/******************************************************************************* Copyright (c) 2019, 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. *******************************************************************************/ #define ASSEMBLER #include "common.h" #define N x0 /* vector length */ #define X x1 /* X vector address */ #define INC_X x2 /* X stride */ #define I x5 /* loop variable */ /******************************************************************************* * Macro definitions *******************************************************************************/ #if !defined(DOUBLE) #define REG0 wzr #define SUMF s0 #define TMPF s1 #define TMPVF {v1.s}[0] #define SZ 4 #else #define REG0 xzr #define SUMF d0 #define TMPF d1 #define TMPVF {v1.d}[0] #define SZ 8 #endif /******************************************************************************/ .macro KERNEL_F1 ldr TMPF, [X], #SZ fadd SUMF, SUMF, TMPF .endm .macro KERNEL_F8 #if !defined(DOUBLE) ld1 {v1.4s, v2.4s}, [X], #32 // Load [X3, X2, X1, X0] fadd v1.4s, v1.4s, v2.4s // [X3+X1, X2+X0] fadd v0.4s, v0.4s, v1.4s // [X3+X1, X2+X0] PRFM PLDL1KEEP, [X, #1024] #else // DOUBLE ld1 {v2.2d, v3.2d, v4.2d, v5.2d}, [X] add X, X, #64 PRFM PLDL1KEEP, [X, #1024] fadd v2.2d, v2.2d, v3.2d fadd v4.2d, v4.2d, v5.2d fadd v0.2d, v0.2d, v2.2d fadd v0.2d, v0.2d, v4.2d #endif .endm .macro KERNEL_F8_FINALIZE #if !defined(DOUBLE) ext v1.16b, v0.16b, v0.16b, #8 fadd v0.2s, v0.2s, v1.2s faddp SUMF, v0.2s #else faddp SUMF, v0.2d #endif .endm .macro INIT_S #if !defined(DOUBLE) lsl INC_X, INC_X, #2 #else lsl INC_X, INC_X, #3 #endif .endm .macro KERNEL_S1 ld1 TMPVF, [X], INC_X fadd SUMF, SUMF, TMPF .endm /******************************************************************************* * End of macro definitions *******************************************************************************/ PROLOGUE fmov SUMF, REG0 #if !defined(DOUBLE) fmov s1, SUMF #else fmov d1, SUMF #endif cmp N, xzr ble .Lsum_kernel_L999 cmp INC_X, xzr ble .Lsum_kernel_L999 cmp INC_X, #1 bne .Lsum_kernel_S_BEGIN .Lsum_kernel_F_BEGIN: asr I, N, #3 cmp I, xzr beq .Lsum_kernel_F1 .Lsum_kernel_F8: KERNEL_F8 subs I, I, #1 bne .Lsum_kernel_F8 KERNEL_F8_FINALIZE .Lsum_kernel_F1: ands I, N, #7 ble .Lsum_kernel_L999 .Lsum_kernel_F10: KERNEL_F1 subs I, I, #1 bne .Lsum_kernel_F10 .Lsum_kernel_L999: ret .Lsum_kernel_S_BEGIN: INIT_S asr I, N, #2 cmp I, xzr ble .Lsum_kernel_S1 .Lsum_kernel_S4: KERNEL_S1 KERNEL_S1 KERNEL_S1 KERNEL_S1 subs I, I, #1 bne .Lsum_kernel_S4 .Lsum_kernel_S1: ands I, N, #3 ble .Lsum_kernel_L999 .Lsum_kernel_S10: KERNEL_S1 subs I, I, #1 bne .Lsum_kernel_S10 ret EPILOGUE