isa-l/igzip/igzip_set_long_icf_fg_06.asm

373 lines
9.6 KiB
NASM
Raw Normal View History

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Copyright(c) 2011-2018 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.
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
%include "reg_sizes.asm"
%include "lz0a_const.asm"
%include "data_struct2.asm"
%include "igzip_compare_types.asm"
%define NEQ 4
%ifdef HAVE_AS_KNOWS_AVX512
%ifidn __OUTPUT_FORMAT__, win64
%define arg1 rcx
%define arg2 rdx
%define arg3 r8
%define arg4 r9
%define len rdi
%define dist rsi
%else
%define arg1 rdi
%define arg2 rsi
%define arg3 rdx
%define arg4 rcx
%define len r8
%define dist r9
%endif
%define next_in arg1
%define end_processed arg2
%define end_in arg3
%define match_lookup arg4
%define match_in rax
%define match_offset r10
%define tmp1 r11
%define end_processed_orig r12
%define dist_code r13
%define tmp2 r13
%define zmatch_lookup zmm0
%define zmatch_lookup2 zmm1
%define zlens zmm2
%define zdist_codes zmm3
%define zdist_extras zmm4
%define zdists zmm5
%define zdists2 zmm6
%define zlens1 zmm7
%define zlens2 zmm8
%define zlookup zmm9
%define zlookup2 zmm10
%define datas zmm11
%define ztmp1 zmm12
%define ztmp2 zmm13
%define zvect_size zmm16
%define zmax_len zmm17
%define ztwofiftyfour zmm18
%define ztwofiftysix zmm19
%define ztwosixtytwo zmm20
%define znlen_mask zmm21
%define zbswap zmm22
%define zqword_shuf zmm23
%define zdatas_perm3 zmm24
%define zdatas_perm2 zmm25
%define zincrement zmm26
%define zdists_mask zmm27
%define zdists_start zmm28
%define zlong_lens2 zmm29
%define zlong_lens zmm30
%define zlens_mask zmm31
%ifidn __OUTPUT_FORMAT__, win64
%define stack_size 8*16 + 4 * 8 + 8
%define func(x) proc_frame x
%macro FUNC_SAVE 0
alloc_stack stack_size
vmovdqa [rsp + 0*16], xmm6
vmovdqa [rsp + 1*16], xmm7
vmovdqa [rsp + 2*16], xmm8
vmovdqa [rsp + 3*16], xmm9
vmovdqa [rsp + 4*16], xmm10
vmovdqa [rsp + 5*16], xmm11
vmovdqa [rsp + 6*16], xmm12
vmovdqa [rsp + 7*16], xmm13
save_reg rsi, 8*16 + 0*8
save_reg rdi, 8*16 + 1*8
save_reg r12, 8*16 + 2*8
save_reg r13, 8*16 + 3*8
end_prolog
%endm
%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]
mov rsi, [rsp + 8*16 + 0*8]
mov rdi, [rsp + 8*16 + 1*8]
mov r12, [rsp + 8*16 + 2*8]
mov r13, [rsp + 8*16 + 3*8]
add rsp, stack_size
%endm
%else
%define func(x) x: endbranch
%macro FUNC_SAVE 0
push r12
push r13
%endm
%macro FUNC_RESTORE 0
pop r13
pop r12
%endm
%endif
%define VECT_SIZE 16
[bits 64]
default rel
section .text
global set_long_icf_fg_06
func(set_long_icf_fg_06)
endbranch
FUNC_SAVE
lea end_in, [next_in + arg3]
add end_processed, next_in
mov end_processed_orig, end_processed
lea tmp1, [end_processed + LA_STATELESS]
cmp end_in, tmp1
cmovg end_in, tmp1
sub end_processed, 15
vpbroadcastd zlong_lens, [long_len]
vpbroadcastd zlong_lens2, [long_len2]
vpbroadcastd zlens_mask, [len_mask]
vmovdqu16 zdists_start, [dist_start]
vpbroadcastd zdists_mask, [dists_mask]
vmovdqu32 zincrement, [increment]
vbroadcasti64x2 zdatas_perm2, [datas_perm2]
vbroadcasti64x2 zdatas_perm3, [datas_perm3]
vmovdqu64 zqword_shuf, [qword_shuf]
vbroadcasti64x2 zbswap, [bswap_shuf]
vpbroadcastd znlen_mask, [nlen_mask]
vpbroadcastd zvect_size, [vect_size]
vpbroadcastd zmax_len, [max_len]
vpbroadcastd ztwofiftyfour, [twofiftyfour]
vpbroadcastd ztwofiftysix, [twofiftysix]
vpbroadcastd ztwosixtytwo, [twosixtytwo]
vmovdqu32 zmatch_lookup, [match_lookup]
.fill_loop: ; Tahiti is a magical place
vmovdqu32 zmatch_lookup2, zmatch_lookup
vmovdqu32 zmatch_lookup, [match_lookup + ICF_CODE_BYTES * VECT_SIZE]
cmp next_in, end_processed
jae .end_fill
.finish_entry:
vpandd zlens, zmatch_lookup2, zlens_mask
vpcmpgtd k3, zlens, zlong_lens
;; Speculatively increment
add next_in, VECT_SIZE
add match_lookup, ICF_CODE_BYTES * VECT_SIZE
ktestw k3, k3
jz .fill_loop
vpsrld zdist_codes, zmatch_lookup2, DIST_OFFSET
vpmovdw zdists %+ y, zdist_codes ; Relies on perm working mod 32
vpermw zdists, zdists, zdists_start
vpmovzxwd zdists, zdists %+ y
vpsrld zdist_extras, zmatch_lookup2, EXTRA_BITS_OFFSET
vpsubd zdist_extras, zincrement, zdist_extras
vpsubd zdists, zdist_extras, zdists
vextracti32x8 zdists2 %+ y, zdists, 1
kmovb k6, k3
kshiftrw k7, k3, 8
vpgatherdq zlens1 {k6}, [next_in + zdists %+ y - 8]
vpgatherdq zlens2 {k7}, [next_in + zdists2 %+ y - 8]
vmovdqu8 datas %+ y, [next_in - 8]
vpermq zlookup, zdatas_perm2, datas
vpshufb zlookup, zlookup, zqword_shuf
vpermq zlookup2, zdatas_perm3, datas
vpshufb zlookup2, zlookup2, zqword_shuf
vpxorq zlens1, zlens1, zlookup
vpxorq zlens2, zlens2, zlookup2
vpshufb zlens1, zlens1, zbswap
vpshufb zlens2, zlens2, zbswap
vplzcntq zlens1, zlens1
vplzcntq zlens2, zlens2
vpmovqd zlens1 %+ y, zlens1
vpmovqd zlens2 %+ y, zlens2
vinserti32x8 zlens1, zlens2 %+ y, 1
vpsrld zlens1 {k3}{z}, zlens1, 3
vpandd zmatch_lookup2 {k3}{z}, zmatch_lookup2, znlen_mask
vpaddd zmatch_lookup2 {k3}{z}, zmatch_lookup2, ztwosixtytwo
vpaddd zmatch_lookup2 {k3}{z}, zmatch_lookup2, zlens1
vmovdqu32 [match_lookup - ICF_CODE_BYTES * VECT_SIZE] {k3}, zmatch_lookup2
vpcmpgtd k3, zlens1, zlong_lens2
ktestw k3, k3
jz .fill_loop
vpsubd zdists, zincrement, zdists
vpcompressd zdists2 {k3}, zdists
vpcompressd zmatch_lookup2 {k3}, zmatch_lookup2
kmovq match_offset, k3
tzcnt match_offset, match_offset
vmovd dist %+ d, zdists2 %+ x
lea next_in, [next_in + match_offset - VECT_SIZE]
lea match_lookup, [match_lookup + ICF_CODE_BYTES * (match_offset - VECT_SIZE)]
mov match_in, next_in
sub match_in, dist
mov len, 16
mov tmp2, end_in
sub tmp2, next_in
compare_z next_in, match_in, len, tmp2, tmp1, k3, ztmp1, ztmp2
vpbroadcastd zlens1, len %+ d
vpsubd zlens1, zlens1, zincrement
vpaddd zlens1, zlens1, ztwofiftyfour
mov tmp2, end_processed
sub tmp2, next_in
cmp len, tmp2
cmovg len, tmp2
add next_in, len
lea match_lookup, [match_lookup + ICF_CODE_BYTES * len]
vmovdqu32 zmatch_lookup, [match_lookup]
vpbroadcastd zmatch_lookup2, zmatch_lookup2 %+ x
vpandd zmatch_lookup2, zmatch_lookup2, znlen_mask
neg len
.update_match_lookup:
vpandd zlens2, zlens_mask, [match_lookup + ICF_CODE_BYTES * len]
vpcmpgtd k3, zlens1, zlens2
vpcmpgtd k4, zlens1, ztwofiftysix
kandw k3, k3, k4
vpcmpgtd k4, zlens1, zmax_len
vmovdqu32 zlens, zlens1
vmovdqu32 zlens {k4}, zmax_len
vpaddd zlens2 {k3}{z}, zlens, zmatch_lookup2
vmovdqu32 [match_lookup + ICF_CODE_BYTES * len] {k3}, zlens2
knotw k3, k3
ktestw k3, k3
jnz .fill_loop
add len, VECT_SIZE
vpsubd zlens1, zlens1, zvect_size
jmp .update_match_lookup
.end_fill:
mov end_processed, end_processed_orig
cmp next_in, end_processed
jge .finish
mov tmp1, end_processed
sub tmp1, next_in
vpbroadcastd ztmp1, tmp1 %+ d
vpcmpd k3, ztmp1, zincrement, 6
vmovdqu32 zmatch_lookup2 {k3}{z}, zmatch_lookup2
jmp .finish_entry
.finish:
FUNC_RESTORE
ret
endproc_frame
section .data
align 64
;; 64 byte data
dist_start:
dw 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0007, 0x0009, 0x000d
dw 0x0011, 0x0019, 0x0021, 0x0031, 0x0041, 0x0061, 0x0081, 0x00c1
dw 0x0101, 0x0181, 0x0201, 0x0301, 0x0401, 0x0601, 0x0801, 0x0c01
dw 0x1001, 0x1801, 0x2001, 0x3001, 0x4001, 0x6001, 0x0000, 0x0000
qword_shuf:
db 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7
db 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8
db 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9
db 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa
db 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb
db 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc
db 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd
db 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe
db 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf
;; 16 byte data
increment:
dd 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7
dd 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf
datas_perm2:
dq 0x0, 0x1
datas_perm3:
dq 0x1, 0x2
bswap_shuf:
db 0x07, 0x06, 0x05, 0x04, 0x03, 0x02, 0x01, 0x00
db 0x0f, 0x0e, 0x0d, 0x0c, 0x0b, 0x0a, 0x09, 0x08
;; 4 byte data
len_mask:
dd LIT_LEN_MASK
dists_mask:
dd LIT_DIST_MASK
long_len:
dd 0x105
long_len2:
dd 0x7
max_len:
dd 0xfe + 0x102
vect_size:
dd VECT_SIZE
twofiftyfour:
dd 0xfe
twofiftysix:
dd 0x100
twosixtytwo:
dd 0x106
nlen_mask:
dd 0xfffffc00
%endif