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
//use crate::common::validate_inputs;
use crate::common::validate_options;
use crate::common_simd::assets::validate_inputs;
use crate::indicators::atr::{
min_data, multiplier, output_length, IndicatorState, State, INPUTS_WIDTH, OPTIONS_WIDTH,
};
use crate::indicators::simd_indicators::atr_simd::SimdState;
use crate::indicators::simd_indicators::road_train::{Asset, Driver, PrimeMover};
use crate::indicators::tr::output_length as tr_output_length;
use crate::types::IndicatorError;
use std::simd::Simd;
/// SIMD driver that advances the Average True Range (ATR) across `N` asset lanes per scheduling
/// epoch.
struct AtrDriver {
/// Pre-computed Wilder smoothing multiplier for the given period.
multipliers: (f64, f64),
/// Whether to also emit the raw True Range (TR) output.
want_optional_outputs: bool,
}
impl Driver<State> for AtrDriver {
/// Processes one epoch of bars for `N` assets simultaneously using SIMD.
///
/// Reads from `inputs[asset][field]` (high, low, close), writes to
/// `outputs[asset][output]`, and updates `states[asset]` in place.
fn next_run<const N: usize>(
&mut self,
inputs: Vec<Vec<&[f64]>>,
mut outputs: Vec<Vec<&mut [f64]>>,
mut states: Vec<&mut State>,
_options: Vec<Option<&()>>,
) {
let mut state = SimdState::<N>::new(&states);
let len = inputs[0][0].len();
let multipliers = (
Simd::splat(self.multipliers.0),
Simd::splat(self.multipliers.1),
);
//collect outputs
let (atr_line_ptr, tr_line_ptr) =
crate::extract_output_ptrs!(outputs, N, atr_line_ptr, tr_line_ptr);
// Optimization 2: Pre-compute all input and output pointers
let (high_ptrs, low_ptrs, close_ptrs) =
crate::extract_input_ptrs!(inputs, N, high_ptrs, low_ptrs, close_ptrs);
// Optimization 3: Simplified main loop with pre-computed offsets
for i in 0..len {
// Get inputs arrays for stocks
let (high, low, close) = crate::extract_simd_inputs_at_index!(
i,
N,
high @ high_ptrs,
low @ low_ptrs,
close @ close_ptrs
);
let (atr, tr) = state.calc_simd(high, low, close, multipliers);
// Store results using pre-computed pointers
crate::write_simd_at_indices!(N, i,
atr_line_ptr => atr
);
crate::store_simd_optional_outputs!(i, N,
self.want_optional_outputs, tr_line_ptr => tr
);
}
// Update states efficiently
state.write_states(&mut states);
}
}
/// Calculates the Average True Range (ATR) for `N` assets simultaneously using SIMD
/// parallelism.
///
/// ATR smooths the True Range over a rolling period using Wilder's smoothing method.
/// All assets share the same `options`. Uses the [`PrimeMover`] scheduler to batch assets
/// into SIMD-width groups.
///
/// # Arguments
/// * `inputs` - An array of `N` asset input sets; `inputs[i]` is `[&[f64]; INPUTS_WIDTH]`
/// containing `[high, low, close]` for asset `i`.
/// * `options` - Shared options applied to all `N` assets: `[period]`.
/// * `optional_outputs` - Optional output flags: `[want_tr]`.
///
/// # Returns
/// `Ok((outputs, states))` where `outputs[i]` contains `[atr, tr?]`
/// for asset `i` and `states[i]` is the final [`IndicatorState`] for asset `i`.
/// Returns `Err(IndicatorError)` if any input is too short or options are invalid.
pub fn indicator_by_assets<const N: usize>(
inputs: &[&[&[f64]; INPUTS_WIDTH]; N], //stock[ fields [ field [f64] ] ]
options: &[f64; OPTIONS_WIDTH],
optional_outputs: Option<&[bool]>,
) -> Result<(Vec<Vec<Vec<f64>>>, Vec<IndicatorState>), IndicatorError> {
validate_inputs::<INPUTS_WIDTH>(inputs, min_data(options))?;
validate_options(options)?;
let period = options[0] as usize;
let multipliers = multiplier(period);
let mut road_train = PrimeMover::<N, State>::new();
let mut want_optional_outputs = false;
let mut output_buffers = Vec::with_capacity(N);
for i in 0..N {
let asset_inputs = vec![
inputs[i][0], // high
inputs[i][1], // low
inputs[i][2], // close
];
let atr_capacity = output_length(inputs[i][0].len(), options);
let atr_line = crate::uninit_vec!(f64, atr_capacity);
let mut tr_line = crate::init_optional_outputs_eff!(
optional_outputs, &[false],
tr_line: tr_output_length(inputs[i][0].len(), options)
);
let state = State::init_state(
inputs[i][0],
inputs[i][1],
inputs[i][2],
period,
&mut tr_line,
false,
);
let mut starts = [0; 2];
starts[1] = crate::slice_outputs_start!(atr_capacity, tr_line);
if i == 0 {
(_, want_optional_outputs) = crate::calc_want_flags!(tr_line);
}
let mut output_buffer = vec![atr_line, tr_line];
//let adosc_len = output_buffer[0].len();
let mut asset_outputs = Vec::with_capacity(output_buffer.len());
for j in 0..output_buffer.len() {
unsafe {
//let slice_len = output_buffer.len() - starts[j];
// Get a mutable reference to the output buffer for this asset
let output_buffer = &mut output_buffer[j];
asset_outputs.push(std::slice::from_raw_parts_mut(
output_buffer.as_mut_ptr().add(starts[j]), //slice from
output_buffer.len() - starts[j], // slice to
));
}
}
road_train.add_asset(Asset::new(
asset_inputs,
asset_outputs,
i,
period,
0,
state,
None,
));
output_buffers.push(output_buffer);
}
let mut driver = AtrDriver {
multipliers,
want_optional_outputs,
};
let states_vec = road_train.drive(&mut driver);
let mut states = Vec::with_capacity(N);
for state in states_vec.into_iter() {
states.push(IndicatorState::new(state, multipliers));
}
Ok((output_buffers, states))
}