📄 zaxpy_sse2.s

📁 Optimized GotoBLAS libraries
💻 S
📖 第 1 页 / 共 2 页
字号:
12 下一页
/*********************************************************************//*                                                                   *//*             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 STACK	16#define ARGS	 0	#define STACK_M	 4 + STACK + ARGS(%esp)#define ALPHA_R	16 + STACK + ARGS(%esp)#define ALPHA_I	24 + STACK + ARGS(%esp)#define STACK_X	32 + STACK + ARGS(%esp)#define STACK_INCX	36 + STACK + ARGS(%esp)#define STACK_Y	40 + STACK + ARGS(%esp)#define STACK_INCY	44 + STACK + ARGS(%esp)#ifdef PENTIUM4#define PREFETCHSIZE	220#endif#ifdef OPTERON#define PREFETCHSIZE	 88#define movsd	movlpd#endif#define M	%ebx#define X	%esi#define	INCX	%ecx#define Y	%edi#define INCY	%edx#define YY	%ebp	PROLOGUE	pushl	%ebp	pushl	%edi	pushl	%esi	pushl	%ebx	PROFCODE	movl	STACK_M,    M	movl	STACK_X,    X	movl	STACK_INCX, INCX	movl	STACK_Y,    Y	movl	STACK_INCY, INCY#ifndef CONJ	movsd	ALPHA_R, %xmm6	movsd	ALPHA_I, %xmm5	pxor	%xmm7, %xmm7	subsd	%xmm5, %xmm7	unpcklpd %xmm6, %xmm6	unpcklpd %xmm5, %xmm7#else	movsd	ALPHA_R, %xmm6	movsd	ALPHA_I, %xmm7	pxor	%xmm5, %xmm5	subsd	%xmm6, %xmm5	unpcklpd %xmm5, %xmm6	unpcklpd %xmm7, %xmm7#endif	leal	(, INCX, SIZE), INCX	leal	(, INCY, SIZE), INCY	addl	INCX, INCX	addl	INCY, INCY	cmpl	$2 * SIZE, INCX	jne	.L50	cmpl	$2 * SIZE, INCY	jne	.L50	testl	$SIZE, Y	jne	.L30	testl	$SIZE, X	jne	.L20	movl	M, %eax	sarl	$3,   %eax	jle	.L15	ALIGN_3.L12:#ifdef OPTERON	prefetcht0     (PREFETCHSIZE + 0) * SIZE(X)	prefetchw      (PREFETCHSIZE + 0) * SIZE(Y)#endif	movapd	 0 * SIZE(X),  %xmm0	movapd	 2 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 0 * SIZE(Y), %xmm4	movapd	 2 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   0 * SIZE(Y)	movapd	%xmm5,   2 * SIZE(Y)	movapd	 4 * SIZE(X),  %xmm0	movapd	 6 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 4 * SIZE(Y), %xmm4	movapd	 6 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   4 * SIZE(Y)	movapd	%xmm5,   6 * SIZE(Y)#ifdef OPTERON	prefetcht0     (PREFETCHSIZE + 8) * SIZE(X)	prefetchw      (PREFETCHSIZE + 8) * SIZE(Y)#endif	movapd	 8 * SIZE(X),  %xmm0	movapd	10 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 8 * SIZE(Y), %xmm4	movapd	10 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   8 * SIZE(Y)	movapd	%xmm5,  10 * SIZE(Y)	movapd	12 * SIZE(X),  %xmm0	movapd	14 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	12 * SIZE(Y), %xmm4	movapd	14 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,  12 * SIZE(Y)	movapd	%xmm5,  14 * SIZE(Y)#ifdef PENTIUM4	prefetcht0     (PREFETCHSIZE + 0) * SIZE(Y)#endif	addl	$16 * SIZE, X	addl	$16 * SIZE, Y	decl	%eax	jg	.L12	ALIGN_3.L15:	testl	$4, M	jle	.L16	movapd	 0 * SIZE(X),  %xmm0	movapd	 2 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 0 * SIZE(Y), %xmm4	movapd	 2 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   0 * SIZE(Y)	movapd	%xmm5,   2 * SIZE(Y)	movapd	 4 * SIZE(X),  %xmm0	movapd	 6 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 4 * SIZE(Y), %xmm4	movapd	 6 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   4 * SIZE(Y)	movapd	%xmm5,   6 * SIZE(Y)	addl	$8 * SIZE, X	addl	$8 * SIZE, Y	ALIGN_3.L16:	testl	$2, M	jle	.L17	movapd	 0 * SIZE(X),  %xmm0	movapd	 2 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 0 * SIZE(Y), %xmm4	movapd	 2 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   0 * SIZE(Y)	movapd	%xmm5,   2 * SIZE(Y)	addl	$4 * SIZE, X	addl	$4 * SIZE, Y	ALIGN_3.L17:	testl	$1, M	jle	.L99	movapd	 0 * SIZE(X),  %xmm0	pshufd	 $0x4e, %xmm0, %xmm1	movapd	 0 * SIZE(Y), %xmm4	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	movapd	%xmm4,   0 * SIZE(Y)	jmp	.L99	ALIGN_3.L20:	movl	M, %eax	sarl	$3,   %eax	jle	.L25	ALIGN_3.L22:#ifdef OPTERON	prefetcht0     (PREFETCHSIZE + 0) * SIZE(X)	prefetchw      (PREFETCHSIZE + 0) * SIZE(Y)#endif	movsd	 0 * SIZE(X),  %xmm0	movhpd	 1 * SIZE(X),  %xmm0	movsd	 2 * SIZE(X),  %xmm2	movhpd	 3 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 0 * SIZE(Y), %xmm4	movapd	 2 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   0 * SIZE(Y)	movapd	%xmm5,   2 * SIZE(Y)	movsd	 4 * SIZE(X),  %xmm0	movhpd	 5 * SIZE(X),  %xmm0	movsd	 6 * SIZE(X),  %xmm2	movhpd	 7 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 4 * SIZE(Y), %xmm4	movapd	 6 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   4 * SIZE(Y)	movapd	%xmm5,   6 * SIZE(Y)#ifdef OPTERON	prefetcht0     (PREFETCHSIZE + 8) * SIZE(X)	prefetchw      (PREFETCHSIZE + 8) * SIZE(Y)#endif	movsd	 8 * SIZE(X),  %xmm0	movhpd	 9 * SIZE(X),  %xmm0	movsd	10 * SIZE(X),  %xmm2	movhpd	11 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 8 * SIZE(Y), %xmm4	movapd	10 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   8 * SIZE(Y)	movapd	%xmm5,  10 * SIZE(Y)	movsd	12 * SIZE(X),  %xmm0	movhpd	13 * SIZE(X),  %xmm0	movsd	14 * SIZE(X),  %xmm2	movhpd	15 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	12 * SIZE(Y), %xmm4	movapd	14 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,  12 * SIZE(Y)	movapd	%xmm5,  14 * SIZE(Y)#ifdef PENTIUM4	prefetcht0     (PREFETCHSIZE + 0) * SIZE(Y)#endif	addl	$16 * SIZE, X	addl	$16 * SIZE, Y	decl	%eax	jg	.L22	ALIGN_3.L25:	testl	$4, M	jle	.L26	movsd	 0 * SIZE(X),  %xmm0	movhpd	 1 * SIZE(X),  %xmm0	movsd	 2 * SIZE(X),  %xmm2	movhpd	 3 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 0 * SIZE(Y), %xmm4	movapd	 2 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   0 * SIZE(Y)	movapd	%xmm5,   2 * SIZE(Y)	movsd	 4 * SIZE(X),  %xmm0	movhpd	 5 * SIZE(X),  %xmm0	movsd	 6 * SIZE(X),  %xmm2	movhpd	 7 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 4 * SIZE(Y), %xmm4	movapd	 6 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   4 * SIZE(Y)	movapd	%xmm5,   6 * SIZE(Y)	addl	$8 * SIZE, X	addl	$8 * SIZE, Y	ALIGN_3.L26:	testl	$2, M	jle	.L27	movsd	 0 * SIZE(X),  %xmm0	movhpd	 1 * SIZE(X),  %xmm0	movsd	 2 * SIZE(X),  %xmm2	movhpd	 3 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movapd	 0 * SIZE(Y), %xmm4	movapd	 2 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movapd	%xmm4,   0 * SIZE(Y)	movapd	%xmm5,   2 * SIZE(Y)	addl	$4 * SIZE, X	addl	$4 * SIZE, Y	ALIGN_3.L27:	testl	$1, M	jle	.L99	movsd	 0 * SIZE(X),  %xmm0	movhpd	 1 * SIZE(X),  %xmm0	pshufd	 $0x4e, %xmm0, %xmm1	movapd	 0 * SIZE(Y), %xmm4	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	movapd	%xmm4,   0 * SIZE(Y)	jmp	.L99	ALIGN_3.L30:	testl	$SIZE, X	jne	.L40	movl	M, %eax	sarl	$3,   %eax	jle	.L35	ALIGN_3.L32:#ifdef OPTERON	prefetcht0     (PREFETCHSIZE + 0) * SIZE(X)	prefetchw      (PREFETCHSIZE + 0) * SIZE(Y)#endif	movapd	 0 * SIZE(X),  %xmm0	movapd	 2 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movsd	 0 * SIZE(Y), %xmm4	movhpd	 1 * SIZE(Y), %xmm4	movsd	 2 * SIZE(Y), %xmm5	movhpd	 3 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3	addpd	%xmm0, %xmm4	addpd	%xmm1, %xmm4	addpd	%xmm2, %xmm5	addpd	%xmm3, %xmm5	movsd	%xmm4,   0 * SIZE(Y)	movhpd	%xmm4,   1 * SIZE(Y)	movsd	%xmm5,   2 * SIZE(Y)	movhpd	%xmm5,   3 * SIZE(Y)	movapd	 4 * SIZE(X),  %xmm0	movapd	 6 * SIZE(X),  %xmm2	pshufd	 $0x4e, %xmm0, %xmm1	pshufd	 $0x4e, %xmm2, %xmm3		movsd	 4 * SIZE(Y), %xmm4	movhpd	 5 * SIZE(Y), %xmm4	movsd	 6 * SIZE(Y), %xmm5	movhpd	 7 * SIZE(Y), %xmm5	mulpd	%xmm6, %xmm0	mulpd	%xmm7, %xmm1	mulpd	%xmm6, %xmm2	mulpd	%xmm7, %xmm3
12 下一页
⌨️ 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 + -