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
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
// SPDX-License-Identifier: AGPL-3.0-or-later
//! `TopK` - GPU-accelerated top-K largest values selection
//!
//! **Deep Debt Principles**:
//! - ✅ Pure WGSL implementation (uses existing shader!)
//! - ✅ Safe Rust wrapper (no unsafe code)
//! - ✅ Hardware-agnostic via WebGPU
//! - ✅ Complete implementation (production-ready for inference)
//!
//! ## Algorithm
//!
//! ```text
//! Find indices of top K largest values in tensor
//! Output: [k] indices (as u32)
//! ```
//!
//! **Implementation**: GPU selection (basic O(n*k), parallel sorting for production)
//!
//! **Key Properties**:
//! - Returns indices, not values
//! - Handles duplicates
//! - Stable ordering for equal values
//!
//! **Used By**: Beam search, retrieval, recommendation systems
//!
//! ## Usage
//!
//! ```rust,ignore
//! use barracuda::tensor::Tensor;
//!
//! let scores = Tensor::from_vec(vec![5.0, 1.0, 9.0, 3.0, 7.0], vec![5]).await?;
//! let top3_indices = scores.topk(3)?; // Returns [2, 4, 0] (indices)
//! ```
use crate::device::{DeviceCapabilities, WorkloadType};
use crate::error::{BarracudaError, Result};
use crate::tensor::Tensor;
/// `TopK` parameters for WGSL shader
#[repr(C)]
#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
struct TopKParams {
k: u32,
_pad0: u32,
_pad1: u32,
_pad2: u32,
}
/// `TopK` operation
///
/// **Deep Debt**: Uses existing WGSL shader with selection algorithm
pub struct TopK {
input: Tensor,
k: usize,
}
impl TopK {
/// Create new `TopK` operation
/// **Deep Debt**: Validates K against tensor size
/// # Errors
/// Returns [`Err`] if `k` is zero, `k` exceeds tensor size, or input is not 1D.
pub fn new(input: Tensor, k: usize) -> Result<Self> {
// Validate K
let size = input.len();
if k == 0 {
return Err(BarracudaError::invalid_op("TopK", "k must be positive"));
}
if k > size {
return Err(BarracudaError::invalid_op(
"TopK",
format!("k ({k}) exceeds tensor size ({size})"),
));
}
// TopK currently only works on 1D tensors (flatten for higher dims)
if input.shape().len() != 1 {
return Err(BarracudaError::invalid_op(
"TopK",
"currently only supports 1D tensors (use flatten() first)",
));
}
Ok(Self { input, k })
}
/// WGSL shader source
fn shader() -> &'static str {
{
static SHADER: std::sync::LazyLock<String> = std::sync::LazyLock::new(|| {
include_str!("../shaders/misc/topk_f64.wgsl").to_string()
});
std::sync::LazyLock::force(&SHADER).as_str()
}
}
/// Execute `TopK` (GPU selection)
/// **Deep Debt**: Basic O(n*k) selection, sufficient for moderate K
/// Returns: Tensor of indices [k] as f32 (cast from u32)
/// # Errors
/// Returns [`Err`] if buffer allocation fails, GPU dispatch fails (e.g., device lost),
/// buffer readback fails, or output tensor creation fails.
pub fn execute(self) -> Result<Tensor> {
let device = self.input.device();
// Create parameters
let params = TopKParams {
k: self.k as u32,
_pad0: 0,
_pad1: 0,
_pad2: 0,
};
let params_buffer = device.device.create_buffer(&wgpu::BufferDescriptor {
label: Some("TopK Params"),
size: std::mem::size_of::<TopKParams>() as u64,
usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
mapped_at_creation: false,
});
device
.queue
.write_buffer(¶ms_buffer, 0, bytemuck::bytes_of(¶ms));
// Output buffer (u32 indices)
let output_buffer = device.device.create_buffer(&wgpu::BufferDescriptor {
label: Some("TopK Output"),
size: (self.k * std::mem::size_of::<u32>()) as u64,
usage: wgpu::BufferUsages::STORAGE
| wgpu::BufferUsages::COPY_SRC
| wgpu::BufferUsages::COPY_DST,
mapped_at_creation: false,
});
// Compile shader
let shader = device.compile_shader(Self::shader(), Some("TopK"));
// Create bind group layout
let bgl = device
.device
.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
label: Some("TopK BGL"),
entries: &[
// Input
wgpu::BindGroupLayoutEntry {
binding: 0,
visibility: wgpu::ShaderStages::COMPUTE,
ty: wgpu::BindingType::Buffer {
ty: wgpu::BufferBindingType::Storage { read_only: true },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
// Output
wgpu::BindGroupLayoutEntry {
binding: 1,
visibility: wgpu::ShaderStages::COMPUTE,
ty: wgpu::BindingType::Buffer {
ty: wgpu::BufferBindingType::Storage { read_only: false },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
// Params
wgpu::BindGroupLayoutEntry {
binding: 2,
visibility: wgpu::ShaderStages::COMPUTE,
ty: wgpu::BindingType::Buffer {
ty: wgpu::BufferBindingType::Uniform,
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
],
});
// Create bind group
let bind_group = device.device.create_bind_group(&wgpu::BindGroupDescriptor {
label: Some("TopK BG"),
layout: &bgl,
entries: &[
wgpu::BindGroupEntry {
binding: 0,
resource: self.input.buffer().as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 1,
resource: output_buffer.as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 2,
resource: params_buffer.as_entire_binding(),
},
],
});
// Create pipeline
let pipeline_layout =
device
.device
.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
label: Some("TopK Pipeline Layout"),
bind_group_layouts: &[&bgl],
immediate_size: 0,
});
let pipeline = device
.device
.create_compute_pipeline(&wgpu::ComputePipelineDescriptor {
label: Some("TopK Pipeline"),
layout: Some(&pipeline_layout),
module: &shader,
entry_point: Some("main"),
cache: None,
compilation_options: Default::default(),
});
// Execute
let mut encoder = device.create_encoder_guarded(&wgpu::CommandEncoderDescriptor {
label: Some("TopK Encoder"),
});
{
let mut pass = encoder.begin_compute_pass(&wgpu::ComputePassDescriptor {
label: Some("TopK Pass"),
timestamp_writes: None,
});
pass.set_pipeline(&pipeline);
pass.set_bind_group(0, Some(&bind_group), &[]);
// Deep Debt Evolution: Capability-based dispatch
let caps = DeviceCapabilities::from_device(device);
let optimal_wg_size = caps.optimal_workgroup_size(WorkloadType::Reduction);
let size = self.input.len();
let workgroups = (size as u32).div_ceil(optimal_wg_size);
pass.dispatch_workgroups(workgroups, 1, 1);
}
device.submit_commands(Some(encoder.finish()));
// Read u32 indices back and convert to f32
let staging_buffer = device.device.create_buffer(&wgpu::BufferDescriptor {
label: Some("TopK Staging"),
size: (self.k * std::mem::size_of::<u32>()) as u64,
usage: wgpu::BufferUsages::MAP_READ | wgpu::BufferUsages::COPY_DST,
mapped_at_creation: false,
});
let mut encoder = device.create_encoder_guarded(&wgpu::CommandEncoderDescriptor {
label: Some("TopK Copy Encoder"),
});
encoder.copy_buffer_to_buffer(
&output_buffer,
0,
&staging_buffer,
0,
(self.k * std::mem::size_of::<u32>()) as u64,
);
device.submit_commands(Some(encoder.finish()));
let indices_u32: Vec<u32> = device.map_staging_buffer(&staging_buffer, self.k)?;
let indices_f32: Vec<f32> = indices_u32.iter().map(|&x| x as f32).collect();
// Create output tensor [k] as f32
let output_tensor = Tensor::from_vec_on_sync(indices_f32, vec![self.k], device.clone())?;
Ok(output_tensor)
}
}
// ═══════════════════════════════════════════════════════════════
// TENSOR API INTEGRATION
// ═══════════════════════════════════════════════════════════════
impl Tensor {
/// Top-K largest values (returns indices)
/// **Deep Debt**: Essential for beam search and retrieval
/// # Arguments
/// - `k`: Number of top values to return
/// # Returns
/// - Indices tensor [k] as f32 (cast from u32)
/// # Example
/// ```rust,ignore
/// let scores = Tensor::from_vec(vec![5.0, 1.0, 9.0, 3.0, 7.0], vec![5]).await?;
/// let top3 = scores.topk(3)?; // [2, 4, 0] (indices of 9.0, 7.0, 5.0)
/// ```
/// # Note
/// Currently only supports 1D tensors. Use `flatten()` for higher dimensions.
/// # Errors
/// Returns [`Err`] if validation fails (invalid `k` or shape), buffer allocation fails,
/// GPU dispatch fails (e.g., device lost), or buffer readback fails.
pub fn topk(self, k: usize) -> Result<Self> {
TopK::new(self, k)?.execute()
}
}
// ═══════════════════════════════════════════════════════════════
// TESTS
// ═══════════════════════════════════════════════════════════════
#[cfg(test)]
mod tests {
use super::*;
#[tokio::test]
async fn test_topk_gpu_basic() {
let device = crate::device::test_pool::get_test_device().await;
let input = Tensor::from_vec_on(vec![5.0, 1.0, 9.0, 3.0, 7.0], vec![5], device)
.await
.unwrap();
let top3 = input.topk(3).unwrap();
assert_eq!(top3.shape(), &[3]);
let indices = top3.to_vec().unwrap();
// Should return indices of [9.0, 7.0, 5.0] = [2, 4, 0]
assert_eq!(indices[0] as u32, 2); // 9.0
assert_eq!(indices[1] as u32, 4); // 7.0
assert_eq!(indices[2] as u32, 0); // 5.0
}
#[tokio::test]
async fn test_topk_gpu_single() {
let device = crate::device::test_pool::get_test_device().await;
let input = Tensor::from_vec_on(vec![1.0, 2.0, 3.0, 4.0], vec![4], device)
.await
.unwrap();
let top1 = input.topk(1).unwrap();
let indices = top1.to_vec().unwrap();
// Largest value is 4.0 at index 3
assert_eq!(indices[0] as u32, 3);
}
#[tokio::test]
async fn test_topk_gpu_all() {
let device = crate::device::test_pool::get_test_device().await;
let input = Tensor::from_vec_on(vec![3.0, 1.0, 4.0, 1.0, 5.0], vec![5], device)
.await
.unwrap();
let top5 = input.topk(5).unwrap();
let indices = top5.to_vec().unwrap();
// All indices, sorted by value: [5.0, 4.0, 3.0, 1.0, 1.0] = [4, 2, 0, 1, 3]
assert_eq!(indices[0] as u32, 4); // 5.0
assert_eq!(indices[1] as u32, 2); // 4.0
assert_eq!(indices[2] as u32, 0); // 3.0
}
#[tokio::test]
async fn test_topk_gpu_negative() {
let device = crate::device::test_pool::get_test_device().await;
let input = Tensor::from_vec_on(vec![-5.0, -1.0, -9.0, -3.0], vec![4], device)
.await
.unwrap();
let top2 = input.topk(2).unwrap();
let indices = top2.to_vec().unwrap();
// Largest (least negative): [-1.0, -3.0] at indices [1, 3]
assert_eq!(indices[0] as u32, 1); // -1.0
assert_eq!(indices[1] as u32, 3); // -3.0
}
#[tokio::test]
async fn test_topk_gpu_validation() {
let device = crate::device::test_pool::get_test_device().await;
let input = Tensor::from_vec_on(vec![1.0, 2.0, 3.0], vec![3], device)
.await
.unwrap();
// k=0 should error
assert!(input.clone().topk(0).is_err());
// k > size should error
assert!(input.topk(10).is_err());
}
#[tokio::test]
async fn test_topk_gpu_duplicates() {
let device = crate::device::test_pool::get_test_device().await;
let input = Tensor::from_vec_on(vec![2.0, 5.0, 5.0, 1.0], vec![4], device)
.await
.unwrap();
let top2 = input.topk(2).unwrap();
let indices = top2.to_vec().unwrap();
// Two 5.0 values at indices 1 and 2
// Should return both (order may vary, but both should be 1 or 2)
let idx0 = indices[0] as u32;
let idx1 = indices[1] as u32;
assert!((idx0 == 1 || idx0 == 2) && (idx1 == 1 || idx1 == 2));
}
#[tokio::test]
async fn test_topk_gpu_large() {
let device = crate::device::test_pool::get_test_device().await;
// Larger tensor (100 elements)
let mut values = vec![0.0; 100];
for (i, v) in values.iter_mut().enumerate() {
*v = (i as f32) * 0.1;
}
let input = Tensor::from_vec_on(values, vec![100], device)
.await
.unwrap();
let top10 = input.topk(10).unwrap();
let indices = top10.to_vec().unwrap();
// Should be indices 99, 98, 97, ..., 90 (highest values)
assert_eq!(indices[0] as u32, 99);
assert_eq!(indices[1] as u32, 98);
assert_eq!(indices[9] as u32, 90);
}
}