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
use core;
use super::super::alloc;
use super::super::alloc::{SliceWrapper, SliceWrapperMut};
use super::interface;
use super::backward_references::BrotliEncoderParams;
use super::input_pair::{InputPair, InputReference, InputReferenceMut};
use super::ir_interpret::{IRInterpreter, push_base};
use super::util::{floatX};
use super::prior_eval::{Prior,Stride1Prior,init_cdfs, DEFAULT_SPEED, STRIDE_PRIOR_SIZE};
pub struct StrideEval<'a,
AllocU16:alloc::Allocator<u16> + 'a,
AllocU32:alloc::Allocator<u32> + 'a,
AllocF:alloc::Allocator<floatX> +'a,
> {
input: InputPair<'a>,
mf: &'a mut AllocF,
m16: &'a mut AllocU16,
_m32: &'a mut AllocU32,
context_map: &'a interface::PredictionModeContextMap<InputReferenceMut<'a>>,
block_type: u8,
local_byte_offset: usize,
_nop: AllocU32::AllocatedMemory,
stride_priors: [AllocU16::AllocatedMemory;8],
score: AllocF::AllocatedMemory,
cur_score_epoch: usize,
stride_speed: [(u16, u16);2],
cur_stride: u8,
}
impl<'a,
AllocU16:alloc::Allocator<u16>,
AllocU32:alloc::Allocator<u32>,
AllocF:alloc::Allocator<floatX>+'a,
> StrideEval<'a, AllocU16, AllocU32, AllocF> {
pub fn new(m16: &'a mut AllocU16,
_m32: &'a mut AllocU32,
mf: &'a mut AllocF,
input: InputPair<'a>,
prediction_mode: &'a interface::PredictionModeContextMap<InputReferenceMut<'a>>,
params: &BrotliEncoderParams,
) -> Self {
let do_alloc = true;
let mut stride_speed = prediction_mode.stride_context_speed();
if stride_speed[0] == (0, 0) {
stride_speed[0] = params.literal_adaptation[0]
}
if stride_speed[0] == (0, 0) {
stride_speed[0] = DEFAULT_SPEED;
}
if stride_speed[1] == (0, 0) {
stride_speed[1] = params.literal_adaptation[1]
}
if stride_speed[1] == (0, 0) {
stride_speed[1] = stride_speed[0];
}
let score = if do_alloc {
mf.alloc_cell(8 * 4)
} else {
AllocF::AllocatedMemory::default()
};
let stride_priors = if do_alloc {
[m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
m16.alloc_cell(STRIDE_PRIOR_SIZE),
]
} else {
[AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
AllocU16::AllocatedMemory::default(),
]
};
let mut ret = StrideEval::<AllocU16, AllocU32, AllocF>{
input: input,
context_map: prediction_mode,
block_type: 0,
m16: m16,
_m32: _m32,
mf: mf,
cur_stride: 1,
cur_score_epoch: 0,
local_byte_offset: 0,
_nop: AllocU32::AllocatedMemory::default(),
stride_priors: stride_priors,
score: score,
stride_speed: stride_speed,
};
for stride_prior in ret.stride_priors.iter_mut() {
init_cdfs(stride_prior.slice_mut());
}
ret
}
pub fn choose_stride(&self, stride_data: &mut[u8]) {
assert_eq!(stride_data.len(), self.cur_score_epoch);
assert!(self.score.slice().len() > stride_data.len());
assert!(self.score.slice().len() > (stride_data.len() << 3) + 7 + 8);
for (index, choice) in stride_data.iter_mut().enumerate() {
let choices = self.score.slice().split_at((1 + index) << 3).1.split_at(8).0;
let mut best_choice: u8 = 0;
let mut best_score = choices[0];
for (cur_index, cur_score) in choices.iter().enumerate() {
if *cur_score + 2.0 < best_score {
best_score = *cur_score;
best_choice = cur_index as u8;
}
}
*choice = best_choice;
}
}
pub fn num_types(&self) -> usize {
self.cur_score_epoch
}
fn update_cost_base(&mut self, stride_prior: [u8;8], selected_bits: u8, cm_prior: usize, literal: u8) {
type CurPrior = Stride1Prior;
{
for i in 0..8 {
let mut cdf = CurPrior::lookup_mut(self.stride_priors[i].slice_mut(),
stride_prior[i], selected_bits, cm_prior, None);
self.score.slice_mut()[self.cur_score_epoch * 8 + i] += cdf.cost(literal>>4);
cdf.update(literal >> 4, self.stride_speed[1]);
}
}
{
for i in 0..8 {
let mut cdf = CurPrior::lookup_mut(self.stride_priors[i].slice_mut(),
stride_prior[i], selected_bits, cm_prior, Some(literal >> 4));
self.score.slice_mut()[self.cur_score_epoch * 8 + i] += cdf.cost(literal&0xf);
cdf.update(literal&0xf, self.stride_speed[0]);
}
}
}
}
impl<'a, AllocU16: alloc::Allocator<u16>,
AllocU32:alloc::Allocator<u32>,
AllocF: alloc::Allocator<floatX>> Drop for StrideEval<'a, AllocU16, AllocU32, AllocF> {
fn drop(&mut self) {
self.mf.free_cell(core::mem::replace(&mut self.score, AllocF::AllocatedMemory::default()));
for i in 0..8 {
self.m16.free_cell(core::mem::replace(&mut self.stride_priors[i], AllocU16::AllocatedMemory::default()));
}
}
}
impl<'a, AllocU16: alloc::Allocator<u16>,
AllocU32:alloc::Allocator<u32>,
AllocF: alloc::Allocator<floatX>> IRInterpreter for StrideEval<'a, AllocU16, AllocU32, AllocF> {
fn inc_local_byte_offset(&mut self, inc: usize) {
self.local_byte_offset += inc;
}
fn local_byte_offset(&self) -> usize {
self.local_byte_offset
}
fn update_block_type(&mut self, new_type: u8, stride: u8) {
self.block_type = new_type;
self.cur_stride = stride;
self.cur_score_epoch += 1;
if self.cur_score_epoch * 8 + 7 >= self.score.slice().len() {
let new_len = self.score.slice().len() * 2;
let mut new_score = self.mf.alloc_cell(new_len);
for (src, dst) in self.score.slice().iter().zip(new_score.slice_mut().split_at_mut(self.score.slice().len()).0.iter_mut()) {
*dst = *src;
}
self.mf.free_cell(core::mem::replace(&mut self.score, new_score));
}
}
fn block_type(&self) -> u8 {
self.block_type
}
fn literal_data_at_offset(&self, index:usize) -> u8 {
self.input[index]
}
fn literal_context_map(&self) -> &[u8] {
self.context_map.literal_context_map.slice()
}
fn prediction_mode(&self) -> ::interface::LiteralPredictionModeNibble {
self.context_map.literal_prediction_mode()
}
fn update_cost(&mut self, stride_prior: [u8;8], stride_prior_offset: usize , selected_bits: u8, cm_prior: usize, literal: u8) {
let reversed_stride_priors = [stride_prior[stride_prior_offset&7],
stride_prior[stride_prior_offset.wrapping_sub(1)&7],
stride_prior[stride_prior_offset.wrapping_sub(2)&7],
stride_prior[stride_prior_offset.wrapping_sub(3)&7],
stride_prior[stride_prior_offset.wrapping_sub(4)&7],
stride_prior[stride_prior_offset.wrapping_sub(5)&7],
stride_prior[stride_prior_offset.wrapping_sub(6)&7],
stride_prior[stride_prior_offset.wrapping_sub(7)&7]];
self.update_cost_base(reversed_stride_priors, selected_bits, cm_prior, literal)
}
}
impl<'a, 'b, AllocU16: alloc::Allocator<u16>,
AllocU32:alloc::Allocator<u32>,
AllocF: alloc::Allocator<floatX>> interface::CommandProcessor<'b> for StrideEval<'a, AllocU16, AllocU32, AllocF> {
fn push<Cb: FnMut(&[interface::Command<InputReference>])>(&mut self,
val: interface::Command<InputReference<'b>>,
callback: &mut Cb) {
push_base(self, val, callback)
}
}