431 lines
7.6 KiB
ArmAsm
431 lines
7.6 KiB
ArmAsm
/***************************************************************************
|
|
Copyright (c) 2013, The OpenBLAS Project
|
|
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.
|
|
3. Neither the name of the OpenBLAS project nor the names of
|
|
its contributors may be used to endorse or promote products
|
|
derived from this software without specific prior written permission.
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "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 OPENBLAS PROJECT 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.
|
|
*****************************************************************************/
|
|
|
|
/**************************************************************************************
|
|
* 2013/11/06 Saar
|
|
* BLASTEST : OK
|
|
* CTEST : OK
|
|
* TEST : OK
|
|
*
|
|
**************************************************************************************/
|
|
|
|
#define ASSEMBLER
|
|
#include "common.h"
|
|
|
|
#define STACKSIZE 256
|
|
|
|
#define OLD_M r0
|
|
#define OLD_N r1
|
|
#define OLD_A r2
|
|
#define OLD_LDA r3
|
|
|
|
|
|
/******************************************************
|
|
* [fp, #-128] - [fp, #-64] is reserved
|
|
* for store and restore of floating point
|
|
* registers
|
|
*******************************************************/
|
|
|
|
#define B [fp, #4 ]
|
|
#define A [fp, #-248 ]
|
|
|
|
#define M r0
|
|
#define N r1
|
|
#define M4 r2
|
|
|
|
#define LDA r5
|
|
|
|
#define AO1 r6
|
|
#define BO1 r7
|
|
#define BO2 r8
|
|
#define BO3 r9
|
|
|
|
#define I r4
|
|
#define J r12
|
|
|
|
#define A_PRE 256
|
|
|
|
/**************************************************************************************
|
|
* Macro definitions
|
|
**************************************************************************************/
|
|
|
|
.macro COPY4x4_1
|
|
|
|
pld [ AO1, #A_PRE ]
|
|
vldmia.f32 AO1, { s0 - s3 }
|
|
|
|
add r3, AO1, LDA
|
|
pld [ r3, #A_PRE ]
|
|
vldmia.f32 r3, { s4 - s7 }
|
|
|
|
add r3, r3, LDA
|
|
pld [ r3, #A_PRE ]
|
|
vldmia.f32 r3, { s8 - s11 }
|
|
|
|
add r3, r3, LDA
|
|
pld [ r3, #A_PRE ]
|
|
vldmia.f32 r3, { s12 - s15 }
|
|
|
|
vstmia.f32 BO1, { s0 - s15 }
|
|
add AO1, AO1, #16
|
|
add BO1, BO1, M4
|
|
|
|
.endm
|
|
|
|
.macro COPY4x4_2
|
|
|
|
vldmia.f32 AO1, { s0 - s3 }
|
|
|
|
add r3, AO1, LDA
|
|
vldmia.f32 r3, { s4 - s7 }
|
|
|
|
add r3, r3, LDA
|
|
vldmia.f32 r3, { s8 - s11 }
|
|
|
|
add r3, r3, LDA
|
|
vldmia.f32 r3, { s12 - s15 }
|
|
|
|
vstmia.f32 BO1, { s0 - s15 }
|
|
add AO1, AO1, #16
|
|
add BO1, BO1, M4
|
|
|
|
.endm
|
|
|
|
|
|
.macro COPY2x4
|
|
|
|
vldmia.f32 AO1, { s0 - s1 }
|
|
|
|
add r3, AO1, LDA
|
|
vldmia.f32 r3, { s2 - s3 }
|
|
|
|
add r3, r3, LDA
|
|
vldmia.f32 r3, { s4 - s5 }
|
|
|
|
add r3, r3, LDA
|
|
vldmia.f32 r3, { s6 - s7 }
|
|
|
|
vstmia.f32 BO2, { s0 - s7 }
|
|
add AO1, AO1, #8
|
|
add BO2, BO2, #32
|
|
|
|
.endm
|
|
|
|
.macro COPY1x4
|
|
|
|
vldmia.f32 AO1, { s0 }
|
|
|
|
add r3, AO1, LDA
|
|
vldmia.f32 r3, { s1 }
|
|
|
|
add r3, r3, LDA
|
|
vldmia.f32 r3, { s2 }
|
|
|
|
add r3, r3, LDA
|
|
vldmia.f32 r3, { s3 }
|
|
|
|
vstmia.f32 BO3, { s0 - s3 }
|
|
add AO1, AO1, #4
|
|
add BO3, BO3, #16
|
|
|
|
.endm
|
|
|
|
/*************************************************************************************************************************/
|
|
|
|
.macro COPY4x2
|
|
|
|
vldmia.f32 AO1, { s0 - s3 }
|
|
|
|
add r3, AO1, LDA
|
|
vldmia.f32 r3, { s4 - s7 }
|
|
|
|
vstmia.f32 BO1, { s0 - s7 }
|
|
add AO1, AO1, #16
|
|
add BO1, BO1, M4
|
|
|
|
.endm
|
|
|
|
.macro COPY2x2
|
|
|
|
vldmia.f32 AO1, { s0 - s1 }
|
|
|
|
add r3, AO1, LDA
|
|
vldmia.f32 r3, { s2 - s3 }
|
|
|
|
vstmia.f32 BO2, { s0 - s3 }
|
|
add AO1, AO1, #8
|
|
add BO2, BO2, #16
|
|
|
|
.endm
|
|
|
|
.macro COPY1x2
|
|
|
|
vldmia.f32 AO1, { s0 }
|
|
|
|
add r3, AO1, LDA
|
|
vldmia.f32 r3, { s1 }
|
|
|
|
vstmia.f32 BO3, { s0 - s1 }
|
|
add AO1, AO1, #4
|
|
add BO3, BO3, #8
|
|
|
|
.endm
|
|
|
|
/*************************************************************************************************************************/
|
|
|
|
.macro COPY4x1
|
|
|
|
vldmia.f32 AO1, { s0 - s3 }
|
|
|
|
vstmia.f32 BO1, { s0 - s3 }
|
|
add AO1, AO1, #16
|
|
add BO1, BO1, M4
|
|
|
|
.endm
|
|
|
|
.macro COPY2x1
|
|
|
|
vldmia.f32 AO1, { s0 - s1 }
|
|
|
|
vstmia.f32 BO2, { s0 - s1 }
|
|
add AO1, AO1, #8
|
|
add BO2, BO2, #8
|
|
|
|
.endm
|
|
|
|
.macro COPY1x1
|
|
|
|
vldmia.f32 AO1, { s0 }
|
|
|
|
vstmia.f32 BO3, { s0 }
|
|
add AO1, AO1, #4
|
|
add BO3, BO3, #4
|
|
|
|
.endm
|
|
|
|
|
|
|
|
/**************************************************************************************
|
|
* End of macro definitions
|
|
**************************************************************************************/
|
|
|
|
PROLOGUE
|
|
|
|
.align 5
|
|
|
|
push {r4 - r9, fp}
|
|
add fp, sp, #24
|
|
sub sp, sp, #STACKSIZE // reserve stack
|
|
|
|
str OLD_A, A // store A
|
|
|
|
lsl LDA, OLD_LDA, #2 // lda = lda * SIZE
|
|
|
|
sub r4, fp, #128
|
|
vstm r4, { s8 - s15} // store floating point registers
|
|
|
|
lsl r4 , M, #2 // M * SIZE
|
|
|
|
ldr r3, B
|
|
|
|
and BO2 , N , #-4
|
|
and BO3 , N , #-2
|
|
|
|
mul BO2, BO2, r4
|
|
mul BO3, BO3, r4
|
|
|
|
add BO2 , BO2, r3
|
|
add BO3 , BO3, r3
|
|
|
|
lsl M4, M, #4 // M4 = M * 4 * SIZE
|
|
|
|
sgemm_tcopy_L4_BEGIN:
|
|
|
|
asrs J, M, #2 // J = N / 4
|
|
ble sgemm_tcopy_L2_BEGIN
|
|
|
|
sgemm_tcopy_L4_M4_BEGIN:
|
|
|
|
ldr AO1, A // AO1 = A
|
|
lsl r3, LDA, #2 // r3 = 4 * LDA
|
|
add r3, r3 , AO1 // A = A + 4 * LDA
|
|
str r3, A // store A
|
|
|
|
ldr BO1, B
|
|
add r3, BO1, #64 // B = B + 16 * SIZE
|
|
str r3, B
|
|
|
|
asrs I, N, #2 // I = M / 4
|
|
ble sgemm_tcopy_L4_M4_40
|
|
|
|
sgemm_tcopy_L4_M4_20:
|
|
|
|
COPY4x4_1
|
|
|
|
subs I , I , #1
|
|
ble sgemm_tcopy_L4_M4_40
|
|
|
|
COPY4x4_2
|
|
|
|
subs I , I , #1
|
|
bne sgemm_tcopy_L4_M4_20
|
|
|
|
|
|
sgemm_tcopy_L4_M4_40:
|
|
|
|
tst N , #2
|
|
ble sgemm_tcopy_L4_M4_60
|
|
|
|
COPY2x4
|
|
|
|
|
|
sgemm_tcopy_L4_M4_60:
|
|
|
|
tst N, #1
|
|
ble sgemm_tcopy_L4_M4_END
|
|
|
|
COPY1x4
|
|
|
|
|
|
sgemm_tcopy_L4_M4_END:
|
|
|
|
subs J , J, #1 // j--
|
|
bne sgemm_tcopy_L4_M4_BEGIN
|
|
|
|
|
|
|
|
/*********************************************************************************************/
|
|
|
|
sgemm_tcopy_L2_BEGIN:
|
|
|
|
tst M, #3
|
|
ble sgemm_tcopy_L999
|
|
|
|
tst M, #2
|
|
ble sgemm_tcopy_L1_BEGIN
|
|
|
|
sgemm_tcopy_L2_M4_BEGIN:
|
|
|
|
ldr AO1, A // AO1 = A
|
|
lsl r3, LDA, #1 // r3 = 2 * LDA
|
|
add r3, r3 , AO1 // A = A + 2 * LDA
|
|
str r3, A // store A
|
|
|
|
ldr BO1, B
|
|
add r3, BO1, #32 // B = B + 8 * SIZE
|
|
str r3, B
|
|
|
|
asrs I, N, #2 // I = M / 4
|
|
ble sgemm_tcopy_L2_M4_40
|
|
|
|
sgemm_tcopy_L2_M4_20:
|
|
|
|
COPY4x2
|
|
|
|
subs I , I , #1
|
|
bne sgemm_tcopy_L2_M4_20
|
|
|
|
|
|
sgemm_tcopy_L2_M4_40:
|
|
|
|
tst N , #2
|
|
ble sgemm_tcopy_L2_M4_60
|
|
|
|
COPY2x2
|
|
|
|
sgemm_tcopy_L2_M4_60:
|
|
|
|
tst N , #1
|
|
ble sgemm_tcopy_L2_M4_END
|
|
|
|
COPY1x2
|
|
|
|
|
|
sgemm_tcopy_L2_M4_END:
|
|
|
|
|
|
/*********************************************************************************************/
|
|
|
|
sgemm_tcopy_L1_BEGIN:
|
|
|
|
tst M, #1
|
|
ble sgemm_tcopy_L999
|
|
|
|
|
|
sgemm_tcopy_L1_M4_BEGIN:
|
|
|
|
ldr AO1, A // AO1 = A
|
|
add r3, LDA , AO1 // A = A + 1 * LDA
|
|
str r3, A // store A
|
|
|
|
ldr BO1, B
|
|
add r3, BO1, #16 // B = B + 4 * SIZE
|
|
str r3, B
|
|
|
|
asrs I, N, #2 // I = M / 4
|
|
ble sgemm_tcopy_L1_M4_40
|
|
|
|
sgemm_tcopy_L1_M4_20:
|
|
|
|
COPY4x1
|
|
|
|
subs I , I , #1
|
|
bne sgemm_tcopy_L1_M4_20
|
|
|
|
|
|
sgemm_tcopy_L1_M4_40:
|
|
|
|
tst N , #2
|
|
ble sgemm_tcopy_L1_M4_60
|
|
|
|
COPY2x1
|
|
|
|
sgemm_tcopy_L1_M4_60:
|
|
|
|
tst N , #1
|
|
ble sgemm_tcopy_L1_M4_END
|
|
|
|
COPY1x1
|
|
|
|
|
|
sgemm_tcopy_L1_M4_END:
|
|
|
|
|
|
|
|
sgemm_tcopy_L999:
|
|
|
|
sub r3, fp, #128
|
|
vldm r3, { s8 - s15} // restore floating point registers
|
|
|
|
mov r0, #0 // set return value
|
|
sub sp, fp, #24
|
|
pop {r4 - r9, fp}
|
|
bx lr
|
|
|
|
EPILOGUE
|
|
|