📄 xgemv_n.s

📁 Optimized GotoBLAS libraries
💻 S
字号:
/*********************************************************************//*                                                                   *//*             Optimized BLAS libraries                              *//*                     By Kazushige Goto <kgoto@tacc.utexas.edu>     *//*                                                                   *//* Copyright (c) The University of Texas, 2005. All rights reserved. *//* UNIVERSITY EXPRESSLY DISCLAIMS ANY AND ALL WARRANTIES CONCERNING  *//* THIS SOFTWARE AND DOCUMENTATION, INCLUDING ANY WARRANTIES OF      *//* MERCHANTABILITY, FITNESS FOR ANY PARTICULAR PURPOSE,              *//* NON-INFRINGEMENT AND WARRANTIES OF PERFORMANCE, AND ANY WARRANTY  *//* THAT MIGHT OTHERWISE ARISE FROM COURSE OF DEALING OR USAGE OF     *//* TRADE. NO WARRANTY IS EITHER EXPRESS OR IMPLIED WITH RESPECT TO   *//* THE USE OF THE SOFTWARE OR DOCUMENTATION.                         *//* Under no circumstances shall University be liable for incidental, *//* special, indirect, direct or consequential damages or loss of     *//* profits, interruption of business, or related expenses which may  *//* arise from use of Software or Documentation, including but not    *//* limited to those resulting from defects in Software and/or        *//* Documentation, or loss or inaccuracy of data of any kind.         *//*********************************************************************/#define ASSEMBLER#include "common.h"#define P		32	#define STACKSIZE	80	#define ALPHA_R		 8 + STACKSIZE(%rsp)#define ALPHA_I		24 + STACKSIZE(%rsp)#define OLD_INCX	40 + STACKSIZE(%rsp)#define OLD_Y		48 + STACKSIZE(%rsp)#define OLD_INCY	56 + STACKSIZE(%rsp)#define BUFFER		64 + STACKSIZE(%rsp)#define PLDA_M	        56            (%rsp)#define IS		64	      (%rsp)#define M	  %rdi#define N	  %rsi#define A	  %rcx#define LDA	  %r8#define X	  %r9#define INCX	  %rdx#define Y	  %rbp#define INCY	  %r10#define TEMP	%rax#define I	%rax#define J	%r11#define A1	%r12#define X1	%r13#define Y1	%r14#define XP	%r15#define MIN_N	%rbx		PROLOGUE	PROFCODE	subq	$STACKSIZE, %rsp	movq	%rbx,  0(%rsp)	movq	%rbp,  8(%rsp)	movq	%r12, 16(%rsp)	movq	%r13, 24(%rsp)	movq	%r14, 32(%rsp)	movq	%r15, 40(%rsp)	movq	OLD_INCX,   INCX	movq	OLD_Y,      Y	movq	OLD_INCY,   INCY	FLD	ALPHA_I	FLD	ALPHA_R	salq	$ZBASE_SHIFT, INCX	salq	$ZBASE_SHIFT, INCY	movq	$0, IS	test	M, M	jle	.L79	test	N, N	jle	.L79	movq	LDA, %rax	imulq	$P,  %rax		# P * lda	subq	M   ,%rax		# P * lda - m	salq	$ZBASE_SHIFT, %rax	movq	%rax, PLDA_M	salq	$ZBASE_SHIFT, LDA	ALIGN_2.L32:	movq	$P,  %rax	movq	N,    MIN_N	subq	IS,   MIN_N	cmpq	%rax, MIN_N	cmovg	%rax, MIN_N	movq	IS, XP	salq	$ZBASE_SHIFT, XP	leaq	(X,XP, 1), XP	cmpq	$2 * SIZE, INCX	je	.L34	movq	BUFFER, XP	movq	XP,     X1	movq	MIN_N, I	sarq	$1,    I	jle	.L35	ALIGN_2.L36:	FLD	0 * SIZE(X)	FLD	1 * SIZE(X)	addq	INCX,X		# x += incx	FLD	0 * SIZE(X)	FLD	1 * SIZE(X)	addq	INCX,X		# x += incx	FST	3 * SIZE(X1)	FST	2 * SIZE(X1)	FST	1 * SIZE(X1)	FST	0 * SIZE(X1)	addq	$4 * SIZE, X1		# xp += 4	decq	I	jg	.L36	ALIGN_3.L35:	movq	MIN_N, I	andq	$1,    I	jle	.L34	FLD	0 * SIZE(X)	FLD	1 * SIZE(X)	addq	INCX,X		# x += incx	FST	1 * SIZE(X1)	FST	0 * SIZE(X1)	ALIGN_3/* Main Routine */.L34:	movq	 Y, Y1		# c_offset	movq	 M, J		# j = m	ALIGN_3.L61:	movq	A, A1			# a_offset = a	addq	$2 * SIZE, A		# a++	fldz	fldz	fldz	fldz	movq	XP, X1	FLD	(X1)			#  bt1 = *(b_offset + 0)	movq	MIN_N, I	sarq	$1,    I	jle	.L64	ALIGN_3.L65:	FLD	 0 * SIZE(A1)		# at1  = *(a_offset + 0)	fmul	%st(1)			# at1 *= bt1	faddp	%st, %st(2)		# ct1 += at1	FLD	 1 * SIZE(A1)		# bt1 *= *(a_offset + 1)	fmulp	 %st, %st(1)#ifndef CONJ	faddp	%st, %st(2)		# ct2 += bt1#else	fsubrp	%st, %st(2)		# ct2 -= bt1#endif	FLD	 1 * SIZE(X1)		# bt1  = *(b_offset + 1)	FLD	 0 * SIZE(A1)		# at1  = *(a_offset + 0)	fmul	%st(1)			# at1 *= bt1	faddp	%st, %st(4)		# ct3 += at1	FLD	 1 * SIZE(A1)		# bt1 *= *(a_offset + 1)	fmulp	 %st, %st(1)	faddp	%st, %st(4)		# ct4 += bt1	FLD	 2 * SIZE(X1)		# bt1  = *(b_offset + 2)	addq	$2 * SIZE, X1		# b_offset += 2	addq	LDA,  A1		# a_offset += lda	FLD	 0 * SIZE(A1)		# at1  = *(a_offset + 0)	fmul	%st(1)			# at1 *= bt1	faddp	%st, %st(2)		# ct1 += at1	FLD	 1 * SIZE(A1)		# bt1 *= *(a_offset + 1)	fmulp	 %st, %st(1)#ifndef CONJ	faddp	%st, %st(2)		# ct2 += bt1#else	fsubrp	%st, %st(2)		# ct2 -= bt1#endif	FLD	 1 * SIZE(X1)		# bt1  = *(b_offset + 1)	FLD	 0 * SIZE(A1)		# at1  = *(a_offset + 0)	fmul	%st(1)			# at1 *= bt1	faddp	%st, %st(4)		# ct3 += at1	FLD	 1 * SIZE(A1)		# bt1 *= *(a_offset + 1)	fmulp	 %st, %st(1)	faddp	%st, %st(4)		# ct4 += bt1	FLD	 2 * SIZE(X1)		# bt1  = *(b_offset + 2)	addq	$2 * SIZE, X1		# b_offset += 2	addq	LDA,  A1		# a_offset += lda	decq	I	jg	.L65.L64:	movq	MIN_N, I	andq	$1, I	jle	.L70	ALIGN_2.L71:	FLD	 0 * SIZE(A1)		# at1  = *(a_offset + 0)	fmul	%st(1)			# at1 *= bt1	faddp	%st, %st(2)		# ct1 += at1	FLD	 1 * SIZE(A1)		# bt1 *= *(a_offset + 1)	fmulp	 %st, %st(1)#ifndef CONJ	faddp	%st, %st(2)		# ct2 += bt1#else	fsubrp	%st, %st(2)		# ct2 -= bt1#endif	FLD	 1 * SIZE(X1)		# bt1  = *(b_offset + 1)	FLD	 0 * SIZE(A1)		# at1  = *(a_offset + 0)	fmul	%st(1)			# at1 *= bt1	faddp	%st, %st(4)		# ct3 += at1	FLD	 1 * SIZE(A1)		# bt1 *= *(a_offset + 1)	fmulp	 %st, %st(1)	faddp	%st, %st(4)		# ct4 += bt1	fldz	ALIGN_2.L70:	ffreep	%st(0)#ifndef XCONJ#ifndef CONJ	fsubp	%st, %st(3)	faddp	%st, %st(1)#else	faddp	%st, %st(3)	faddp	%st, %st(1)#endif#else#ifndef CONJ	faddp	%st, %st(3)	fsubp	%st, %st(1)#else	fsubp	%st, %st(3)	fsubp	%st, %st(1)#endif#endif	fld	%st(0)		# ct4 = ct2	fmul	%st(4)	fld	%st(2)	fmul	%st(4)	fsubp	%st, %st(1)	FLD	0 * SIZE(Y1)	faddp	 %st, %st(1)	FST	0 * SIZE(Y1)	fmul	%st(2)	fxch	%st(1)	fmul	%st(3)	faddp	%st, %st(1)	FLD	1 * SIZE(Y1)	faddp	 %st, %st(1)	FST	1 * SIZE(Y1)	addq	INCY, Y1	decq	J	jg	.L61.L60:	addq	PLDA_M, A	addq	$P, IS	cmpq	N,  IS	jl	.L32.L79:	ffreep	%st	ffreep	%st	movq	  0(%rsp), %rbx	movq	  8(%rsp), %rbp	movq	 16(%rsp), %r12	movq	 24(%rsp), %r13	movq	 32(%rsp), %r14	movq	 40(%rsp), %r15	addq	$STACKSIZE, %rsp	ret	EPILOGUE
⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -