2843 lines
52 KiB
ArmAsm
2843 lines
52 KiB
ArmAsm
/*********************************************************************/
|
|
/* 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 "l2param.h"
|
|
|
|
#if GEMV_UNROLL < 2
|
|
#undef GEMV_UNROLL
|
|
#define GEMV_UNROLL 2
|
|
#endif
|
|
|
|
#ifndef WINDOWS_ABI
|
|
|
|
#define STACKSIZE 64
|
|
|
|
#define OLD_M %rdi
|
|
#define OLD_N %rsi
|
|
#define OLD_A %rcx
|
|
#define OLD_LDA %r8
|
|
#define STACK_INCX 8 + STACKSIZE(%rsp)
|
|
#define STACK_Y 16 + STACKSIZE(%rsp)
|
|
#define STACK_INCY 24 + STACKSIZE(%rsp)
|
|
#define STACK_BUFFER 32 + STACKSIZE(%rsp)
|
|
#define ALPHA 48 (%rsp)
|
|
|
|
#else
|
|
|
|
#define STACKSIZE 256
|
|
|
|
#define OLD_M %rcx
|
|
#define OLD_N %rdx
|
|
#define OLD_A 40 + STACKSIZE(%rsp)
|
|
#define OLD_LDA 48 + STACKSIZE(%rsp)
|
|
#define OLD_X 56 + STACKSIZE(%rsp)
|
|
#define STACK_INCX 64 + STACKSIZE(%rsp)
|
|
#define STACK_Y 72 + STACKSIZE(%rsp)
|
|
#define STACK_INCY 80 + STACKSIZE(%rsp)
|
|
#define STACK_BUFFER 88 + STACKSIZE(%rsp)
|
|
#define ALPHA 224 (%rsp)
|
|
|
|
#endif
|
|
|
|
#define LDA %r8
|
|
#define X %r9
|
|
|
|
#define INCX %rsi
|
|
#define INCY %rdi
|
|
|
|
#define M %r10
|
|
#define N %r11
|
|
#define A %r12
|
|
#define Y %r14
|
|
#define BUFFER %r13
|
|
|
|
#define I %rax
|
|
#define A1 %rbx
|
|
#define A2 %rcx
|
|
#define LDA3 %rdx
|
|
#define Y1 %rbp
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
#define MM %r15
|
|
#else
|
|
#define MM M
|
|
#endif
|
|
|
|
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)
|
|
|
|
#ifdef WINDOWS_ABI
|
|
movq %rdi, 48(%rsp)
|
|
movq %rsi, 56(%rsp)
|
|
movups %xmm6, 64(%rsp)
|
|
movups %xmm7, 80(%rsp)
|
|
movups %xmm8, 96(%rsp)
|
|
movups %xmm9, 112(%rsp)
|
|
movups %xmm10, 128(%rsp)
|
|
movups %xmm11, 144(%rsp)
|
|
movups %xmm12, 160(%rsp)
|
|
movups %xmm13, 176(%rsp)
|
|
movups %xmm14, 192(%rsp)
|
|
movups %xmm15, 208(%rsp)
|
|
|
|
movq OLD_M, M
|
|
movq OLD_N, N
|
|
movq OLD_A, A
|
|
movq OLD_LDA, LDA
|
|
movq OLD_X, X
|
|
#else
|
|
movq OLD_M, M
|
|
movq OLD_N, N
|
|
movq OLD_A, A
|
|
movq OLD_LDA, LDA
|
|
#endif
|
|
|
|
movq STACK_INCX, INCX
|
|
movq STACK_Y, Y
|
|
movq STACK_INCY, INCY
|
|
movq STACK_BUFFER, BUFFER
|
|
|
|
#ifndef WINDOWS_ABI
|
|
movsd %xmm0, ALPHA
|
|
#else
|
|
movsd %xmm3, ALPHA
|
|
#endif
|
|
|
|
leaq -1(INCY), %rax
|
|
|
|
leaq (,INCX, SIZE), INCX
|
|
leaq (,INCY, SIZE), INCY
|
|
leaq (,LDA, SIZE), LDA
|
|
|
|
leaq (LDA, LDA, 2), LDA3
|
|
|
|
subq $-16 * SIZE, A
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
leaq -1 (M), MM
|
|
testq $SIZE, A
|
|
cmoveq M, MM
|
|
#endif
|
|
|
|
testq N, N # if n <= 0 goto END
|
|
jle .L999
|
|
testq M, M # if n <= 0 goto END
|
|
jle .L999
|
|
|
|
#if !defined(COPY_FORCE) && !defined(ALIGNED_ACCESS)
|
|
#ifndef NOCOPY_UNALIGNED
|
|
movq Y, Y1
|
|
andq $0xf, Y1
|
|
orq Y1, %rax
|
|
#endif
|
|
testq %rax, %rax
|
|
cmoveq Y, BUFFER
|
|
je .L10
|
|
#endif
|
|
|
|
movq BUFFER, Y1
|
|
|
|
pxor %xmm4, %xmm4
|
|
|
|
movq M, %rax
|
|
addq $16, %rax
|
|
sarq $4, %rax
|
|
ALIGN_3
|
|
|
|
.L01:
|
|
movapd %xmm4, 0 * SIZE(Y1)
|
|
movapd %xmm4, 2 * SIZE(Y1)
|
|
movapd %xmm4, 4 * SIZE(Y1)
|
|
movapd %xmm4, 6 * SIZE(Y1)
|
|
movapd %xmm4, 8 * SIZE(Y1)
|
|
movapd %xmm4, 10 * SIZE(Y1)
|
|
movapd %xmm4, 12 * SIZE(Y1)
|
|
movapd %xmm4, 14 * SIZE(Y1)
|
|
subq $-16 * SIZE, Y1
|
|
decq %rax
|
|
jg .L01
|
|
ALIGN_3
|
|
|
|
.L10:
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
leaq SIZE(BUFFER), %rax
|
|
testq $SIZE, A
|
|
cmovne %rax, BUFFER
|
|
|
|
testq $SIZE, LDA
|
|
jne .L50
|
|
#endif
|
|
|
|
#if GEMV_UNROLL >= 8
|
|
|
|
cmpq $8, N
|
|
jl .L20
|
|
ALIGN_3
|
|
|
|
.L11:
|
|
subq $8, N
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
leaq (A, LDA, 4), A2
|
|
leaq (A, LDA, 8), A
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm8
|
|
addq INCX, X
|
|
movddup (X), %xmm9
|
|
addq INCX, X
|
|
movddup (X), %xmm10
|
|
addq INCX, X
|
|
movddup (X), %xmm11
|
|
addq INCX, X
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
movddup (X), %xmm13
|
|
addq INCX, X
|
|
movddup (X), %xmm14
|
|
addq INCX, X
|
|
movddup (X), %xmm15
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm8
|
|
unpcklpd %xmm8, %xmm8
|
|
addq INCX, X
|
|
movsd (X), %xmm9
|
|
unpcklpd %xmm9, %xmm9
|
|
addq INCX, X
|
|
movsd (X), %xmm10
|
|
unpcklpd %xmm10, %xmm10
|
|
addq INCX, X
|
|
movsd (X), %xmm11
|
|
unpcklpd %xmm11, %xmm11
|
|
addq INCX, X
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
movsd (X), %xmm13
|
|
unpcklpd %xmm13, %xmm13
|
|
addq INCX, X
|
|
movsd (X), %xmm14
|
|
unpcklpd %xmm14, %xmm14
|
|
addq INCX, X
|
|
movsd (X), %xmm15
|
|
unpcklpd %xmm15, %xmm15
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm8
|
|
mulpd %xmm0, %xmm9
|
|
mulpd %xmm0, %xmm10
|
|
mulpd %xmm0, %xmm11
|
|
mulpd %xmm0, %xmm12
|
|
mulpd %xmm0, %xmm13
|
|
mulpd %xmm0, %xmm14
|
|
mulpd %xmm0, %xmm15
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
testq $SIZE, A
|
|
je .L1X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(A1, LDA), %xmm5
|
|
movsd -16 * SIZE(A1, LDA, 2), %xmm6
|
|
movsd -16 * SIZE(A1, LDA3), %xmm7
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm8, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
movsd -16 * SIZE(A2), %xmm4
|
|
mulsd %xmm9, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
movsd -16 * SIZE(A2, LDA), %xmm5
|
|
mulsd %xmm10, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
movsd -16 * SIZE(A2, LDA, 2), %xmm6
|
|
mulsd %xmm11, %xmm7
|
|
addsd %xmm7, %xmm0
|
|
movsd -16 * SIZE(A2, LDA3), %xmm7
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
mulsd %xmm14, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
mulsd %xmm15, %xmm7
|
|
addsd %xmm7, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, A2
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L1X:
|
|
#endif
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L15
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm5)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm6)
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm7)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
decq I
|
|
jle .L14
|
|
ALIGN_3
|
|
|
|
.L13:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm8, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm4)
|
|
mulpd %xmm8, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm8, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA, 1, %xmm6)
|
|
mulpd %xmm8, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA, 1, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A1, LDA, 1)
|
|
#endif
|
|
|
|
mulpd %xmm9, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 2, %xmm4)
|
|
mulpd %xmm9, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 2, %xmm5)
|
|
|
|
mulpd %xmm9, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA, 2, %xmm6)
|
|
mulpd %xmm9, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA, 2, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A1, LDA, 2)
|
|
#endif
|
|
|
|
mulpd %xmm10, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA3, 1, %xmm4)
|
|
mulpd %xmm10, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA3, 1, %xmm5)
|
|
|
|
mulpd %xmm10, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA3, 1, %xmm6)
|
|
mulpd %xmm10, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA3, 1, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A1, LDA3)
|
|
#endif
|
|
|
|
mulpd %xmm11, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
mulpd %xmm11, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
|
|
mulpd %xmm11, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm6)
|
|
mulpd %xmm11, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A2)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm4)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm12, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA, 1, %xmm6)
|
|
mulpd %xmm12, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A2, LDA, 1)
|
|
#endif
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 2, %xmm4)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 2, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA, 2, %xmm6)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA, 2, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A2, LDA, 2)
|
|
#endif
|
|
|
|
mulpd %xmm14, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA3, 1, %xmm4)
|
|
mulpd %xmm14, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA3, 1, %xmm5)
|
|
|
|
mulpd %xmm14, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA3, 1, %xmm6)
|
|
mulpd %xmm14, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA3, 1, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) - 128 + PREOFFSET(A2, LDA3)
|
|
#endif
|
|
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm4)
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm5)
|
|
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm6)
|
|
mulpd %xmm15, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A1( -2 * SIZE, A1, %xmm7)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm3)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L13
|
|
ALIGN_3
|
|
|
|
.L14:
|
|
mulpd %xmm8, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm4)
|
|
mulpd %xmm8, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm8, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA, 1, %xmm6)
|
|
mulpd %xmm8, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA, 1, %xmm7)
|
|
|
|
mulpd %xmm9, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 2, %xmm4)
|
|
mulpd %xmm9, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 2, %xmm5)
|
|
|
|
mulpd %xmm9, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA, 2, %xmm6)
|
|
mulpd %xmm9, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA, 2, %xmm7)
|
|
|
|
mulpd %xmm10, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA3, 1, %xmm4)
|
|
mulpd %xmm10, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA3, 1, %xmm5)
|
|
|
|
mulpd %xmm10, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA3, 1, %xmm6)
|
|
mulpd %xmm10, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA3, 1, %xmm7)
|
|
|
|
mulpd %xmm11, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
mulpd %xmm11, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
|
|
mulpd %xmm11, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm6)
|
|
mulpd %xmm11, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm7)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm4)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm12, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA, 1, %xmm6)
|
|
mulpd %xmm12, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 2, %xmm4)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 2, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA, 2, %xmm6)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA, 2, %xmm7)
|
|
|
|
mulpd %xmm14, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA3, 1, %xmm4)
|
|
mulpd %xmm14, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA3, 1, %xmm5)
|
|
|
|
mulpd %xmm14, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA3, 1, %xmm6)
|
|
mulpd %xmm14, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA3, 1, %xmm7)
|
|
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm2)
|
|
mulpd %xmm15, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L15:
|
|
testq $4, MM
|
|
je .L16
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm5)
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm6)
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 1, %xmm7)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
mulpd %xmm8, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 2, %xmm4)
|
|
mulpd %xmm8, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 2, %xmm5)
|
|
|
|
mulpd %xmm9, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA3, 1, %xmm6)
|
|
mulpd %xmm9, %xmm7
|
|
addpd %xmm7, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA3, 1, %xmm7)
|
|
|
|
mulpd %xmm10, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
mulpd %xmm10, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
|
|
mulpd %xmm11, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm6)
|
|
mulpd %xmm11, %xmm7
|
|
addpd %xmm7, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 2, %xmm4)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 2, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA3, 1, %xmm6)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm1
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA3, 1, %xmm7)
|
|
|
|
mulpd %xmm14, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
mulpd %xmm14, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
mulpd %xmm15, %xmm7
|
|
addpd %xmm7, %xmm1
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, A2
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L16:
|
|
testq $2, MM
|
|
je .L17
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm5)
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 2, %xmm6)
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA3, 1, %xmm7)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm8, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
mulpd %xmm9, %xmm5
|
|
addpd %xmm5, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm10, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 2, %xmm6)
|
|
mulpd %xmm11, %xmm7
|
|
addpd %xmm7, %xmm0
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA3, 1, %xmm7)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm0
|
|
mulpd %xmm14, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
mulpd %xmm15, %xmm7
|
|
addpd %xmm7, %xmm0
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, A2
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L17:
|
|
testq $1, MM
|
|
je .L18
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(A1, LDA), %xmm5
|
|
movsd -16 * SIZE(A1, LDA, 2), %xmm6
|
|
movsd -16 * SIZE(A1, LDA3), %xmm7
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm8, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
movsd -16 * SIZE(A2), %xmm4
|
|
mulsd %xmm9, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
movsd -16 * SIZE(A2, LDA), %xmm5
|
|
mulsd %xmm10, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
movsd -16 * SIZE(A2, LDA, 2), %xmm6
|
|
mulsd %xmm11, %xmm7
|
|
addsd %xmm7, %xmm0
|
|
movsd -16 * SIZE(A2, LDA3), %xmm7
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
mulsd %xmm14, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
mulsd %xmm15, %xmm7
|
|
addsd %xmm7, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
ALIGN_3
|
|
|
|
.L18:
|
|
cmpq $8, N
|
|
jge .L11
|
|
ALIGN_3
|
|
|
|
.L20:
|
|
#endif
|
|
|
|
#if GEMV_UNROLL >= 4
|
|
|
|
cmpq $4, N
|
|
jl .L30
|
|
|
|
#if GEMV_UNROLL == 4
|
|
ALIGN_3
|
|
|
|
.L21:
|
|
#endif
|
|
|
|
subq $4, N
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
leaq (A, LDA, 2), A2
|
|
leaq (A, LDA, 4), A
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
movddup (X), %xmm13
|
|
addq INCX, X
|
|
movddup (X), %xmm14
|
|
addq INCX, X
|
|
movddup (X), %xmm15
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
movsd (X), %xmm13
|
|
unpcklpd %xmm13, %xmm13
|
|
addq INCX, X
|
|
movsd (X), %xmm14
|
|
unpcklpd %xmm14, %xmm14
|
|
addq INCX, X
|
|
movsd (X), %xmm15
|
|
unpcklpd %xmm15, %xmm15
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm12
|
|
mulpd %xmm0, %xmm13
|
|
mulpd %xmm0, %xmm14
|
|
mulpd %xmm0, %xmm15
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
testq $SIZE, A
|
|
je .L2X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(A1, LDA), %xmm5
|
|
movsd -16 * SIZE(A2), %xmm6
|
|
movsd -16 * SIZE(A2, LDA), %xmm7
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
mulsd %xmm14, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
mulsd %xmm15, %xmm7
|
|
addsd %xmm7, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, A2
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L2X:
|
|
#endif
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L25
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm2)
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm3)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm4)
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 1, %xmm5)
|
|
MOVUPS_A2(-12 * SIZE, A1, LDA, 1, %xmm6)
|
|
MOVUPS_A2(-10 * SIZE, A1, LDA, 1, %xmm7)
|
|
|
|
decq I
|
|
jle .L24
|
|
ALIGN_3
|
|
|
|
.L23:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm0)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm2)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm3)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET(A1, LDA)
|
|
#endif
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm4)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm10
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA, 1, %xmm6)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm11
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET(A2)
|
|
#endif
|
|
|
|
mulpd %xmm14, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm0)
|
|
mulpd %xmm14, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm1)
|
|
|
|
mulpd %xmm14, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm2)
|
|
mulpd %xmm14, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_A1( -2 * SIZE, A1, %xmm3)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET(A2, LDA)
|
|
#endif
|
|
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_A2( -8 * SIZE, A1, LDA, 1, %xmm4)
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_A2( -6 * SIZE, A1, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm10
|
|
MOVUPS_A2( -4 * SIZE, A1, LDA, 1, %xmm6)
|
|
mulpd %xmm15, %xmm7
|
|
addpd %xmm7, %xmm11
|
|
MOVUPS_A2( -2 * SIZE, A1, LDA, 1, %xmm7)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 2 - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L23
|
|
ALIGN_3
|
|
|
|
.L24:
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm0)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm2)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm3)
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm4)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm10
|
|
MOVUPS_A2(-12 * SIZE, A2, LDA, 1, %xmm6)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm11
|
|
MOVUPS_A2(-10 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
mulpd %xmm14, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
mulpd %xmm14, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
|
|
mulpd %xmm14, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
mulpd %xmm14, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm10
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
mulpd %xmm15, %xmm7
|
|
addpd %xmm7, %xmm11
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L25:
|
|
testq $4, MM
|
|
je .L26
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm4)
|
|
MOVUPS_A2(-14 * SIZE, A1, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm1)
|
|
|
|
mulpd %xmm14, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
mulpd %xmm14, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm4)
|
|
MOVUPS_A2(-14 * SIZE, A2, LDA, 1, %xmm5)
|
|
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, A2
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L26:
|
|
testq $2, MM
|
|
je .L27
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm8)
|
|
MOVUPS_A2(-16 * SIZE, A1, LDA, 1, %xmm9)
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm10)
|
|
MOVUPS_A2(-16 * SIZE, A2, LDA, 1, %xmm11)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm12, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
mulpd %xmm13, %xmm9
|
|
addpd %xmm9, %xmm0
|
|
mulpd %xmm14, %xmm10
|
|
addpd %xmm10, %xmm0
|
|
mulpd %xmm15, %xmm11
|
|
addpd %xmm11, %xmm0
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, A2
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L27:
|
|
testq $1, MM
|
|
#if GEMV_UNROLL == 4
|
|
je .L28
|
|
#else
|
|
je .L30
|
|
#endif
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
movsd -16 * SIZE(A1), %xmm8
|
|
movsd -16 * SIZE(A1, LDA), %xmm9
|
|
movsd -16 * SIZE(A2), %xmm10
|
|
movsd -16 * SIZE(A2, LDA), %xmm11
|
|
|
|
mulsd %xmm12, %xmm8
|
|
addsd %xmm8, %xmm0
|
|
mulsd %xmm13, %xmm9
|
|
addsd %xmm9, %xmm0
|
|
mulsd %xmm14, %xmm10
|
|
addsd %xmm10, %xmm0
|
|
mulsd %xmm15, %xmm11
|
|
addsd %xmm11, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
ALIGN_3
|
|
|
|
#if GEMV_UNROLL == 4
|
|
.L28:
|
|
cmpq $4, N
|
|
jge .L21
|
|
ALIGN_3
|
|
|
|
#endif
|
|
|
|
.L30:
|
|
#endif
|
|
|
|
#if GEMV_UNROLL >= 2
|
|
|
|
cmpq $2, N
|
|
jl .L40
|
|
|
|
#if GEMV_UNROLL == 2
|
|
ALIGN_3
|
|
|
|
.L31:
|
|
#endif
|
|
|
|
subq $2, N
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
leaq (A, LDA), A2
|
|
leaq (A, LDA, 2), A
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
movddup (X), %xmm13
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
movsd (X), %xmm13
|
|
unpcklpd %xmm13, %xmm13
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm12
|
|
mulpd %xmm0, %xmm13
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
testq $SIZE, A
|
|
je .L3X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(A2), %xmm5
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, A2
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L3X:
|
|
#endif
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L35
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm2)
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm3)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm6)
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm7)
|
|
|
|
decq I
|
|
jle .L34
|
|
ALIGN_3
|
|
|
|
.L33:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 4 - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm0)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm2)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_A1( -2 * SIZE, A1, %xmm3)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 4 - 128 + PREOFFSET(A2)
|
|
#endif
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_A1( -8 * SIZE, A2, %xmm4)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_A1( -6 * SIZE, A2, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm10
|
|
MOVUPS_A1( -4 * SIZE, A2, %xmm6)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm11
|
|
MOVUPS_A1( -2 * SIZE, A2, %xmm7)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 4 - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L33
|
|
ALIGN_3
|
|
|
|
.L34:
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm10
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
mulpd %xmm13, %xmm7
|
|
addpd %xmm7, %xmm11
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L35:
|
|
testq $4, MM
|
|
je .L36
|
|
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, A2
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L36:
|
|
testq $2, MM
|
|
je .L37
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm8)
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm9)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm12, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
mulpd %xmm13, %xmm9
|
|
addpd %xmm9, %xmm0
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, A2
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L37:
|
|
testq $1, MM
|
|
#if GEMV_UNROLL == 2
|
|
je .L38
|
|
#else
|
|
je .L40
|
|
#endif
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
movsd -16 * SIZE(A1), %xmm8
|
|
movsd -16 * SIZE(A2), %xmm9
|
|
|
|
mulsd %xmm12, %xmm8
|
|
addsd %xmm8, %xmm0
|
|
mulsd %xmm13, %xmm9
|
|
addsd %xmm9, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
ALIGN_3
|
|
|
|
#if GEMV_UNROLL == 2
|
|
.L38:
|
|
cmpq $2, N
|
|
jge .L31
|
|
ALIGN_3
|
|
|
|
#endif
|
|
|
|
.L40:
|
|
cmpq $1, N
|
|
jl .L900
|
|
#endif
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm12
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
testq $SIZE, A
|
|
je .L4X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L4X:
|
|
#endif
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L45
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm2)
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm3)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
decq I
|
|
jle .L44
|
|
ALIGN_3
|
|
|
|
.L43:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 8 - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm0)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm2)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_A1( -2 * SIZE, A1, %xmm3)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 8 - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L43
|
|
ALIGN_3
|
|
|
|
.L44:
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L45:
|
|
testq $4, MM
|
|
je .L46
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L46:
|
|
testq $2, MM
|
|
je .L47
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm8)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm12, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L47:
|
|
testq $1, MM
|
|
je .L900
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
movsd -16 * SIZE(A1), %xmm8
|
|
|
|
mulsd %xmm12, %xmm8
|
|
addsd %xmm8, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
ALIGN_3
|
|
|
|
#ifdef ALIGNED_ACCESS
|
|
jmp .L900
|
|
ALIGN_3
|
|
|
|
.L50:
|
|
#if GEMV_UNROLL >= 4
|
|
|
|
cmpq $4, N
|
|
jl .L60
|
|
ALIGN_3
|
|
|
|
.L51:
|
|
|
|
subq $4, N
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
leaq (A, LDA, 2), A2
|
|
leaq (A, LDA, 4), A
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
movddup (X), %xmm13
|
|
addq INCX, X
|
|
movddup (X), %xmm14
|
|
addq INCX, X
|
|
movddup (X), %xmm15
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
movsd (X), %xmm13
|
|
unpcklpd %xmm13, %xmm13
|
|
addq INCX, X
|
|
movsd (X), %xmm14
|
|
unpcklpd %xmm14, %xmm14
|
|
addq INCX, X
|
|
movsd (X), %xmm15
|
|
unpcklpd %xmm15, %xmm15
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm12
|
|
mulpd %xmm0, %xmm13
|
|
mulpd %xmm0, %xmm14
|
|
mulpd %xmm0, %xmm15
|
|
|
|
testq $SIZE, A
|
|
je .L5X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(A1, LDA), %xmm5
|
|
movsd -16 * SIZE(A2), %xmm6
|
|
movsd -16 * SIZE(A2, LDA), %xmm7
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
mulsd %xmm14, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
mulsd %xmm15, %xmm7
|
|
addsd %xmm7, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, A2
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L5X:
|
|
movhpd -16 * SIZE(A1, LDA), %xmm8
|
|
movhpd -16 * SIZE(A2, LDA), %xmm9
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L55
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm5)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm6)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
decq I
|
|
jle .L54
|
|
ALIGN_3
|
|
|
|
.L53:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm7)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-15 * SIZE, A1, LDA, 1, %xmm4)
|
|
|
|
mulpd %xmm12, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-13 * SIZE, A1, LDA, 1, %xmm5)
|
|
mulpd %xmm12, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-11 * SIZE, A1, LDA, 1, %xmm6)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET + 8(A1, LDA)
|
|
#endif
|
|
|
|
shufpd $1, %xmm4, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
MOVUPS_A2( -9 * SIZE, A1, LDA, 1, %xmm8)
|
|
shufpd $1, %xmm5, %xmm4
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm1
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
|
|
shufpd $1, %xmm6, %xmm5
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm2
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
shufpd $1, %xmm8, %xmm6
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm3
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm6)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET(A2)
|
|
#endif
|
|
|
|
mulpd %xmm14, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm7)
|
|
mulpd %xmm14, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-15 * SIZE, A2, LDA, 1, %xmm4)
|
|
|
|
mulpd %xmm14, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-13 * SIZE, A2, LDA, 1, %xmm5)
|
|
mulpd %xmm14, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-11 * SIZE, A2, LDA, 1, %xmm6)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 2 - 128 + PREOFFSET + 8(A2, LDA)
|
|
#endif
|
|
|
|
shufpd $1, %xmm4, %xmm9
|
|
mulpd %xmm15, %xmm9
|
|
addpd %xmm9, %xmm0
|
|
MOVUPS_A2( -9 * SIZE, A2, LDA, 1, %xmm9)
|
|
shufpd $1, %xmm5, %xmm4
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm1
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm4)
|
|
|
|
shufpd $1, %xmm6, %xmm5
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm2
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm5)
|
|
shufpd $1, %xmm9, %xmm6
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm3
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm6)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 2 - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm3)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L53
|
|
ALIGN_3
|
|
|
|
.L54:
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm7)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-15 * SIZE, A1, LDA, 1, %xmm4)
|
|
|
|
mulpd %xmm12, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-13 * SIZE, A1, LDA, 1, %xmm5)
|
|
mulpd %xmm12, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-11 * SIZE, A1, LDA, 1, %xmm6)
|
|
|
|
shufpd $1, %xmm4, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
MOVUPS_A2( -9 * SIZE, A1, LDA, 1, %xmm8)
|
|
shufpd $1, %xmm5, %xmm4
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm1
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
|
|
shufpd $1, %xmm6, %xmm5
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm2
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
shufpd $1, %xmm8, %xmm6
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm3
|
|
MOVUPS_A1(-12 * SIZE, A2, %xmm6)
|
|
|
|
mulpd %xmm14, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-10 * SIZE, A2, %xmm7)
|
|
mulpd %xmm14, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A2(-15 * SIZE, A2, LDA, 1, %xmm4)
|
|
|
|
mulpd %xmm14, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A2(-13 * SIZE, A2, LDA, 1, %xmm5)
|
|
mulpd %xmm14, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A2(-11 * SIZE, A2, LDA, 1, %xmm6)
|
|
|
|
shufpd $1, %xmm4, %xmm9
|
|
mulpd %xmm15, %xmm9
|
|
addpd %xmm9, %xmm0
|
|
MOVUPS_A2( -9 * SIZE, A2, LDA, 1, %xmm9)
|
|
|
|
shufpd $1, %xmm5, %xmm4
|
|
mulpd %xmm15, %xmm4
|
|
addpd %xmm4, %xmm1
|
|
shufpd $1, %xmm6, %xmm5
|
|
mulpd %xmm15, %xmm5
|
|
addpd %xmm5, %xmm2
|
|
shufpd $1, %xmm9, %xmm6
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm3
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L55:
|
|
testq $4, MM
|
|
je .L56
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm5)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
|
|
MOVUPS_A2(-15 * SIZE, A1, LDA, 1, %xmm6)
|
|
MOVUPS_A2(-13 * SIZE, A1, LDA, 1, %xmm7)
|
|
|
|
shufpd $1, %xmm6, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
movaps %xmm7, %xmm8
|
|
shufpd $1, %xmm7, %xmm6
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm1
|
|
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A2, %xmm5)
|
|
|
|
mulpd %xmm14, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
mulpd %xmm14, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
|
|
MOVUPS_A2(-15 * SIZE, A2, LDA, 1, %xmm6)
|
|
MOVUPS_A2(-13 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
shufpd $1, %xmm6, %xmm9
|
|
mulpd %xmm15, %xmm9
|
|
addpd %xmm9, %xmm0
|
|
movaps %xmm7, %xmm9
|
|
shufpd $1, %xmm7, %xmm6
|
|
mulpd %xmm15, %xmm6
|
|
addpd %xmm6, %xmm1
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, A2
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L56:
|
|
testq $2, MM
|
|
je .L57
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A2(-15 * SIZE, A1, LDA, 1, %xmm5)
|
|
MOVUPS_A1(-16 * SIZE, A2, %xmm6)
|
|
MOVUPS_A2(-15 * SIZE, A2, LDA, 1, %xmm7)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
shufpd $1, %xmm5, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
movaps %xmm5, %xmm8
|
|
mulpd %xmm14, %xmm6
|
|
addpd %xmm6, %xmm0
|
|
shufpd $1, %xmm7, %xmm9
|
|
mulpd %xmm15, %xmm9
|
|
addpd %xmm9, %xmm0
|
|
movaps %xmm7, %xmm9
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, A2
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L57:
|
|
testq $1, MM
|
|
je .L58
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
shufpd $1, %xmm8, %xmm8
|
|
movsd -16 * SIZE(A2), %xmm6
|
|
shufpd $1, %xmm9, %xmm9
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm8
|
|
addsd %xmm8, %xmm0
|
|
mulsd %xmm14, %xmm6
|
|
addsd %xmm6, %xmm0
|
|
mulsd %xmm15, %xmm9
|
|
addsd %xmm9, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
ALIGN_3
|
|
|
|
.L58:
|
|
cmpq $4, N
|
|
jge .L51
|
|
ALIGN_3
|
|
|
|
.L60:
|
|
#endif
|
|
|
|
#if GEMV_UNROLL >= 2
|
|
|
|
cmpq $2, N
|
|
jl .L70
|
|
|
|
#if GEMV_UNROLL == 2
|
|
ALIGN_3
|
|
|
|
.L61:
|
|
#endif
|
|
|
|
subq $2, N
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
leaq (A, LDA), A2
|
|
leaq (A, LDA, 2), A
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
movddup (X), %xmm13
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
movsd (X), %xmm13
|
|
unpcklpd %xmm13, %xmm13
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm12
|
|
mulpd %xmm0, %xmm13
|
|
|
|
testq $SIZE, A
|
|
je .L6X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(A2), %xmm5
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm5
|
|
addsd %xmm5, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, A2
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L6X:
|
|
movhpd -16 * SIZE(A2), %xmm8
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L65
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm5)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm6)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
decq I
|
|
jle .L64
|
|
ALIGN_3
|
|
|
|
.L63:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 4 - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm7)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A1(-15 * SIZE, A2, %xmm4)
|
|
|
|
mulpd %xmm12, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A1(-13 * SIZE, A2, %xmm5)
|
|
mulpd %xmm12, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A1(-11 * SIZE, A2, %xmm6)
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 4 - 128 + PREOFFSET + 8(A2)
|
|
#endif
|
|
|
|
shufpd $1, %xmm4, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
MOVUPS_A1( -9 * SIZE, A2, %xmm8)
|
|
shufpd $1, %xmm5, %xmm4
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm1
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm4)
|
|
|
|
shufpd $1, %xmm6, %xmm5
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm2
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm5)
|
|
shufpd $1, %xmm8, %xmm6
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm3
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm6)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 4 - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm3)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L63
|
|
ALIGN_3
|
|
|
|
.L64:
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm7)
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
MOVUPS_A1(-15 * SIZE, A2, %xmm4)
|
|
|
|
mulpd %xmm12, %xmm6
|
|
addpd %xmm6, %xmm2
|
|
MOVUPS_A1(-13 * SIZE, A2, %xmm5)
|
|
mulpd %xmm12, %xmm7
|
|
addpd %xmm7, %xmm3
|
|
MOVUPS_A1(-11 * SIZE, A2, %xmm6)
|
|
|
|
shufpd $1, %xmm4, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
MOVUPS_A1( -9 * SIZE, A2, %xmm8)
|
|
shufpd $1, %xmm5, %xmm4
|
|
mulpd %xmm13, %xmm4
|
|
addpd %xmm4, %xmm1
|
|
|
|
shufpd $1, %xmm6, %xmm5
|
|
mulpd %xmm13, %xmm5
|
|
addpd %xmm5, %xmm2
|
|
shufpd $1, %xmm8, %xmm6
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm3
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm2)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm3)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, A2
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L65:
|
|
testq $4, MM
|
|
je .L66
|
|
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm5)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
mulpd %xmm12, %xmm5
|
|
addpd %xmm5, %xmm1
|
|
|
|
MOVUPS_A1(-15 * SIZE, A2, %xmm6)
|
|
MOVUPS_A1(-13 * SIZE, A2, %xmm7)
|
|
|
|
shufpd $1, %xmm6, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
movaps %xmm7, %xmm8
|
|
shufpd $1, %xmm7, %xmm6
|
|
mulpd %xmm13, %xmm6
|
|
addpd %xmm6, %xmm1
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm1)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, A2
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L66:
|
|
testq $2, MM
|
|
je .L67
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm4)
|
|
MOVUPS_A1(-15 * SIZE, A2, %xmm5)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm12, %xmm4
|
|
addpd %xmm4, %xmm0
|
|
shufpd $1, %xmm5, %xmm8
|
|
mulpd %xmm13, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
movaps %xmm5, %xmm8
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, A2
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L67:
|
|
testq $1, MM
|
|
#if GEMV_UNROLL == 2
|
|
je .L68
|
|
#else
|
|
je .L70
|
|
#endif
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
shufpd $1, %xmm8, %xmm8
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
mulsd %xmm13, %xmm8
|
|
addsd %xmm8, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
ALIGN_3
|
|
|
|
#if GEMV_UNROLL == 2
|
|
.L68:
|
|
cmpq $2, N
|
|
jge .L61
|
|
ALIGN_3
|
|
|
|
#endif
|
|
|
|
.L70:
|
|
cmpq $1, N
|
|
jl .L900
|
|
|
|
#endif
|
|
|
|
leaq 16 * SIZE(BUFFER), Y1
|
|
movq A, A1
|
|
|
|
#ifdef HAVE_SSE3
|
|
movddup (X), %xmm12
|
|
addq INCX, X
|
|
|
|
movddup ALPHA, %xmm0
|
|
#else
|
|
movsd (X), %xmm12
|
|
unpcklpd %xmm12, %xmm12
|
|
addq INCX, X
|
|
|
|
movsd ALPHA, %xmm0
|
|
unpcklpd %xmm0, %xmm0
|
|
#endif
|
|
|
|
mulpd %xmm0, %xmm12
|
|
|
|
testq $SIZE, A
|
|
je .L7X
|
|
|
|
movsd -16 * SIZE(A1), %xmm4
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
|
|
mulsd %xmm12, %xmm4
|
|
addsd %xmm4, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
|
|
addq $SIZE, A1
|
|
addq $SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L7X:
|
|
|
|
movq MM, I
|
|
sarq $3, I
|
|
jle .L75
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
MOVUPS_A1(-12 * SIZE, A1, %xmm2)
|
|
MOVUPS_A1(-10 * SIZE, A1, %xmm3)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
decq I
|
|
jle .L74
|
|
ALIGN_3
|
|
|
|
.L73:
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 8 - 128 + PREOFFSET(A1)
|
|
#endif
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_A1( -8 * SIZE, A1, %xmm0)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_A1( -6 * SIZE, A1, %xmm1)
|
|
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_A1( -4 * SIZE, A1, %xmm2)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_A1( -2 * SIZE, A1, %xmm3)
|
|
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 8 - 128 + PREOFFSET(Y1)
|
|
#endif
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
MOVUPS_YL1( -8 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1( -6 * SIZE, Y1, %xmm9)
|
|
MOVUPS_YL1( -4 * SIZE, Y1, %xmm10)
|
|
MOVUPS_YL1( -2 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, Y1
|
|
|
|
subq $1, I
|
|
BRANCH
|
|
jg .L73
|
|
ALIGN_3
|
|
|
|
.L74:
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
mulpd %xmm12, %xmm2
|
|
addpd %xmm2, %xmm10
|
|
MOVUPS_YS1(-12 * SIZE, Y1, %xmm10)
|
|
mulpd %xmm12, %xmm3
|
|
addpd %xmm3, %xmm11
|
|
MOVUPS_YS1(-10 * SIZE, Y1, %xmm11)
|
|
|
|
subq $-8 * SIZE, A1
|
|
subq $-8 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L75:
|
|
testq $4, MM
|
|
je .L76
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm0)
|
|
MOVUPS_A1(-14 * SIZE, A1, %xmm1)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm8)
|
|
MOVUPS_YL1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
mulpd %xmm12, %xmm0
|
|
addpd %xmm0, %xmm8
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm8)
|
|
mulpd %xmm12, %xmm1
|
|
addpd %xmm1, %xmm9
|
|
MOVUPS_YS1(-14 * SIZE, Y1, %xmm9)
|
|
|
|
addq $4 * SIZE, A1
|
|
addq $4 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L76:
|
|
testq $2, MM
|
|
je .L77
|
|
|
|
MOVUPS_A1(-16 * SIZE, A1, %xmm8)
|
|
|
|
MOVUPS_YL1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
mulpd %xmm12, %xmm8
|
|
addpd %xmm8, %xmm0
|
|
|
|
MOVUPS_YS1(-16 * SIZE, Y1, %xmm0)
|
|
|
|
addq $2 * SIZE, A1
|
|
addq $2 * SIZE, Y1
|
|
ALIGN_3
|
|
|
|
.L77:
|
|
testq $1, MM
|
|
je .L900
|
|
|
|
movsd -16 * SIZE(Y1), %xmm0
|
|
movsd -16 * SIZE(A1), %xmm8
|
|
|
|
mulsd %xmm12, %xmm8
|
|
addsd %xmm8, %xmm0
|
|
|
|
movsd %xmm0, -16 * SIZE(Y1)
|
|
#endif
|
|
ALIGN_3
|
|
|
|
|
|
.L900:
|
|
#ifndef COPY_FORCE
|
|
cmpq Y, BUFFER
|
|
je .L999
|
|
#endif
|
|
|
|
cmpq $SIZE, INCY
|
|
jne .L950
|
|
|
|
testq $SIZE, Y
|
|
je .L910
|
|
|
|
movsd (Y), %xmm0
|
|
addsd (BUFFER), %xmm0
|
|
movsd %xmm0, (Y)
|
|
|
|
addq $SIZE, Y
|
|
addq $SIZE, BUFFER
|
|
|
|
decq M
|
|
jle .L999
|
|
ALIGN_4
|
|
|
|
.L910:
|
|
testq $SIZE, BUFFER
|
|
jne .L920
|
|
|
|
movq M, %rax
|
|
sarq $3, %rax
|
|
jle .L914
|
|
ALIGN_3
|
|
|
|
.L912:
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 4 + PREOFFSET(Y)
|
|
#endif
|
|
|
|
movapd 0 * SIZE(Y), %xmm0
|
|
movapd 2 * SIZE(Y), %xmm1
|
|
movapd 4 * SIZE(Y), %xmm2
|
|
movapd 6 * SIZE(Y), %xmm3
|
|
|
|
movapd 0 * SIZE(BUFFER), %xmm4
|
|
movapd 2 * SIZE(BUFFER), %xmm5
|
|
movapd 4 * SIZE(BUFFER), %xmm6
|
|
movapd 6 * SIZE(BUFFER), %xmm7
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 4 + PREOFFSET(BUFFER)
|
|
#endif
|
|
|
|
addpd %xmm4, %xmm0
|
|
addpd %xmm5, %xmm1
|
|
addpd %xmm6, %xmm2
|
|
addpd %xmm7, %xmm3
|
|
|
|
movapd %xmm0, 0 * SIZE(Y)
|
|
movapd %xmm1, 2 * SIZE(Y)
|
|
movapd %xmm2, 4 * SIZE(Y)
|
|
movapd %xmm3, 6 * SIZE(Y)
|
|
|
|
addq $8 * SIZE, Y
|
|
addq $8 * SIZE, BUFFER
|
|
|
|
decq %rax
|
|
jg .L912
|
|
ALIGN_3
|
|
|
|
.L914:
|
|
testq $7, M
|
|
jle .L999
|
|
|
|
testq $4, M
|
|
jle .L915
|
|
|
|
movapd 0 * SIZE(Y), %xmm0
|
|
movapd 2 * SIZE(Y), %xmm1
|
|
|
|
movapd 0 * SIZE(BUFFER), %xmm4
|
|
movapd 2 * SIZE(BUFFER), %xmm5
|
|
|
|
addpd %xmm4, %xmm0
|
|
addpd %xmm5, %xmm1
|
|
|
|
movapd %xmm0, 0 * SIZE(Y)
|
|
movapd %xmm1, 2 * SIZE(Y)
|
|
|
|
addq $4 * SIZE, Y
|
|
addq $4 * SIZE, BUFFER
|
|
ALIGN_3
|
|
|
|
.L915:
|
|
testq $2, M
|
|
jle .L916
|
|
|
|
movapd (Y), %xmm0
|
|
|
|
movapd (BUFFER), %xmm4
|
|
|
|
addpd %xmm4, %xmm0
|
|
|
|
movapd %xmm0, (Y)
|
|
|
|
addq $2 * SIZE, Y
|
|
addq $2 * SIZE, BUFFER
|
|
ALIGN_3
|
|
|
|
.L916:
|
|
testq $1, M
|
|
jle .L999
|
|
|
|
movsd (Y), %xmm0
|
|
|
|
movsd 0 * SIZE(BUFFER), %xmm4
|
|
|
|
addsd %xmm4, %xmm0
|
|
|
|
movlpd %xmm0, (Y)
|
|
ALIGN_3
|
|
|
|
jmp .L999
|
|
ALIGN_4
|
|
|
|
.L920:
|
|
movapd -1 * SIZE(BUFFER), %xmm4
|
|
|
|
movq M, %rax
|
|
sarq $3, %rax
|
|
jle .L924
|
|
ALIGN_3
|
|
|
|
.L922:
|
|
#ifdef PREFETCHW
|
|
PREFETCHW (PREFETCHSIZE) * 4 + PREOFFSET(Y)
|
|
#endif
|
|
|
|
movapd 0 * SIZE(Y), %xmm0
|
|
movapd 2 * SIZE(Y), %xmm1
|
|
movapd 4 * SIZE(Y), %xmm2
|
|
movapd 6 * SIZE(Y), %xmm3
|
|
|
|
movapd 1 * SIZE(BUFFER), %xmm5
|
|
movapd 3 * SIZE(BUFFER), %xmm6
|
|
movapd 5 * SIZE(BUFFER), %xmm7
|
|
movapd 7 * SIZE(BUFFER), %xmm8
|
|
|
|
shufpd $1, %xmm5, %xmm4
|
|
shufpd $1, %xmm6, %xmm5
|
|
shufpd $1, %xmm7, %xmm6
|
|
shufpd $1, %xmm8, %xmm7
|
|
|
|
#ifdef PREFETCH
|
|
PREFETCH (PREFETCHSIZE) * 4 + PREOFFSET(BUFFER)
|
|
#endif
|
|
|
|
addpd %xmm4, %xmm0
|
|
addpd %xmm5, %xmm1
|
|
addpd %xmm6, %xmm2
|
|
addpd %xmm7, %xmm3
|
|
|
|
movapd %xmm0, 0 * SIZE(Y)
|
|
movapd %xmm1, 2 * SIZE(Y)
|
|
movapd %xmm2, 4 * SIZE(Y)
|
|
movapd %xmm3, 6 * SIZE(Y)
|
|
|
|
movapd %xmm8, %xmm4
|
|
|
|
addq $8 * SIZE, Y
|
|
addq $8 * SIZE, BUFFER
|
|
|
|
decq %rax
|
|
jg .L922
|
|
ALIGN_3
|
|
|
|
.L924:
|
|
testq $7, M
|
|
jle .L999
|
|
|
|
testq $4, M
|
|
jle .L925
|
|
|
|
movapd 0 * SIZE(Y), %xmm0
|
|
movapd 2 * SIZE(Y), %xmm1
|
|
|
|
movapd 1 * SIZE(BUFFER), %xmm5
|
|
movapd 3 * SIZE(BUFFER), %xmm6
|
|
|
|
shufpd $1, %xmm5, %xmm4
|
|
shufpd $1, %xmm6, %xmm5
|
|
|
|
addpd %xmm4, %xmm0
|
|
addpd %xmm5, %xmm1
|
|
|
|
movapd %xmm0, 0 * SIZE(Y)
|
|
movapd %xmm1, 2 * SIZE(Y)
|
|
|
|
movapd %xmm6, %xmm4
|
|
|
|
addq $4 * SIZE, Y
|
|
addq $4 * SIZE, BUFFER
|
|
ALIGN_3
|
|
|
|
.L925:
|
|
testq $2, M
|
|
jle .L926
|
|
|
|
movapd (Y), %xmm0
|
|
|
|
movapd 1 * SIZE(BUFFER), %xmm5
|
|
|
|
shufpd $1, %xmm5, %xmm4
|
|
|
|
addpd %xmm4, %xmm0
|
|
|
|
movapd %xmm0, (Y)
|
|
|
|
movaps %xmm5, %xmm4
|
|
|
|
addq $2 * SIZE, Y
|
|
addq $2 * SIZE, BUFFER
|
|
ALIGN_3
|
|
|
|
.L926:
|
|
testq $1, M
|
|
jle .L999
|
|
|
|
movsd (Y), %xmm0
|
|
|
|
shufpd $1, %xmm4, %xmm4
|
|
|
|
addsd %xmm4, %xmm0
|
|
|
|
movlpd %xmm0, (Y)
|
|
ALIGN_3
|
|
|
|
jmp .L999
|
|
ALIGN_4
|
|
|
|
.L950:
|
|
testq $SIZE, BUFFER
|
|
je .L960
|
|
|
|
movsd (Y), %xmm0
|
|
addsd (BUFFER), %xmm0
|
|
movsd %xmm0, (Y)
|
|
|
|
addq INCY, Y
|
|
addq $SIZE, BUFFER
|
|
|
|
decq M
|
|
jle .L999
|
|
ALIGN_4
|
|
|
|
.L960:
|
|
movq Y, Y1
|
|
|
|
movq M, %rax
|
|
sarq $3, %rax
|
|
jle .L964
|
|
ALIGN_3
|
|
|
|
.L962:
|
|
movsd (Y), %xmm0
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm0
|
|
addq INCY, Y
|
|
|
|
movapd 0 * SIZE(BUFFER), %xmm4
|
|
|
|
movsd (Y), %xmm1
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm1
|
|
addq INCY, Y
|
|
|
|
movapd 2 * SIZE(BUFFER), %xmm5
|
|
|
|
movsd (Y), %xmm2
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm2
|
|
addq INCY, Y
|
|
|
|
movapd 4 * SIZE(BUFFER), %xmm6
|
|
|
|
addpd %xmm4, %xmm0
|
|
|
|
movsd (Y), %xmm3
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm3
|
|
addq INCY, Y
|
|
|
|
movapd 6 * SIZE(BUFFER), %xmm7
|
|
|
|
addpd %xmm5, %xmm1
|
|
|
|
movlpd %xmm0, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm0, (Y1)
|
|
addq INCY, Y1
|
|
|
|
addpd %xmm6, %xmm2
|
|
|
|
movlpd %xmm1, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm1, (Y1)
|
|
addq INCY, Y1
|
|
|
|
addpd %xmm7, %xmm3
|
|
|
|
movlpd %xmm2, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm2, (Y1)
|
|
addq INCY, Y1
|
|
movlpd %xmm3, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm3, (Y1)
|
|
addq INCY, Y1
|
|
|
|
addq $8 * SIZE, BUFFER
|
|
decq %rax
|
|
jg .L962
|
|
ALIGN_3
|
|
|
|
.L964:
|
|
testq $7, M
|
|
jle .L999
|
|
|
|
testq $4, M
|
|
jle .L965
|
|
|
|
movsd (Y), %xmm0
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm0
|
|
addq INCY, Y
|
|
|
|
movapd 0 * SIZE(BUFFER), %xmm4
|
|
|
|
movsd (Y), %xmm1
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm1
|
|
addq INCY, Y
|
|
|
|
movapd 2 * SIZE(BUFFER), %xmm5
|
|
|
|
addpd %xmm4, %xmm0
|
|
addpd %xmm5, %xmm1
|
|
|
|
movlpd %xmm0, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm0, (Y1)
|
|
addq INCY, Y1
|
|
movlpd %xmm1, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm1, (Y1)
|
|
addq INCY, Y1
|
|
|
|
addq $4 * SIZE, BUFFER
|
|
ALIGN_3
|
|
|
|
.L965:
|
|
testq $2, M
|
|
jle .L966
|
|
|
|
movsd (Y), %xmm0
|
|
addq INCY, Y
|
|
movhpd (Y), %xmm0
|
|
addq INCY, Y
|
|
|
|
movapd 0 * SIZE(BUFFER), %xmm4
|
|
|
|
addpd %xmm4, %xmm0
|
|
|
|
movlpd %xmm0, (Y1)
|
|
addq INCY, Y1
|
|
movhpd %xmm0, (Y1)
|
|
addq INCY, Y1
|
|
|
|
addq $2 * SIZE, BUFFER
|
|
ALIGN_3
|
|
|
|
.L966:
|
|
testq $1, M
|
|
jle .L999
|
|
|
|
movsd (Y), %xmm0
|
|
|
|
movsd 0 * SIZE(BUFFER), %xmm4
|
|
|
|
addsd %xmm4, %xmm0
|
|
|
|
movlpd %xmm0, (Y1)
|
|
ALIGN_3
|
|
|
|
.L999:
|
|
movq 0(%rsp), %rbx
|
|
movq 8(%rsp), %rbp
|
|
movq 16(%rsp), %r12
|
|
movq 24(%rsp), %r13
|
|
movq 32(%rsp), %r14
|
|
movq 40(%rsp), %r15
|
|
|
|
#ifdef WINDOWS_ABI
|
|
movq 48(%rsp), %rdi
|
|
movq 56(%rsp), %rsi
|
|
movups 64(%rsp), %xmm6
|
|
movups 80(%rsp), %xmm7
|
|
movups 96(%rsp), %xmm8
|
|
movups 112(%rsp), %xmm9
|
|
movups 128(%rsp), %xmm10
|
|
movups 144(%rsp), %xmm11
|
|
movups 160(%rsp), %xmm12
|
|
movups 176(%rsp), %xmm13
|
|
movups 192(%rsp), %xmm14
|
|
movups 208(%rsp), %xmm15
|
|
#endif
|
|
|
|
addq $STACKSIZE, %rsp
|
|
|
|
ret
|
|
EPILOGUE
|