1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; 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_3vect_mad_avx512(len, vec, vec_i, mul_array, src, dest);
;;;
%include "reg_sizes.asm"
%ifdef HAVE_AS_KNOWS_AVX512
%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 return rax
%define func(x) x:
%define FUNC_SAVE
%define FUNC_RESTORE
%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 return rax
%define return.w eax
%define stack_size 16*10 + 3*8
%define arg(x) [rsp + stack_size + PS + PS*x]
%define func(x) proc_frame x
%macro FUNC_SAVE 0
sub rsp, stack_size
vmovdqa [rsp+16*0],xmm6
vmovdqa [rsp+16*1],xmm7
vmovdqa [rsp+16*2],xmm8
vmovdqa [rsp+16*3],xmm9
vmovdqa [rsp+16*4],xmm10
vmovdqa [rsp+16*5],xmm11
vmovdqa [rsp+16*6],xmm12
vmovdqa [rsp+16*7],xmm13
vmovdqa [rsp+16*8],xmm14
vmovdqa [rsp+16*9],xmm15
save_reg r12, 10*16 + 0*8
save_reg r15, 10*16 + 1*8
end_prolog
mov arg4, arg(4)
mov arg5, arg(5)
%endmacro
%macro FUNC_RESTORE 0
vmovdqa xmm6, [rsp+16*0]
vmovdqa xmm7, [rsp+16*1]
vmovdqa xmm8, [rsp+16*2]
vmovdqa xmm9, [rsp+16*3]
vmovdqa xmm10, [rsp+16*4]
vmovdqa xmm11, [rsp+16*5]
vmovdqa xmm12, [rsp+16*6]
vmovdqa xmm13, [rsp+16*7]
vmovdqa xmm14, [rsp+16*8]
vmovdqa xmm15, [rsp+16*9]
mov r12, [rsp + 10*16 + 0*8]
mov r15, [rsp + 10*16 + 1*8]
add rsp, stack_size
%endmacro
%endif
%define PS 8
%define len arg0
%define vec arg1
%define vec_i arg2
%define mul_array arg3
%define src arg4
%define dest1 arg5
%define pos return
%define dest2 mul_array
%define dest3 vec_i
%ifndef EC_ALIGNED_ADDR
;;; Use Un-aligned load/store
%define XLDR vmovdqu8
%define XSTR vmovdqu8
%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 x0 zmm0
%define xtmpa zmm1
%define xtmph1 zmm2
%define xtmpl1 zmm3
%define xtmph2 zmm4
%define xtmpl2 zmm5
%define xtmph3 zmm6
%define xtmpl3 zmm7
%define xgft1_hi zmm8
%define xgft1_lo zmm9
%define xgft1_loy ymm9
%define xgft2_hi zmm10
%define xgft2_lo zmm11
%define xgft2_loy ymm11
%define xgft3_hi zmm12
%define xgft3_lo zmm13
%define xgft3_loy ymm13
%define xd1 zmm14
%define xd2 zmm15
%define xd3 zmm16
%define xmask0f zmm17
align 16
global gf_3vect_mad_avx512:ISAL_SYM_TYPE_FUNCTION
func(gf_3vect_mad_avx512)
%ifidn __OUTPUT_FORMAT__, macho64
global _gf_3vect_mad_avx512:ISAL_SYM_TYPE_FUNCTION
func(_gf_3vect_mad_avx512)
%endif
FUNC_SAVE
sub len, 64
jl .return_fail
xor pos, pos
mov tmp, 0x0f
vpbroadcastb xmask0f, tmp ;Construct mask 0x0f0f0f...
sal vec_i, 5 ;Multiply by 32
sal vec, 5
lea tmp, [mul_array + vec_i]
vmovdqu xgft1_loy, [tmp] ;Load array Ax{00}..{0f}, Ax{00}..{f0}
vmovdqu xgft2_loy, [tmp+vec] ;Load array Bx{00}..{0f}, Bx{00}..{f0}
vmovdqu xgft3_loy, [tmp+2*vec] ;Load array Cx{00}..{0f}, Cx{00}..{f0}
vshufi64x2 xgft1_hi, xgft1_lo, xgft1_lo, 0x55
vshufi64x2 xgft1_lo, xgft1_lo, xgft1_lo, 0x00
vshufi64x2 xgft2_hi, xgft2_lo, xgft2_lo, 0x55
vshufi64x2 xgft2_lo, xgft2_lo, xgft2_lo, 0x00
vshufi64x2 xgft3_hi, xgft3_lo, xgft3_lo, 0x55
vshufi64x2 xgft3_lo, xgft3_lo, xgft3_lo, 0x00
mov dest2, [dest1+PS] ; reuse mul_array
mov dest3, [dest1+2*PS] ; reuse vec_i
mov dest1, [dest1]
mov tmp, -1
kmovq k1, tmp
.loop64:
XLDR x0, [src+pos] ;Get next source vector
XLDR xd1, [dest1+pos] ;Get next dest vector
XLDR xd2, [dest2+pos] ;Get next dest vector
XLDR xd3, [dest3+pos] ;Get next dest vector
vpandq xtmpa, x0, xmask0f ;Mask low src nibble in bits 4-0
vpsraw x0, x0, 4 ;Shift to put high nibble into bits 4-0
vpandq x0, x0, xmask0f ;Mask high src nibble in bits 4-0
; dest1
vpshufb xtmph1 {k1}{z}, xgft1_hi, x0 ;Lookup mul table of high nibble
vpshufb xtmpl1 {k1}{z}, xgft1_lo, xtmpa ;Lookup mul table of low nibble
vpxorq xtmph1, xtmph1, xtmpl1 ;GF add high and low partials
vpxorq xd1, xd1, xtmph1 ;xd1 += partial
; dest2
vpshufb xtmph2 {k1}{z}, xgft2_hi, x0 ;Lookup mul table of high nibble
vpshufb xtmpl2 {k1}{z}, xgft2_lo, xtmpa ;Lookup mul table of low nibble
vpxorq xtmph2, xtmph2, xtmpl2 ;GF add high and low partials
vpxorq xd2, xd2, xtmph2 ;xd2 += partial
; dest3
vpshufb xtmph3 {k1}{z}, xgft3_hi, x0 ;Lookup mul table of high nibble
vpshufb xtmpl3 {k1}{z}, xgft3_lo, xtmpa ;Lookup mul table of low nibble
vpxorq xtmph3, xtmph3, xtmpl3 ;GF add high and low partials
vpxorq xd3, xd3, xtmph3 ;xd2 += partial
XSTR [dest1+pos], xd1
XSTR [dest2+pos], xd2
XSTR [dest3+pos], xd3
add pos, 64 ;Loop on 64 bytes at a time
cmp pos, len
jle .loop64
lea tmp, [len + 64]
cmp pos, tmp
je .return_pass
;; Tail len
mov pos, (1 << 63)
lea tmp, [len + 64 - 1]
and tmp, 63
sarx pos, pos, tmp
kmovq k1, pos
mov pos, len ;Overlapped offset length-64
jmp .loop64 ;Do one more overlap pass
.return_pass:
mov return, 0
FUNC_RESTORE
ret
.return_fail:
mov return, 1
FUNC_RESTORE
ret
endproc_frame
%else
%ifidn __OUTPUT_FORMAT__, win64
global no_gf_3vect_mad_avx512
no_gf_3vect_mad_avx512:
%endif
%endif ; ifdef HAVE_AS_KNOWS_AVX512
|
