Add SPARC implementation of ?sum

as trivial copy of ?asum with the fabs replaced by fmov to preserve code structure

Author: martin-frbg

kernel/sparc/sum.S

@@ -0,0 +1,325 @@
+/*********************************************************************/
+/* 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"
+
+#define N	%i0
+#define X	%i1
+#define INCX	%i2
+#define I	%i3
+
+#ifdef DOUBLE
+#define c1	%f0
+#define c2	%f2
+#define t1	%f8
+#define t2	%f10
+#define t3	%f12
+#define t4	%f14
+
+#define a1	%f16
+#define a2	%f18
+#define a3	%f20
+#define a4	%f22
+#define a5	%f24
+#define a6	%f26
+#define a7	%f28
+#define a8	%f30
+#else
+#define c1	%f0
+#define c2	%f1
+#define t1	%f4
+#define t2	%f5
+#define t3	%f6
+#define t4	%f7
+
+#define a1	%f8
+#define a2	%f9
+#define a3	%f10
+#define a4	%f11
+#define a5	%f12
+#define a6	%f13
+#define a7	%f14
+#define a8	%f15
+#endif
+
+	PROLOGUE
+	SAVESP
+
+	FCLR(0)
+
+	sll	INCX, BASE_SHIFT, INCX
+
+	FMOV	c1, c2
+	FMOV	c1, t1
+	FMOV	c1, t2
+	FMOV	c1, t3
+	FMOV	c1, t4
+
+	cmp	INCX, 0
+	ble	.LL19
+	cmp	INCX, SIZE
+	bne	.LL50
+
+	sra	N, 3, I
+	cmp	I, 0
+	ble,pn	%icc, .LL15
+	nop
+
+	LDF	[X +  0 * SIZE], a1
+	add	I, -1, I
+	LDF	[X +  1 * SIZE], a2
+	cmp	I, 0
+	LDF	[X +  2 * SIZE], a3
+	LDF	[X +  3 * SIZE], a4
+	LDF	[X +  4 * SIZE], a5
+	LDF	[X +  5 * SIZE], a6
+	LDF	[X +  6 * SIZE], a7
+	LDF	[X +  7 * SIZE], a8
+
+	ble,pt	%icc, .LL12
+	add	X, 8 * SIZE, X
+
+#define PREFETCHSIZE 128
+
+.LL11:
+	FADD	c1, t1, c1
+	prefetch [X  + PREFETCHSIZE * SIZE], 0
+	FMOV	a1, t1
+	LDF	[X +  0 * SIZE], a1
+
+	FADD	c2, t2, c2
+	add	I, -1, I
+	FMOV	a2, t2
+	LDF	[X +  1 * SIZE], a2
+
+	FADD	c1, t3, c1
+	cmp	I, 0
+	FMOV	a3, t3
+	LDF	[X +  2 * SIZE], a3
+
+	FADD	c2, t4, c2
+	nop
+	FMOV	a4, t4
+	LDF	[X +  3 * SIZE], a4
+
+	FADD	c1, t1, c1
+	nop
+	FMOV	a5, t1
+	LDF	[X +  4 * SIZE], a5
+
+	FADD	c2, t2, c2
+	nop
+	FMOV	a6, t2
+	LDF	[X +  5 * SIZE], a6
+
+	FADD	c1, t3, c1
+	FMOV	a7, t3
+	LDF	[X +  6 * SIZE], a7
+	add	X, 8 * SIZE, X
+
+	FADD	c2, t4, c2
+	FMOV	a8, t4
+	bg,pt	%icc, .LL11
+	LDF	[X -  1 * SIZE], a8
+
+.LL12:
+	FADD	c1, t1, c1
+	FMOV	a1, t1
+	FADD	c2, t2, c2
+	FMOV	a2, t2
+
+	FADD	c1, t3, c1
+	FMOV	a3, t3
+	FADD	c2, t4, c2
+	FMOV	a4, t4
+
+	FADD	c1, t1, c1
+	FMOV	a5, t1
+	FADD	c2, t2, c2
+	FMOV	a6, t2
+
+	FADD	c1, t3, c1
+	FMOV	a7, t3
+	FADD	c2, t4, c2
+	FMOV	a8, t4
+
+.LL15:
+	and	N, 7, I
+	cmp	I,  0
+	ble,a,pn %icc, .LL19
+	nop
+
+.LL16:
+	LDF	[X +  0 * SIZE], a1
+	add	I, -1, I
+	cmp	I, 0
+	FADD	c1, t1, c1
+	FMOV	a1, t1
+	bg,pt	%icc, .LL16
+	add	X, 1 * SIZE, X
+
+.LL19:
+	FADD	c1, t1, c1
+	FADD	c2, t2, c2
+	FADD	c1, t3, c1
+	FADD	c2, t4, c2
+
+	FADD	c1, c2, c1
+	return	%i7 + 8
+	clr	%g0
+
+.LL50:
+	sra	N, 3, I
+	cmp	I, 0
+	ble,pn	%icc, .LL55
+	nop
+
+	LDF	[X +  0 * SIZE], a1
+	add	X, INCX, X
+	LDF	[X +  0 * SIZE], a2
+	add	X, INCX, X
+	LDF	[X +  0 * SIZE], a3
+	add	X, INCX, X
+	LDF	[X +  0 * SIZE], a4
+	add	X, INCX, X
+	LDF	[X +  0 * SIZE], a5
+	add	X, INCX, X
+	LDF	[X +  0 * SIZE], a6
+	add	X, INCX, X
+	add	I, -1, I
+	LDF	[X +  0 * SIZE], a7
+	cmp	I, 0
+	add	X, INCX, X
+	LDF	[X +  0 * SIZE], a8
+
+	ble,pt	%icc, .LL52
+	add	X, INCX, X
+
+.LL51:
+	FADD	c1, t1, c1
+	add	I, -1, I
+	FMOV	a1, t1
+	LDF	[X +  0 * SIZE], a1
+	add	X, INCX, X
+
+	FADD	c2, t2, c2
+	cmp	I, 0
+	FMOV	a2, t2
+	LDF	[X +  0 * SIZE], a2
+	add	X, INCX, X
+
+	FADD	c1, t3, c1
+	FMOV	a3, t3
+	LDF	[X +  0 * SIZE], a3
+	add	X, INCX, X
+
+	FADD	c2, t4, c2
+	FMOV	a4, t4
+	LDF	[X +  0 * SIZE], a4
+	add	X, INCX, X
+
+	FADD	c1, t1, c1
+	FMOV	a5, t1
+	LDF	[X +  0 * SIZE], a5
+	add	X, INCX, X
+
+	FADD	c2, t2, c2
+	FMOV	a6, t2
+	LDF	[X +  0 * SIZE], a6
+	add	X, INCX, X
+
+	FADD	c1, t3, c1
+	FMOV	a7, t3
+	LDF	[X +  0 * SIZE], a7
+	add	X, INCX, X
+
+	FADD	c2, t4, c2
+	FMOV	a8, t4
+	LDF	[X +  0 * SIZE], a8
+
+	bg,pt	%icc, .LL51
+	add	X, INCX, X
+
+.LL52:
+	FADD	c1, t1, c1
+	FMOV	a1, t1
+	FADD	c2, t2, c2
+	FMOV	a2, t2
+
+	FADD	c1, t3, c1
+	FMOV	a3, t3
+	FADD	c2, t4, c2
+	FMOV	a4, t4
+
+	FADD	c1, t1, c1
+	FMOV	a5, t1
+	FADD	c2, t2, c2
+	FMOV	a6, t2
+
+	FADD	c1, t3, c1
+	FMOV	a7, t3
+	FADD	c2, t4, c2
+	FMOV	a8, t4
+
+.LL55:
+	and	N, 7, I
+	cmp	I,  0
+	ble,a,pn %icc, .LL59
+	nop
+
+.LL56:
+	LDF	[X +  0 * SIZE], a1
+	FADD	c1, t1, c1
+	add	I, -1, I
+	FMOV	a1, t1
+	cmp	I, 0
+	bg,pt	%icc, .LL56
+	add	X, INCX, X
+
+.LL59:
+	FADD	c1, t1, c1
+	FADD	c2, t2, c2
+	FADD	c1, t3, c1
+	FADD	c2, t4, c2
+
+	FADD	c1, c2, c1
+	return	%i7 + 8
+	clr	%o0
+
+	EPILOGUE