generic-library/isa-l
1
0
Fork 0
mirror of https://github.com/intel/isa-l.git synced 2024-12-12 17:33:50 +01:00
Code Issues Projects Releases Wiki Activity

igzip: Optimize two pass first pass

Browse Source 
Change-Id: I45efa8304f929d91238c8173c0fac00ec64ab323
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
This commit is contained in:
Roy Oursler 2017-02-06 19:40:41 -05:00 committed by Greg Tucker
parent 4725626847
commit 7c7272e89c
1 changed files with 115 additions and 154 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

269
igzip/igzip_icf_body.asm
View File

@ -50,55 +50,44 @@ global %1
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
%define tmp2 rcx
%define hash2 rcx
%define file_start rdi
%define file_length r15
%define stream r14
%define f_i r10
%define m_out_buf r11
%define curr_data rax
%define code rax
%define tmp5 rax
%define tmp4 rbx
%define tmp2 rcx
%define dist rbx
%define code2 rbx
%define dist_code2 rbx
%define lit_code2 rbx
%define hash rdx
%define len rdx
%define code_len3 rdx
%define tmp8 rdx
%define dist2 r12
%define dist_code r12
%define tmp1 rsi
%define code_len2 rsi
%define code5 rsi
%define file_start rdi
%define lit_code rsi
%define curr_data2 r8
%define len2 r8
%define tmp6 r8
%define tmp4 r8
%define f_i r10
%define m_out_buf r11
%define f_end_i r12
%define dist2 r12
%define tmp7 r12
%define code4 r12
%define len rdx
%define len_code rdx
%define hash3 rdx
%define tmp3 r13
%define code3 r13
%define stream r14
%define hufftables r15
%define hash rbp
%define hash2 r9
;; GPR r8 & r15 can be used
%define xtmp0 xmm0 ; tmp
%define xtmp1 xmm1 ; tmp
%define xhash xmm2
%define xmask xmm3
%define xdata xmm4
%define ytmp0 ymm0 ; tmp
@ -109,13 +98,13 @@ global %1
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
m_out_end equ 0 ; local variable (8 bytes)
m_out_start equ 8
f_end_i_mem_offset equ 16
gpr_save_mem_offset equ 24 ; gpr save area (8*8 bytes)
xmm_save_mem_offset equ 24 + 8*8 ; xmm save area (4*16 bytes) (16 byte aligned)
stack_size equ 3*8 + 8*8 + 4*16
;;; 8 because stack address is odd multiple of 8 after a function call and
;;; we want it aligned to 16 bytes
@ -163,8 +152,6 @@ skip1:
mov stream, rcx
mov dword [stream + _internal_state_has_eob], 0
MOVDQU xmask, [mask]
; state->bitbuf.set_buf(stream->next_out, stream->avail_out);
mov tmp1, [stream + _level_buf]
mov m_out_buf, [tmp1 + _icf_buf_next]
@ -176,30 +163,27 @@ skip1:
mov [rsp + m_out_end], tmp1
mov hufftables, [stream + _hufftables]
mov file_start, [stream + _next_in]
mov f_i %+ d, dword [stream + _total_in]
sub file_start, f_i
mov f_end_i %+ d, [stream + _avail_in]
add f_end_i, f_i
mov file_length %+ d, [stream + _avail_in]
add file_length, f_i
; f_end_i -= LA;
sub f_end_i, LA
mov [rsp + f_end_i_mem_offset], f_end_i
; if (f_end_i <= 0) continue;
; file_length -= LA;
sub file_length, LA
; if (file_length <= 0) continue;
cmp f_end_i, f_i
cmp file_length, f_i
jle input_end
; for (f_i = f_start_i; f_i < f_end_i; f_i++) {
; for (f_i = f_start_i; f_i < file_length; f_i++) {
MARK __body_compute_hash_ %+ ARCH
MOVDQU xdata, [file_start + f_i]
mov curr_data, [file_start + f_i]
mov tmp3, curr_data
mov tmp6, curr_data
mov tmp4, curr_data
compute_hash hash, curr_data
@ -233,10 +217,10 @@ loop2:
inc f_i
MOVQ tmp6, xdata
shr tmp5, 16
mov tmp8, tmp5
compute_hash tmp6, tmp5
mov tmp2, curr_data
shr curr_data, 16
compute_hash hash, curr_data
and hash %+ d, HASH_MASK
mov dist2 %+ w, f_i %+ w
dec dist2
@ -247,8 +231,9 @@ loop2:
and dist %+ d, (D-1)
neg dist
shr tmp8, 8
compute_hash tmp2, tmp8
shr tmp2, 24
compute_hash hash2, tmp2
and hash2 %+ d, HASH_MASK
and dist2 %+ d, (D-1)
neg dist2
@ -261,19 +246,20 @@ MARK __body_compare_ %+ ARCH
xor len, [tmp1 + dist - 1]
jz compare_loop
MOVD xhash, tmp6 %+ d
PINSRD xhash, tmp2 %+ d, 1
PAND xhash, xhash, xmask
;; Check for len/dist match (>7) with second literal
MOVQ len2, xdata
xor len2, [tmp1 + dist2]
jz compare_loop2
movzx lit_code, curr_data %+ b
shr curr_data, 8
;; Check for len/dist match for first literal
test len %+ d, 0xFFFFFFFF
jz len_dist_huffman_pre
inc word [stream + _internal_state_hist_lit_len + 2*lit_code]
movzx lit_code2, curr_data %+ b
;; Check for len/dist match for second literal
test len2 %+ d, 0xFFFFFFFF
jnz write_lit_bits
@ -284,33 +270,30 @@ len_dist_lit_huffman_pre:
shr len2, 3
len_dist_lit_huffman:
or lit_code, LIT
movnti dword [m_out_buf], lit_code %+ d
neg dist2
%ifndef LONGER_HUFFTABLE
mov tmp4, dist2
get_dist_icf_code tmp4, code4, tmp1 ;; clobbers dist, rcx
%else
get_dist_icf_code dist2, code4, tmp1
%endif
movzx code5, curr_data %+ b
;; get_len_code
add f_i, len2
neg len2
MOVQ tmp5, xdata
shr tmp5, 24
compute_hash tmp4, tmp5
and tmp4, HASH_MASK
get_dist_icf_code dist2, dist_code2, tmp1
;; Setup for updating hash
lea tmp3, [f_i + len2 + 1] ; tmp3 <= k
lea tmp3, [f_i + 1] ; tmp3 <= k
MOVDQU xdata, [file_start + f_i]
mov curr_data, [file_start + f_i]
add file_start, f_i
MOVDQU xdata, [file_start + len2]
mov tmp1, [file_start + len2]
shr curr_data, 24
compute_hash hash3, curr_data
and hash3, HASH_MASK
mov curr_data, tmp1
shr tmp1, 8
sub file_start, f_i
add f_i, len2
MOVD hash %+ d, xhash
PEXTRD hash2 %+ d, xhash, 1
mov [stream + _internal_state_head + 2 * hash], tmp3 %+ w
compute_hash hash, curr_data
@ -318,35 +301,29 @@ len_dist_lit_huffman:
add tmp3,1
mov [stream + _internal_state_head + 2 * hash2], tmp3 %+ w
compute_hash hash2, tmp1
add tmp3, 1
mov [stream + _internal_state_head + 2 * tmp4], tmp3 %+ w
mov [stream + _internal_state_head + 2 * hash3], tmp3 %+ w
neg len2
add len2, 254
or code4, len2
add dist_code2, 254
add dist_code2, len2
inc word [stream + _internal_state_hist_lit_len + 2*code5]
or code5, LIT
inc word [stream + _internal_state_hist_lit_len + 2*len2]
write_dword code5, m_out_buf
write_dword code4, m_out_buf
shr code4, DIST_OFFSET
and code4, 0x1F
inc word [stream + _internal_state_hist_dist + 2*code4]
inc word [stream + _internal_state_hist_lit_len + 2*(len2 + 254)]
movnti dword [m_out_buf + 4], dist_code2 %+ d
add m_out_buf, 8
mov f_end_i, [rsp + f_end_i_mem_offset]
mov curr_data2, curr_data
shr curr_data2, 8
compute_hash hash2, curr_data2
shr dist_code2, DIST_OFFSET
and dist_code2, 0x1F
inc word [stream + _internal_state_hist_dist + 2*dist_code2]
; hash = compute_hash(state->file_start + f_i) & HASH_MASK;
and hash %+ d, HASH_MASK
and hash2 %+ d, HASH_MASK
; continue
cmp f_i, f_end_i
cmp f_i, file_length
jl loop2
jmp input_end
;; encode as dist/len
@ -358,96 +335,78 @@ len_dist_huffman_pre:
len_dist_huffman:
dec f_i
;; Setup for updateing hash
lea tmp3, [f_i + 2] ; tmp3 <= k
neg dist
; get_dist_code(dist, &code2, &code_len2);
%ifndef LONGER_HUFFTABLE
mov tmp3, dist ; since code2 and dist are rbx
get_dist_icf_code tmp3, code2, tmp1 ;; clobbers dist, rcx
%else
get_dist_icf_code dist, code2, tmp1
%endif
get_dist_icf_code dist, dist_code, tmp1
; get_len_code(len, &code, &code_len);
lea code5, [len + 254]
or code2, code5
;; Setup for updateing hash
lea tmp3, [f_i + 2] ; tmp3 <= k
add file_start, f_i
MOVDQU xdata, [file_start + len]
mov curr_data2, [file_start + len]
mov curr_data, curr_data2
sub file_start, f_i
add f_i, len
; get_len_code(len, &code, &code_len);
lea len_code, [len + 254]
or dist_code, len_code
MOVD hash %+ d, xhash
PEXTRD hash2 %+ d, xhash, 1
mov [stream + _internal_state_head + 2 * hash], tmp3 %+ w
add tmp3,1
mov [stream + _internal_state_head + 2 * hash2], tmp3 %+ w
MOVDQU xdata, [file_start + f_i]
mov curr_data, [file_start + f_i]
mov curr_data2, curr_data
compute_hash hash, curr_data
inc word [stream + _internal_state_hist_lit_len + 2*code5]
write_dword code2, m_out_buf
shr code2, DIST_OFFSET
and code2, 0x1F
inc word [stream + _internal_state_hist_dist + 2*code2]
mov f_end_i, [rsp + f_end_i_mem_offset]
shr curr_data2, 8
compute_hash hash2, curr_data2
inc word [stream + _internal_state_hist_lit_len + 2*len_code]
movnti dword [m_out_buf], dist_code %+ d
add m_out_buf, 4
shr dist_code, DIST_OFFSET
and dist_code, 0x1F
inc word [stream + _internal_state_hist_dist + 2*dist_code]
; hash = compute_hash(state->file_start + f_i) & HASH_MASK;
and hash %+ d, HASH_MASK
and hash2 %+ d, HASH_MASK
; continue
cmp f_i, f_end_i
cmp f_i, file_length
jl loop2
jmp input_end
MARK __body_write_lit_bits_ %+ ARCH
write_lit_bits:
movzx code3, curr_data %+ b
shr curr_data, 8
and curr_data, 0xff
mov code2, curr_data
MOVDQU xdata, [file_start + f_i + 1]
mov f_end_i, [rsp + f_end_i_mem_offset]
add f_i, 1
mov curr_data, [file_start + f_i]
MOVQ curr_data, xdata
MOVD hash %+ d, xhash
inc word [stream + _internal_state_hist_lit_len + 2*lit_code2]
inc word [stream + _internal_state_hist_lit_len + 2*code3]
inc word [stream + _internal_state_hist_lit_len + 2*code2]
shl lit_code2, DIST_OFFSET
lea lit_code, [lit_code + lit_code2 + (31 << DIST_OFFSET)]
add code2, 31
shl code2, DIST_OFFSET
or code3, code2
write_dword code3, m_out_buf
PEXTRD hash2 %+ d, xhash, 1
and hash %+ d, HASH_MASK
and hash2 %+ d, HASH_MASK
movnti dword [m_out_buf], lit_code %+ d
add m_out_buf, 4
; continue
cmp f_i, f_end_i
cmp f_i, file_length
jl loop2
input_end:
mov tmp1, ZSTATE_FLUSH_READ_BUFFER
mov tmp5, ZSTATE_BODY
mov tmp2, ZSTATE_BODY
cmp dword [stream + _end_of_stream], 0
cmovne tmp5, tmp1
cmovne tmp2, tmp1
cmp dword [stream + _flush], _NO_FLUSH
cmovne tmp5, tmp1
mov dword [stream + _internal_state_state], tmp5 %+ d
cmovne tmp2, tmp1
mov dword [stream + _internal_state_state], tmp2 %+ d
jmp end
output_end:
@ -455,12 +414,12 @@ output_end:
end:
;; update input buffer
add f_end_i, LA
add file_length, LA
mov [stream + _total_in], f_i %+ d
add file_start, f_i
mov [stream + _next_in], file_start
sub f_end_i, f_i
mov [stream + _avail_in], f_end_i %+ d
sub file_length, f_i
mov [stream + _avail_in], file_length %+ d
;; update output buffer
mov tmp1, [stream + _level_buf]
@ -487,9 +446,6 @@ end:
MARK __body_compare_loops_ %+ ARCH
compare_loop:
MOVD xhash, tmp6 %+ d
PINSRD xhash, tmp2 %+ d, 1
PAND xhash, xhash, xmask
lea tmp2, [tmp1 + dist - 1]
%if (COMPARE_TYPE == 1)
compare250 tmp1, tmp2, len, tmp3
@ -516,6 +472,9 @@ compare_loop2:
%error Unknown Compare type COMPARE_TYPE
% error
%endif
movzx lit_code, curr_data %+ b
shr curr_data, 8
inc word [stream + _internal_state_hist_lit_len + 2*lit_code]
jmp len_dist_lit_huffman
MARK __write_first_byte_ %+ ARCH
@ -528,21 +487,23 @@ write_first_byte:
mov [stream + _internal_state_head + 2 * hash], f_i %+ w
mov hash, hash2
shr tmp6, 16
compute_hash hash2, tmp6
shr tmp4, 16
compute_hash hash2, tmp4
and curr_data, 0xff
inc word [stream + _internal_state_hist_lit_len + 2*curr_data]
or curr_data, LIT
write_dword curr_data, m_out_buf
movnti dword [m_out_buf], curr_data %+ d
add m_out_buf, 4
MOVDQU xdata, [file_start + f_i + 1]
add f_i, 1
mov curr_data, [file_start + f_i]
and hash %+ d, HASH_MASK
and hash2 %+ d, HASH_MASK
cmp f_i, [rsp + f_end_i_mem_offset]
cmp f_i, file_length
jl loop2
jmp input_end