trueno-gpu 0.4.33

Pure Rust PTX generation for NVIDIA CUDA - no LLVM, no nvcc
Documentation 
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

//! Batched RoPE kernel for multi-sequence processing

#![allow(clippy::similar_names)]

use crate::kernels::Kernel;
use crate::ptx::builder::{PtxArithmetic, PtxComparison, PtxControl};
use crate::ptx::{PtxKernel, PtxReg, PtxType};

// ============================================================================
// PAR-114: Batched RoPE Kernel (processes M sequences in parallel)
// ============================================================================

/// Batched RoPE Kernel: Apply rotary position embeddings to M sequences
///
/// Processes M sequences in parallel using Grid.y for batch index.
/// Each sequence can have a different position.
///
/// # Grid Configuration
///
/// - Grid: (num_heads, batch_size, 1)
/// - Block: (head_dim / 2, 1, 1)
#[derive(Debug, Clone)]
pub struct BatchedRopeKernel {
    /// Number of heads
    pub num_heads: u32,
    /// Head dimension
    pub head_dim: u32,
    /// Batch size (M)
    pub batch_size: u32,
    /// Rope theta base (typically 10000.0)
    pub theta: f32,
}

impl BatchedRopeKernel {
    /// Create a new batched RoPE kernel
    #[must_use]
    pub fn new(num_heads: u32, head_dim: u32, batch_size: u32, theta: f32) -> Self {
        Self { num_heads, head_dim, batch_size, theta }
    }
}

impl Kernel for BatchedRopeKernel {
    fn name(&self) -> &str {
        "batched_rope"
    }

    fn build_ptx(&self) -> PtxKernel {
        let head_dim = self.head_dim;
        let num_heads = self.num_heads;
        let theta = self.theta;

        PtxKernel::new("batched_rope")
            .param(PtxType::U64, "x_ptr")
            .param(PtxType::U64, "out_ptr")
            .param(PtxType::U64, "positions_ptr")
            .build(move |ctx| {
                let tid = ctx.special_reg(PtxReg::TidX);
                let head_idx = ctx.special_reg(PtxReg::CtaIdX);
                let batch_idx = ctx.special_reg(PtxReg::CtaIdY);

                let x_ptr = ctx.load_param_u64("x_ptr");
                let out_ptr = ctx.load_param_u64("out_ptr");
                let positions_ptr = ctx.load_param_u64("positions_ptr");

                let pair_idx = tid;

                let half_dim = ctx.mov_u32_imm(head_dim / 2);
                let in_bounds = ctx.setp_lt_u32(pair_idx, half_dim);
                ctx.branch_if_not(in_bounds, "exit");

                let four = ctx.mov_u32_imm(4);
                let pos_byte_offset = ctx.mul_lo_u32(batch_idx, four);
                let pos_byte_offset_64 = ctx.cvt_u64_u32(pos_byte_offset);
                let pos_addr = ctx.add_u64(positions_ptr, pos_byte_offset_64);
                let pos = ctx.ld_global_u32(pos_addr);

                let two = ctx.mov_u32_imm(2);
                let elem0 = ctx.mul_lo_u32(pair_idx, two);
                let one = ctx.mov_u32_imm(1);
                let elem1 = ctx.add_u32_reg(elem0, one);

                let heads_per_seq = ctx.mov_u32_imm(num_heads);
                let dim = ctx.mov_u32_imm(head_dim);
                let seq_stride = ctx.mul_lo_u32(heads_per_seq, dim);
                let batch_offset = ctx.mul_lo_u32(batch_idx, seq_stride);
                let head_offset = ctx.mul_lo_u32(head_idx, dim);
                let base_offset = ctx.add_u32_reg(batch_offset, head_offset);
                let offset0 = ctx.add_u32_reg(base_offset, elem0);
                let offset1 = ctx.add_u32_reg(base_offset, elem1);

                let bytes0 = ctx.mul_lo_u32(offset0, four);
                let bytes1 = ctx.mul_lo_u32(offset1, four);
                let bytes0_64 = ctx.cvt_u64_u32(bytes0);
                let bytes1_64 = ctx.cvt_u64_u32(bytes1);

                let addr0 = ctx.add_u64(x_ptr, bytes0_64);
                let addr1 = ctx.add_u64(x_ptr, bytes1_64);
                let out_addr0 = ctx.add_u64(out_ptr, bytes0_64);
                let out_addr1 = ctx.add_u64(out_ptr, bytes1_64);

                let x0 = ctx.ld_global_f32(addr0);
                let x1 = ctx.ld_global_f32(addr1);

                let pair_f32 = ctx.cvt_f32_u32(pair_idx);
                let dim_f32 = ctx.mov_f32_imm(head_dim as f32);
                let neg_two = ctx.mov_f32_imm(-2.0);
                let exponent = ctx.mul_f32(pair_f32, neg_two);
                let exponent_scaled = ctx.div_f32(exponent, dim_f32);
                let log2_theta = ctx.mov_f32_imm(theta.log2());
                let power = ctx.mul_f32(exponent_scaled, log2_theta);
                let freq_base = ctx.ex2_f32(power);

                let pos_f32 = ctx.cvt_f32_u32(pos);
                let angle = ctx.mul_f32(pos_f32, freq_base);

                let cos_val = ctx.cos_f32(angle);
                let sin_val = ctx.sin_f32(angle);

                let x0_cos = ctx.mul_f32(x0, cos_val);
                let x1_sin = ctx.mul_f32(x1, sin_val);
                let new_x0 = ctx.sub_f32(x0_cos, x1_sin);

                let x0_sin = ctx.mul_f32(x0, sin_val);
                let x1_cos = ctx.mul_f32(x1, cos_val);
                let new_x1 = ctx.add_f32(x0_sin, x1_cos);

                ctx.st_global_f32(out_addr0, new_x0);
                ctx.st_global_f32(out_addr1, new_x1);

                ctx.label("exit");
                ctx.ret();
            })
    }
}

// ============================================================================
// ALB-106: Batched RoPE Backward Kernel (inverse rotation for gradient flow)
// ============================================================================

/// Batched RoPE Backward Kernel: Apply inverse rotary embeddings to gradients
///
/// Identical to `BatchedRopeKernel` but applies the transpose rotation R^T(-θ):
/// - Forward: x0' = x0*cos - x1*sin, x1' = x0*sin + x1*cos
/// - Backward: x0' = x0*cos + x1*sin, x1' = -x0*sin + x1*cos
///
/// # Grid Configuration
///
/// - Grid: (num_heads, batch_size, 1)
/// - Block: (head_dim / 2, 1, 1)
#[derive(Debug, Clone)]
pub struct BatchedRopeBackwardKernel {
    /// Number of heads
    pub num_heads: u32,
    /// Head dimension
    pub head_dim: u32,
    /// Batch size (M)
    pub batch_size: u32,
    /// Rope theta base (typically 10000.0)
    pub theta: f32,
}

impl BatchedRopeBackwardKernel {
    /// Create a new batched RoPE backward kernel
    #[must_use]
    pub fn new(num_heads: u32, head_dim: u32, batch_size: u32, theta: f32) -> Self {
        Self { num_heads, head_dim, batch_size, theta }
    }
}

impl Kernel for BatchedRopeBackwardKernel {
    fn name(&self) -> &str {
        "batched_rope_backward"
    }

    fn build_ptx(&self) -> PtxKernel {
        let head_dim = self.head_dim;
        let num_heads = self.num_heads;
        let theta = self.theta;

        PtxKernel::new("batched_rope_backward")
            .param(PtxType::U64, "x_ptr")
            .param(PtxType::U64, "out_ptr")
            .param(PtxType::U64, "positions_ptr")
            .build(move |ctx| {
                let tid = ctx.special_reg(PtxReg::TidX);
                let head_idx = ctx.special_reg(PtxReg::CtaIdX);
                let batch_idx = ctx.special_reg(PtxReg::CtaIdY);

                let x_ptr = ctx.load_param_u64("x_ptr");
                let out_ptr = ctx.load_param_u64("out_ptr");
                let positions_ptr = ctx.load_param_u64("positions_ptr");

                let pair_idx = tid;

                let half_dim = ctx.mov_u32_imm(head_dim / 2);
                let in_bounds = ctx.setp_lt_u32(pair_idx, half_dim);
                ctx.branch_if_not(in_bounds, "exit");

                let four = ctx.mov_u32_imm(4);
                let pos_byte_offset = ctx.mul_lo_u32(batch_idx, four);
                let pos_byte_offset_64 = ctx.cvt_u64_u32(pos_byte_offset);
                let pos_addr = ctx.add_u64(positions_ptr, pos_byte_offset_64);
                let pos = ctx.ld_global_u32(pos_addr);

                let two = ctx.mov_u32_imm(2);
                let elem0 = ctx.mul_lo_u32(pair_idx, two);
                let one = ctx.mov_u32_imm(1);
                let elem1 = ctx.add_u32_reg(elem0, one);

                let heads_per_seq = ctx.mov_u32_imm(num_heads);
                let dim = ctx.mov_u32_imm(head_dim);
                let seq_stride = ctx.mul_lo_u32(heads_per_seq, dim);
                let batch_offset = ctx.mul_lo_u32(batch_idx, seq_stride);
                let head_offset = ctx.mul_lo_u32(head_idx, dim);
                let base_offset = ctx.add_u32_reg(batch_offset, head_offset);
                let offset0 = ctx.add_u32_reg(base_offset, elem0);
                let offset1 = ctx.add_u32_reg(base_offset, elem1);

                let bytes0 = ctx.mul_lo_u32(offset0, four);
                let bytes1 = ctx.mul_lo_u32(offset1, four);
                let bytes0_64 = ctx.cvt_u64_u32(bytes0);
                let bytes1_64 = ctx.cvt_u64_u32(bytes1);

                let addr0 = ctx.add_u64(x_ptr, bytes0_64);
                let addr1 = ctx.add_u64(x_ptr, bytes1_64);
                let out_addr0 = ctx.add_u64(out_ptr, bytes0_64);
                let out_addr1 = ctx.add_u64(out_ptr, bytes1_64);

                let x0 = ctx.ld_global_f32(addr0);
                let x1 = ctx.ld_global_f32(addr1);

                let pair_f32 = ctx.cvt_f32_u32(pair_idx);
                let dim_f32 = ctx.mov_f32_imm(head_dim as f32);
                let neg_two = ctx.mov_f32_imm(-2.0);
                let exponent = ctx.mul_f32(pair_f32, neg_two);
                let exponent_scaled = ctx.div_f32(exponent, dim_f32);
                let log2_theta = ctx.mov_f32_imm(theta.log2());
                let power = ctx.mul_f32(exponent_scaled, log2_theta);
                let freq_base = ctx.ex2_f32(power);

                let pos_f32 = ctx.cvt_f32_u32(pos);
                let angle = ctx.mul_f32(pos_f32, freq_base);

                let cos_val = ctx.cos_f32(angle);
                let sin_val = ctx.sin_f32(angle);

                // Inverse rotation: negate sin terms (R^T = R(-θ))
                let x0_cos = ctx.mul_f32(x0, cos_val);
                let x1_sin = ctx.mul_f32(x1, sin_val);
                let new_x0 = ctx.add_f32(x0_cos, x1_sin); // + instead of -

                let x0_sin = ctx.mul_f32(x0, sin_val);
                let x1_cos = ctx.mul_f32(x1, cos_val);
                let new_x1 = ctx.sub_f32(x1_cos, x0_sin); // swapped: cos*x1 - sin*x0

                ctx.st_global_f32(out_addr0, new_x0);
                ctx.st_global_f32(out_addr1, new_x1);

                ctx.label("exit");
                ctx.ret();
            })
    }
}