📄 zrot_sse.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	12#define ARGS     0#define STACK_N		 4 + STACK + ARGS(%esp)#define STACK_X		 8 + STACK + ARGS(%esp)#define STACK_INCX	12 + STACK + ARGS(%esp)#define STACK_Y		16 + STACK + ARGS(%esp)#define STACK_INCY	20 + STACK + ARGS(%esp)#define STACK_C		24 + STACK + ARGS(%esp)#define STACK_S		28 + STACK + ARGS(%esp)#define N	%ebx#define X	%esi#define INCX	%ecx#define Y	%edi#define INCY	%edx#define	I	%eax#undef MOVSD#if defined(CORE2) || defined(PENRYN)#define MOVSD		movsd#endif#ifdef PENTIUM4#define PREFETCH	prefetcht0#define PREFETCH_SIZE	144#define MOVSD		movsd#endif#ifdef OPTERON#define PREFETCH	prefetchw#define PREFETCH_SIZE	144#define MOVSD		movlps#endif#ifndef HAVE_SSE2#define movsd	movlps#define MOVSD	movlps#endif#define C	%xmm6#define S	%xmm7	PROLOGUE	pushl	%edi	pushl	%esi	pushl	%ebx	PROFCODE	movl	STACK_N,     N	movl	STACK_X,     X	movl	STACK_INCX,  INCX	movl	STACK_Y,     Y	movl	STACK_INCY,  INCY	sall	$ZBASE_SHIFT, INCX	sall	$ZBASE_SHIFT, INCY	movss	STACK_C, C	movss	STACK_S, S	shufps	$0x0, C, C	shufps	$0x0, S, S	cmpl	$0, N	jle	.L999	cmpl	$2 * SIZE, INCX	jne	.L50	cmpl	$2 * SIZE, INCY	jne	.L50	testl	$2 * SIZE, X	je	.L10#ifndef HAVE_SSE2	xorps	%xmm0, %xmm0	xorps	%xmm1, %xmm1#endif	movsd	0 * SIZE(Y), %xmm1	movsd	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movlps	%xmm0, 0 * SIZE(X)	movlps	%xmm2, 0 * SIZE(Y)	addl	$2 * SIZE, X	addl	$2 * SIZE, Y	decl	N	jle	.L999.L10:	testl	$1 * SIZE, X	jne	.L30	testl	$3 * SIZE, Y	jne	.L20	movl	N,  I	sarl	$4, I	jle	.L14	ALIGN_3.L11:#ifdef PENTIUM4	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(X)#endif#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(X)#endif	movaps	0 * SIZE(Y), %xmm1	movaps	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 0 * SIZE(X)	movaps	%xmm2, 0 * SIZE(Y)	movaps	4 * SIZE(Y), %xmm1	movaps	4 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 4 * SIZE(X)	movaps	%xmm2, 4 * SIZE(Y)#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(Y)#endif	movaps	8 * SIZE(Y), %xmm1	movaps	8 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 8 * SIZE(X)	movaps	%xmm2, 8 * SIZE(Y)	movaps	12 * SIZE(Y), %xmm1	movaps	12 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 12 * SIZE(X)	movaps	%xmm2, 12 * SIZE(Y)#ifdef PENTIUM4	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(Y)#endif#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 16) * SIZE(X)#endif	movaps	16 * SIZE(Y), %xmm1	movaps	16 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 16 * SIZE(X)	movaps	%xmm2, 16 * SIZE(Y)	movaps	20 * SIZE(Y), %xmm1	movaps	20 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 20 * SIZE(X)	movaps	%xmm2, 20 * SIZE(Y)#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 16) * SIZE(Y)#endif	movaps	24 * SIZE(Y), %xmm1	movaps	24 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 24 * SIZE(X)	movaps	%xmm2, 24 * SIZE(Y)	movaps	28 * SIZE(Y), %xmm1	movaps	28 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 28 * SIZE(X)	movaps	%xmm2, 28 * SIZE(Y)	addl	$32 * SIZE, X	addl	$32 * SIZE, Y	decl	I	jg	.L11	ALIGN_3.L14:	testl	$15, N	jle	.L999	testl	$8, N	jle	.L15	movaps	0 * SIZE(Y), %xmm1	movaps	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 0 * SIZE(X)	movaps	%xmm2, 0 * SIZE(Y)	movaps	4 * SIZE(Y), %xmm1	movaps	4 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 4 * SIZE(X)	movaps	%xmm2, 4 * SIZE(Y)	movaps	8 * SIZE(Y), %xmm1	movaps	8 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 8 * SIZE(X)	movaps	%xmm2, 8 * SIZE(Y)	movaps	12 * SIZE(Y), %xmm1	movaps	12 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 12 * SIZE(X)	movaps	%xmm2, 12 * SIZE(Y)	addl	$16 * SIZE, X	addl	$16 * SIZE, Y	ALIGN_3.L15:	testl	$4, N	jle	.L16	movaps	0 * SIZE(Y), %xmm1	movaps	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 0 * SIZE(X)	movaps	%xmm2, 0 * SIZE(Y)	movaps	4 * SIZE(Y), %xmm1	movaps	4 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 4 * SIZE(X)	movaps	%xmm2, 4 * SIZE(Y)	addl	$8 * SIZE, X	addl	$8 * SIZE, Y	ALIGN_3.L16:	testl	$2, N	jle	.L17	movaps	0 * SIZE(Y), %xmm1	movaps	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 0 * SIZE(X)	movaps	%xmm2, 0 * SIZE(Y)	addl	$4 * SIZE, X	addl	$4 * SIZE, Y	ALIGN_3.L17:	testl	$1, N	jle	.L999#ifndef HAVE_SSE2	xorps	%xmm0, %xmm0	xorps	%xmm1, %xmm1#endif	movsd	0 * SIZE(Y), %xmm1	movsd	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movlps	%xmm0, 0 * SIZE(X)	movlps	%xmm2, 0 * SIZE(Y)	jmp	.L999	ALIGN_3.L20:	movl	N,  I	sarl	$4, I	jle	.L24	ALIGN_3.L21:#ifdef PENTIUM4	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(X)#endif#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(X)#endif	MOVSD	0 * SIZE(Y), %xmm1	movhps	2 * SIZE(Y), %xmm1	movaps	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 0 * SIZE(X)	movlps	%xmm2, 0 * SIZE(Y)	movhps	%xmm2, 2 * SIZE(Y)	MOVSD	4 * SIZE(Y), %xmm1	movhps	6 * SIZE(Y), %xmm1	movaps	4 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 4 * SIZE(X)	movlps	%xmm2, 4 * SIZE(Y)	movhps	%xmm2, 6 * SIZE(Y)#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(Y)#endif	MOVSD	 8 * SIZE(Y), %xmm1	movhps	10 * SIZE(Y), %xmm1	movaps	 8 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0,  8 * SIZE(X)	movlps	%xmm2,  8 * SIZE(Y)	movhps	%xmm2, 10 * SIZE(Y)	MOVSD	12 * SIZE(Y), %xmm1	movhps	14 * SIZE(Y), %xmm1	movaps	12 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 12 * SIZE(X)	movlps	%xmm2, 12 * SIZE(Y)	movhps	%xmm2, 14 * SIZE(Y)#ifdef PENTIUM4	PREFETCH	(PREFETCH_SIZE + 0) * SIZE(Y)#endif#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 16) * SIZE(X)#endif	MOVSD	16 * SIZE(Y), %xmm1	movhps	18 * SIZE(Y), %xmm1	movaps	16 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 16 * SIZE(X)	movlps	%xmm2, 16 * SIZE(Y)	movhps	%xmm2, 18 * SIZE(Y)	MOVSD	20 * SIZE(Y), %xmm1	movhps	22 * SIZE(Y), %xmm1	movaps	20 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 20 * SIZE(X)	movlps	%xmm2, 20 * SIZE(Y)	movhps	%xmm2, 22 * SIZE(Y)#ifdef OPTERON	PREFETCH	(PREFETCH_SIZE + 16) * SIZE(Y)#endif	MOVSD	24 * SIZE(Y), %xmm1	movhps	26 * SIZE(Y), %xmm1	movaps	24 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 24 * SIZE(X)	movlps	%xmm2, 24 * SIZE(Y)	movhps	%xmm2, 26 * SIZE(Y)	MOVSD	28 * SIZE(Y), %xmm1	movhps	30 * SIZE(Y), %xmm1	movaps	28 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %xmm3	mulps	C, %xmm0	mulps	S, %xmm1	mulps	C, %xmm2	mulps	S, %xmm3	addps	%xmm1, %xmm0	subps	%xmm3, %xmm2	movaps	%xmm0, 28 * SIZE(X)	movlps	%xmm2, 28 * SIZE(Y)	movhps	%xmm2, 30 * SIZE(Y)	addl	$32 * SIZE, X	addl	$32 * SIZE, Y	decl	I	jg	.L21	ALIGN_3.L24:	testl	$15, N	jle	.L999	testl	$8, N	jle	.L25	MOVSD	0 * SIZE(Y), %xmm1	movhps	2 * SIZE(Y), %xmm1	movaps	0 * SIZE(X), %xmm0	movaps	%xmm1, %xmm2	movaps	%xmm0, %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 + -