minio/pkg/erasure/gf_5vect_dot_prod_avx.asm
2015-09-05 20:19:43 -07:00

343 lines
8.8 KiB
NASM

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Copyright(c) 2011-2015 Intel Corporation All rights reserved.
;
; Redistribution and use in source and binary forms, with or without
; modification, are permitted provided that the following conditions
; are met:
; * Redistributions of source code must retain the above copyright
; notice, this list of conditions and the following disclaimer.
; * 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.
; * Neither the name of Intel Corporation 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 COPYRIGHT
; OWNER 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.
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;
;;; gf_5vect_dot_prod_avx(len, vec, *g_tbls, **buffs, **dests);
;;;
%ifidn __OUTPUT_FORMAT__, elf64
%define arg0 rdi
%define arg1 rsi
%define arg2 rdx
%define arg3 rcx
%define arg4 r8
%define arg5 r9
%define tmp r11
%define tmp2 r10
%define tmp3 r13 ; must be saved and restored
%define tmp4 r12 ; must be saved and restored
%define tmp5 r14 ; must be saved and restored
%define tmp6 r15 ; must be saved and restored
%define return rax
%define PS 8
%define LOG_PS 3
%define func(x) x:
%macro FUNC_SAVE 0
push r12
push r13
push r14
push r15
%endmacro
%macro FUNC_RESTORE 0
pop r15
pop r14
pop r13
pop r12
%endmacro
%endif
%ifidn __OUTPUT_FORMAT__, macho64
%define arg0 rdi
%define arg1 rsi
%define arg2 rdx
%define arg3 rcx
%define arg4 r8
%define arg5 r9
%define tmp r11
%define tmp2 r10
%define tmp3 r13 ; must be saved and restored
%define tmp4 r12 ; must be saved and restored
%define tmp5 r14 ; must be saved and restored
%define tmp6 r15 ; must be saved and restored
%define return rax
%define PS 8
%define LOG_PS 3
%define func(x) x:
%macro FUNC_SAVE 0
push r12
push r13
push r14
push r15
%endmacro
%macro FUNC_RESTORE 0
pop r15
pop r14
pop r13
pop r12
%endmacro
%endif
%ifidn __OUTPUT_FORMAT__, win64
%define arg0 rcx
%define arg1 rdx
%define arg2 r8
%define arg3 r9
%define arg4 r12 ; must be saved, loaded and restored
%define arg5 r15 ; must be saved and restored
%define tmp r11
%define tmp2 r10
%define tmp3 r13 ; must be saved and restored
%define tmp4 r14 ; must be saved and restored
%define tmp5 rdi ; must be saved and restored
%define tmp6 rsi ; must be saved and restored
%define return rax
%define PS 8
%define LOG_PS 3
%define stack_size 10*16 + 7*8 ; must be an odd multiple of 8
%define arg(x) [rsp + stack_size + PS + PS*x]
%define func(x) proc_frame x
%macro FUNC_SAVE 0
alloc_stack stack_size
save_xmm128 xmm6, 0*16
save_xmm128 xmm7, 1*16
save_xmm128 xmm8, 2*16
save_xmm128 xmm9, 3*16
save_xmm128 xmm10, 4*16
save_xmm128 xmm11, 5*16
save_xmm128 xmm12, 6*16
save_xmm128 xmm13, 7*16
save_xmm128 xmm14, 8*16
save_xmm128 xmm15, 9*16
save_reg r12, 10*16 + 0*8
save_reg r13, 10*16 + 1*8
save_reg r14, 10*16 + 2*8
save_reg r15, 10*16 + 3*8
save_reg rdi, 10*16 + 4*8
save_reg rsi, 10*16 + 5*8
end_prolog
mov arg4, arg(4)
%endmacro
%macro FUNC_RESTORE 0
vmovdqa xmm6, [rsp + 0*16]
vmovdqa xmm7, [rsp + 1*16]
vmovdqa xmm8, [rsp + 2*16]
vmovdqa xmm9, [rsp + 3*16]
vmovdqa xmm10, [rsp + 4*16]
vmovdqa xmm11, [rsp + 5*16]
vmovdqa xmm12, [rsp + 6*16]
vmovdqa xmm13, [rsp + 7*16]
vmovdqa xmm14, [rsp + 8*16]
vmovdqa xmm15, [rsp + 9*16]
mov r12, [rsp + 10*16 + 0*8]
mov r13, [rsp + 10*16 + 1*8]
mov r14, [rsp + 10*16 + 2*8]
mov r15, [rsp + 10*16 + 3*8]
mov rdi, [rsp + 10*16 + 4*8]
mov rsi, [rsp + 10*16 + 5*8]
add rsp, stack_size
%endmacro
%endif
%define len arg0
%define vec arg1
%define mul_array arg2
%define src arg3
%define dest arg4
%define ptr arg5
%define vec_i tmp2
%define dest1 tmp3
%define dest2 tmp4
%define vskip1 tmp5
%define vskip3 tmp6
%define pos return
%ifndef EC_ALIGNED_ADDR
;;; Use Un-aligned load/store
%define XLDR vmovdqu
%define XSTR vmovdqu
%else
;;; Use Non-temporal load/stor
%ifdef NO_NT_LDST
%define XLDR vmovdqa
%define XSTR vmovdqa
%else
%define XLDR vmovntdqa
%define XSTR vmovntdq
%endif
%endif
default rel
[bits 64]
section .text
%define xmask0f xmm15
%define xgft1_lo xmm14
%define xgft1_hi xmm13
%define xgft2_lo xmm12
%define xgft2_hi xmm11
%define xgft3_lo xmm10
%define xgft3_hi xmm9
%define xgft4_lo xmm8
%define xgft4_hi xmm7
%define x0 xmm0
%define xtmpa xmm1
%define xp1 xmm2
%define xp2 xmm3
%define xp3 xmm4
%define xp4 xmm5
%define xp5 xmm6
align 16
global gf_5vect_dot_prod_avx:function
func(gf_5vect_dot_prod_avx)
FUNC_SAVE
sub len, 16
jl .return_fail
xor pos, pos
vmovdqa xmask0f, [mask0f] ;Load mask of lower nibble in each byte
mov vskip1, vec
imul vskip1, 32
mov vskip3, vec
imul vskip3, 96
sal vec, LOG_PS ;vec *= PS. Make vec_i count by PS
mov dest1, [dest]
mov dest2, [dest+PS]
.loop16:
mov tmp, mul_array
xor vec_i, vec_i
vpxor xp1, xp1
vpxor xp2, xp2
vpxor xp3, xp3
vpxor xp4, xp4
vpxor xp5, xp5
.next_vect:
mov ptr, [src+vec_i]
add vec_i, PS
XLDR x0, [ptr+pos] ;Get next source vector
vmovdqu xgft1_lo, [tmp] ;Load array Ax{00}, Ax{01}, ..., Ax{0f}
vmovdqu xgft1_hi, [tmp+16] ; " Ax{00}, Ax{10}, ..., Ax{f0}
vmovdqu xgft2_lo, [tmp+vskip1*1] ;Load array Bx{00}, Bx{01}, ..., Bx{0f}
vmovdqu xgft2_hi, [tmp+vskip1*1+16] ; " Bx{00}, Bx{10}, ..., Bx{f0}
vmovdqu xgft3_lo, [tmp+vskip1*2] ;Load array Cx{00}, Cx{01}, ..., Cx{0f}
vmovdqu xgft3_hi, [tmp+vskip1*2+16] ; " Cx{00}, Cx{10}, ..., Cx{f0}
vmovdqu xgft4_lo, [tmp+vskip3] ;Load array Dx{00}, Dx{01}, ..., Dx{0f}
vmovdqu xgft4_hi, [tmp+vskip3+16] ; " Dx{00}, Dx{10}, ..., Dx{f0}
vpand xtmpa, x0, xmask0f ;Mask low src nibble in bits 4-0
vpsraw x0, x0, 4 ;Shift to put high nibble into bits 4-0
vpand x0, x0, xmask0f ;Mask high src nibble in bits 4-0
vpshufb xgft1_hi, x0 ;Lookup mul table of high nibble
vpshufb xgft1_lo, xtmpa ;Lookup mul table of low nibble
vpxor xgft1_hi, xgft1_lo ;GF add high and low partials
vpxor xp1, xgft1_hi ;xp1 += partial
vpshufb xgft2_hi, x0 ;Lookup mul table of high nibble
vpshufb xgft2_lo, xtmpa ;Lookup mul table of low nibble
vpxor xgft2_hi, xgft2_lo ;GF add high and low partials
vpxor xp2, xgft2_hi ;xp2 += partial
vmovdqu xgft1_lo, [tmp+vskip1*4] ;Load array Ex{00}, Ex{01}, ..., Ex{0f}
vmovdqu xgft1_hi, [tmp+vskip1*4+16] ; " Ex{00}, Ex{10}, ..., Ex{f0}
add tmp, 32
vpshufb xgft3_hi, x0 ;Lookup mul table of high nibble
vpshufb xgft3_lo, xtmpa ;Lookup mul table of low nibble
vpxor xgft3_hi, xgft3_lo ;GF add high and low partials
vpxor xp3, xgft3_hi ;xp3 += partial
vpshufb xgft4_hi, x0 ;Lookup mul table of high nibble
vpshufb xgft4_lo, xtmpa ;Lookup mul table of low nibble
vpxor xgft4_hi, xgft4_lo ;GF add high and low partials
vpxor xp4, xgft4_hi ;xp4 += partial
vpshufb xgft1_hi, x0 ;Lookup mul table of high nibble
vpshufb xgft1_lo, xtmpa ;Lookup mul table of low nibble
vpxor xgft1_hi, xgft1_lo ;GF add high and low partials
vpxor xp5, xgft1_hi ;xp5 += partial
cmp vec_i, vec
jl .next_vect
mov tmp, [dest+2*PS]
mov ptr, [dest+3*PS]
mov vec_i, [dest+4*PS]
XSTR [dest1+pos], xp1
XSTR [dest2+pos], xp2
XSTR [tmp+pos], xp3
XSTR [ptr+pos], xp4
XSTR [vec_i+pos], xp5
add pos, 16 ;Loop on 16 bytes at a time
cmp pos, len
jle .loop16
lea tmp, [len + 16]
cmp pos, tmp
je .return_pass
;; Tail len
mov pos, len ;Overlapped offset length-16
jmp .loop16 ;Do one more overlap pass
.return_pass:
FUNC_RESTORE
mov return, 0
ret
.return_fail:
FUNC_RESTORE
mov return, 1
ret
endproc_frame
section .data
align 16
mask0f: ddq 0x0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f
%macro slversion 4
global %1_slver_%2%3%4
global %1_slver
%1_slver:
%1_slver_%2%3%4:
dw 0x%4
db 0x%3, 0x%2
%endmacro
;;; func core, ver, snum
slversion gf_5vect_dot_prod_avx, 02, 03, 0194