#else
	unop
#endif

	ADD	c05, t3, c05
	unop
	MUL	a1, b2, t3
#ifndef TRMMKERNEL
	LD	c10,  1 * SIZE(C1)
#else
	unop
#endif

	ADD	c06, t4, c06
	unop
	MUL	a2, b2, t4
#ifndef TRMMKERNEL
	LD	c11,  0 * SIZE(C2)
#else
	unop
#endif

	ADD	c01, t1, c01
	unop
	MUL	alpha, c01, c01
#ifndef TRMMKERNEL
	LD	c12,  1 * SIZE(C2)
#else
	unop
#endif

	ADD	c02, t2, c02
	lda	C1,   2 * SIZE(C1)
	MUL	alpha, c02, c02
	lda	C2,   2 * SIZE(C2)

	ADD	c05, t3, c05
	MUL	alpha, c05, c05
	ADD	c06, t4, c06
	MUL	alpha, c06, c06

#ifndef TRMMKERNEL
	ADD	c01,  c09, c01
	ADD	c02,  c10, c02
	ADD	c05,  c11, c05
	ADD	c06,  c12, c06
#endif

	ST	c01, -2 * SIZE(C1)
	fclr	t1
	ST	c02, -1 * SIZE(C1)
	fclr	t2
	ST	c05, -2 * SIZE(C2)
	fclr	t3
	ST	c06, -1 * SIZE(C2)
	fclr	t4

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 2, TMP1
#else
	subq	TMP1, 2, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 1, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 1, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 2, KK
#endif
	.align 4

$L70:
	and	M,  1, I
	ble	I, $L79

#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 1, TMP1
#else
	addq	KK, 2, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	c01
	LD	a2,  1 * SIZE(AO)
	fclr	c05

	LD	b1,  0 * SIZE(B)
 	fclr	c02
	LD	b2,  1 * SIZE(B)
	fclr	c06

#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif

	LD	b3,  2 * SIZE(B)
	lda	AO,  1 * SIZE(AO)
	LD	b4,  3 * SIZE(B)
	lda	BO,  2 * SIZE(B)
	ble	L, $L75
#else
	sll	KK, BASE_SHIFT + 0, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 1, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	c01
	LD	a2,  1 * SIZE(AO)
	fclr	c05

	LD	b1,  0 * SIZE(BO)
 	fclr	c02
	LD	b2,  1 * SIZE(BO)
	fclr	c06

#ifndef TRMMKERNEL
	lda	L,        -2(K)
#else
	lda	L,        -2(TMP1)
#endif

	LD	b3,  2 * SIZE(BO)
	lda	AO,  1 * SIZE(AO)
	LD	b4,  3 * SIZE(BO)
	lda	BO,  2 * SIZE(BO)
	ble	L, $L75
#endif
	.align	4

$L72:
	ADD	c01, t1, c01
	lda	L,        -2(L)
	MUL	a1, b1, t1
	LD	b1,  2 * SIZE(BO)

	ADD	c05, t2, c05
	MUL	a1, b2, t2
	LD	a1,  1 * SIZE(AO)
	LD	b2,  3 * SIZE(BO)

	ADD	c02, t3, c02
	lda	AO,    2 * SIZE(AO)
	MUL	a2, b3, t3
	LD	b3,  4 * SIZE(BO)

	ADD	c06, t4, c06
	MUL	a2, b4, t4
	LD	a2,  0 * SIZE(AO)
	LD	b4,  5 * SIZE(BO)

	lda	BO,    4 * SIZE(BO)
	unop
	unop
	bgt	L,  $L72
	.align 4

$L75:
	ADD	c01, t1, c01
	ldt	alpha, ALPHA
	MUL	a1, b1, t1
#ifndef TRMMKERNEL
	blbs	K, $L78
#else
	blbs	TMP1, $L78
#endif
	.align 4

	ADD	c05, t2, c05
	MUL	a1, b2, t2
	LD	a1,  0 * SIZE(AO)
	LD	b1,  0 * SIZE(BO)

	ADD	c01, t1, c01
	LD	b2,  1 * SIZE(BO)
	lda	AO,  1 * SIZE(AO)
	MUL	a1, b1, t1
	lda	BO,  2 * SIZE(BO)
	.align 4

$L78:
	ADD	c05, t2, c05
	MUL	a1, b2, t2
#ifndef TRMMKERNEL
	LD	a5,  0 * SIZE(C1)
#else
	unop
#endif

	ADD	c02, t3, c02
	ADD	c06, t4, c06
#ifndef TRMMKERNEL
	LD	b5,  0 * SIZE(C2)
#else
	unop
#endif

	ADD	c01, c02, c01
	ADD	c05, c06, c05

	ADD	c01, t1, c01
	ADD	c05, t2, c05

	MUL	alpha, c01, c01
	MUL	alpha, c05, c05

#ifndef TRMMKERNEL
	ADD	c01,  a5, c01
	ADD	c05,  b5, c05
#endif

	ST	c01,  0 * SIZE(C1)
	ST	c05,  0 * SIZE(C2)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 1, TMP1
#else
	subq	TMP1, 2, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 0, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 1, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 1, KK
#endif
	.align 4

$L79:
 	mov	BO, B
#if defined(TRMMKERNEL) && !defined(LEFT)
	addq	KK, 2, KK
#else
	unop
#endif
	unop
	unop
	.align 4

$L80:
	and	N, 1, J
	ble	J, $L999

	mov	C,  C1
	mov	A, AO

#if defined(TRMMKERNEL) &&  defined(LEFT)
	mov	OFFSET, KK
#endif

	sra	M,  2, I
	ble	I, $L100
	.align 4

$L91:
#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 4, TMP1
#else
	addq	KK, 1, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	t1
	LD	a2,  1 * SIZE(AO)
	fclr	t2
	LD	a3,  2 * SIZE(AO)
	fclr	t3
	LD	a4,  3 * SIZE(AO)
	fclr	t4

	LD	b1,  0 * SIZE(B)
	fclr	c01
	LD	b2,  1 * SIZE(B)
	fclr	c02
	LD	b3,  2 * SIZE(B)
 	fclr	c03
	LD	b4,  3 * SIZE(B)
	fclr	c04

#ifndef TRMMKERNEL
	sra	K, 2, L
#else
	sra	TMP1, 2, L
#endif
	mov	B, BO
	unop
	ble	L, $L95
#else
	sll	KK, BASE_SHIFT + 2, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 0, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	t1
	LD	a2,  1 * SIZE(AO)
	fclr	t2
	LD	a3,  2 * SIZE(AO)
	fclr	t3
	LD	a4,  3 * SIZE(AO)
	fclr	t4

	LD	b1,  0 * SIZE(BO)
	fclr	c01
	LD	b2,  1 * SIZE(BO)
	fclr	c02
	LD	b3,  2 * SIZE(BO)
 	fclr	c03
	LD	b4,  3 * SIZE(BO)
	fclr	c04

#ifndef TRMMKERNEL
	sra	K, 2, L
#else
	sra	TMP1, 2, L
#endif
	unop
	ble	L, $L95
#endif
	.align	5

$L92:
	ADD	c01, t1, c01
	unop
	MUL	a1, b1, t1
	LD	a1,  4 * SIZE(AO)

	ADD	c02, t2, c02
	lda	L,   -1(L)
	MUL	a2, b1, t2
	LD	a2,  5 * SIZE(AO)

	ADD	c03, t3, c03
	unop
	MUL	a3, b1, t3
	LD	a3,  6 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a4, b1, t4
	LD	a4,  7 * SIZE(AO)
	LD	b1,  4 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b2, t1
	LD	a1,  8 * SIZE(AO)

	ADD	c02, t2, c02
	unop
	MUL	a2, b2, t2
	LD	a2,  9 * SIZE(AO)

	ADD	c03, t3, c03
	unop
	MUL	a3, b2, t3
	LD	a3, 10 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a4, b2, t4
	LD	a4, 11 * SIZE(AO)
	LD	b2,  5 * SIZE(BO)

	ADD	c01, t1, c01
	unop
	MUL	a1, b3, t1
	LD	a1, 12 * SIZE(AO)

	ADD	c02, t2, c02
	unop
	MUL	a2, b3, t2
	LD	a2, 13 * SIZE(AO)

	ADD	c03, t3, c03
	unop
	MUL	a3, b3, t3
	LD	a3, 14 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a4, b3, t4
	LD	a5, 15 * SIZE(AO)
	LD	b3,  6 * SIZE(BO)

	ADD	c01, t1, c01
	MUL	a1, b4, t1
	LD	a1, 16 * SIZE(AO)
	lda	AO, 16 * SIZE(AO)

	ADD	c02, t2, c02
	lda	BO,  4 * SIZE(BO)
	MUL	a2, b4, t2
	LD	a2,  1 * SIZE(AO)

	ADD	c03, t3, c03
	LD	a4,  3 * SIZE(AO)
	MUL	a3, b4, t3
	LD	a3,  2 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a5, b4, t4
	LD	b4,  3 * SIZE(BO)
	bgt	L,  $L92
	.align 4

$L95:
#ifndef TRMMKERNEL
	and	K, 3, L
#else
	and	TMP1, 3, L
#endif
	ldt	alpha, ALPHA
	unop
	ble	L, $L98
	.align 4

$L96:
	ADD	c01, t1, c01
	lda	L,   -1(L)
	MUL	a1, b1, t1
	LD	a1,  4 * SIZE(AO)

	ADD	c02, t2, c02
	lda	BO,  1 * SIZE(BO)
	MUL	a2, b1, t2
	LD	a2,  5 * SIZE(AO)

	ADD	c03, t3, c03
	unop
	MUL	a3, b1, t3
	LD	a3,  6 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a4, b1, t4
	LD	a4,  7 * SIZE(AO)
	LD	b1,  0 * SIZE(BO)

	lda	AO,  4 * SIZE(AO)
	bgt	L,  $L96
	.align 4

$L98:
#ifndef TRMMKERNEL
	ADD	c01, t1, c01
	LD	c05,  0 * SIZE(C1)
	ADD	c02, t2, c02
	LD	c06,  1 * SIZE(C1)
	ADD	c03, t3, c03
	LD	c07,  2 * SIZE(C1)
	ADD	c04, t4, c04
	LD	c08,  3 * SIZE(C1)
#else
	ADD	c01, t1, c01
	ADD	c02, t2, c02
	ADD	c03, t3, c03
	ADD	c04, t4, c04
#endif

	MUL	alpha, c01, c01
	MUL	alpha, c02, c02
	MUL	alpha, c03, c03
	MUL	alpha, c04, c04

#ifndef TRMMKERNEL
	ADD	c01,  c05, c01
	ADD	c02,  c06, c02
	ADD	c03,  c07, c03
	ADD	c04,  c08, c04
#endif

	ST	c01,  0 * SIZE(C1)
	ST	c02,  1 * SIZE(C1)
	ST	c03,  2 * SIZE(C1)
	ST	c04,  3 * SIZE(C1)

	lda	C1,   4 * SIZE(C1)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 4, TMP1
#else
	subq	TMP1, 1, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 2, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 0, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 4, KK
#endif

	lda	I,        -1(I)
	bgt	I, $L91
	.align 4

$L100:
	and	M,  2, I
	unop
	unop
	ble	I, $L110
	.align 4

$L101:
#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 2, TMP1
#else
	addq	KK, 1, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	t1
	LD	a2,  1 * SIZE(AO)
	fclr	t2
	LD	a3,  2 * SIZE(AO)
	fclr	t3
	LD	a4,  3 * SIZE(AO)
	fclr	t4

	LD	b1,  0 * SIZE(B)
	fclr	c01
	LD	b2,  1 * SIZE(B)
	fclr	c02
	LD	b3,  2 * SIZE(B)
 	fclr	c03
	LD	b4,  3 * SIZE(B)
	fclr	c04

#ifndef TRMMKERNEL
	sra	K, 2, L
#else
	sra	TMP1, 2, L
#endif
	mov	B, BO
	unop
	ble	L, $L105
#else
	sll	KK, BASE_SHIFT + 1, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 0, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	t1
	LD	a2,  1 * SIZE(AO)
	fclr	t2
	LD	a3,  2 * SIZE(AO)
	fclr	t3
	LD	a4,  3 * SIZE(AO)
	fclr	t4

	LD	b1,  0 * SIZE(BO)
	fclr	c01
	LD	b2,  1 * SIZE(BO)
	fclr	c02
	LD	b3,  2 * SIZE(BO)
 	fclr	c03
	LD	b4,  3 * SIZE(BO)
	fclr	c04

#ifndef TRMMKERNEL
	sra	K, 2, L
#else
	sra	TMP1, 2, L
#endif
	unop
	ble	L, $L105
#endif
	.align	5

$L102:
	ADD	c01, t1, c01
	lda	L,        -1(L)
	MUL	a1, b1, t1
	LD	a1,  4 * SIZE(AO)

	ADD	c02, t2, c02
	MUL	a2, b1, t2
	LD	a2,  5 * SIZE(AO)
	LD	b1,  4 * SIZE(BO)

	ADD	c03, t3, c03
	lda	BO,    4 * SIZE(BO)
	MUL	a3, b2, t3
	LD	a3,  6 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a4, b2, t4
	LD	a5,  7 * SIZE(AO)
	LD	b2,  1 * SIZE(BO)

	ADD	c01, t1, c01
	MUL	a1, b3, t1
	LD	a1,  8 * SIZE(AO)
	lda	AO,  8 * SIZE(AO)

	ADD	c02, t2, c02
	MUL	a2, b3, t2
	LD	b3,  2 * SIZE(BO)
	LD	a2,  1 * SIZE(AO)

	ADD	c03, t3, c03
	LD	a4,  3 * SIZE(AO)
	MUL	a3, b4, t3
	LD	a3,  2 * SIZE(AO)

	ADD	c04, t4, c04
	MUL	a5, b4, t4
	LD	b4,  3 * SIZE(BO)
	bgt	L,  $L102
	.align 4

$L105:
#ifndef TRMMKERNEL
	and	K, 3, L
#else
	and	TMP1, 3, L
#endif
	ldt	alpha, ALPHA
#ifndef TRMMKERNEL
	LD	a3,  0 * SIZE(C1)
	LD	a4,  1 * SIZE(C1)
#endif
	ble	L, $L108
	.align 4

$L106:
	ADD	c01, t1, c01
	lda	L,        -1(L)
	MUL	a1, b1, t1
	LD	a1,  2 * SIZE(AO)

	ADD	c02, t2, c02
	MUL	a2, b1, t2
	LD	a2,  3 * SIZE(AO)
	LD	b1,  1 * SIZE(BO)

	lda	AO,  2 * SIZE(AO)
	unop
	lda	BO,  1 * SIZE(BO)
	bgt	L,  $L106
	.align 4

$L108:
	ADD	c01, t1, c01
	fclr	t1
	ADD	c02, t2, c02
	fclr	t2
	ADD	c03, t3, c03
	fclr	t3
	ADD	c04, t4, c04
	fclr	t4

	ADD	c01, c03, c01
	ADD	c02, c04, c02

	MUL	alpha, c01, c01
	MUL	alpha, c02, c02

#ifndef TRMMKERNEL
	ADD	c01,  a3, c01
	ADD	c02,  a4, c02
#endif

	ST	c01,  0 * SIZE(C1)
	ST	c02,  1 * SIZE(C1)
	lda	C1,   2 * SIZE(C1)

#if (defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
    (defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))
	subq	K, KK, TMP1
#ifdef LEFT
	subq	TMP1, 2, TMP1
#else
	subq	TMP1, 1, TMP1
#endif
	sll	TMP1, BASE_SHIFT + 1, TMP2
	addq	AO, TMP2, AO
	sll	TMP1, BASE_SHIFT + 0, TMP2
	addq	BO, TMP2, BO
#endif

#if defined(TRMMKERNEL) && defined(LEFT)
	addq	KK, 2, KK
#endif
	.align 4

$L110:
	and	M,  1, I
	ble	I, $L999
	.align 4

$L111:
#if !defined(TRMMKERNEL) || \
	(defined(TRMMKERNEL) &&  defined(LEFT) &&  defined(TRANSA)) || \
	(defined(TRMMKERNEL) && !defined(LEFT) && !defined(TRANSA))

#ifdef TRMMKERNEL
#ifdef LEFT
	addq	KK, 1, TMP1
#else
	addq	KK, 1, TMP1
#endif
#endif

	LD	a1,  0 * SIZE(AO)
	fclr	t1
	LD	a2,  1 * SIZE(AO)
	fclr	t2
	LD	a3,  2 * SIZE(AO)
	fclr	t3
	LD	a4,  3 * SIZE(AO)
	fclr	t4

	LD	b1,  0 * SIZE(B)
	fclr	c01
	LD	b2,  1 * SIZE(B)
	fclr	c02
	LD	b3,  2 * SIZE(B)
 	fclr	c03
	LD	b4,  3 * SIZE(B)
	fclr	c04

#ifndef TRMMKERNEL
	sra	K, 2, L
#else
	sra	TMP1, 2, L
#endif
	mov	B, BO
	unop
	ble	L, $L115
#else
	sll	KK, BASE_SHIFT + 0, TMP1
	addq	AO, TMP1, AO
	sll	KK, BASE_SHIFT + 0, TMP2
	addq	B,  TMP2, BO
	subq	K, KK, TMP1

	LD	a1,  0 * SIZE(AO)
	fclr	t1
	LD	a2,  1 * SIZE(AO)
	fclr	t2
	LD	a3,  2 * SIZE(AO)
	fclr	t3
	LD	a4,  3 * SIZE(AO)
	fclr	t4

	LD	b1,  0 * SIZE(BO)
	fclr	c01
	LD	b2,  1 * SIZE(BO)
	fclr	c02
	LD	b3,  2 * SIZE(BO)
 	fclr	c03
	LD	b4,  3 * SIZE(BO)
	fclr	c04

#ifndef TRMMKERNEL
	sra	K, 2, L
#else
	sra	TMP1, 2, L
#endif
	unop
	ble	L, $L115
#endif
	.align	4

$L112:
	ADD	c01, t1, c01
	MUL	a1,  b1, t1
	LD	a1,  4 * SIZE(AO)
	LD	b1,  4 * SIZE(BO)

	ADD	c02, t2, c02
	MUL	a2,  b2, t2
	LD	a2,  5 * SIZE(AO)
	LD	b2,  5 * SIZE(BO)

	ADD	c03, t3, c03
	MUL	a3,  b3, t3
	LD	a3,  6 * SIZE(AO)
	LD	b3,  6 * SIZE(BO)

	ADD	c04, t4, c04
	MUL	a4,  b4, t4
	LD	a4,  7 * SIZE(AO)
	LD	b4,  7 * SIZE(BO)

	lda	L,        -1(L)
	lda	AO,    4 * SIZE(AO)
	lda	BO,    4 * SIZE(BO)
	bgt	L,  $L112
	.align 4

$L115:
#ifndef TRMMKERNEL
	and	K, 3, L
#else
	and	TMP1, 3, L
#endif
	ldt	alpha, ALPHA
#ifndef TRMMKERNEL
	LD	a2,  0 * SIZE(C1)
#endif
	ble	L, $L118
	.align	4

$L116:
	ADD	c01, t1, c01
	MUL	a1,  b1, t1
	LD	a1,  1 * SIZE(AO)
	LD	b1,  1 * SIZE(BO)

	lda	L,        -1(L)
	lda	AO,  1 * SIZE(AO)
	lda	BO,  1 * SIZE(BO)
	bgt	L,  $L116
	.align 4

$L118:
	ADD	c01, t1, c01
	ADD	c02, t2, c02
	ADD	c03, t3, c03
	ADD	c04, t4, c04

	ADD	c01, c02, c01
	ADD	c03, c04, c03
	ADD	c01, c03, c01

	MUL	alpha, c01, c01
#ifndef TRMMKERNEL
	ADD	c01,  a2, c01
#endif
	ST	c01,  0 * SIZE(C1)
	.align 4

$L999:
	ldt	$f2,   0($sp)
	ldt	$f3,   8($sp)
	ldt	$f4,  16($sp)
	ldt	$f5,  24($sp)
	ldt	$f6,  32($sp)
	ldt	$f7,  40($sp)
	ldt	$f8,  48($sp)
	ldt	$f9,  56($sp)
	clr	$0
	lda	$sp, STACKSIZE($sp)
	ret
	EPILOGUE