finish range-clip-gate #67

S1/uucoco_#67/prompt.txt

@@ -0,0 +1,77 @@
+
+You have complete freedom to choose the set of operators you want to replace. You may make the decision to replace some operators with custom CUDA kernels and leave others unchanged. You may replace multiple operators with custom implementations, consider operator fusion opportunities (combining multiple operators into a single kernel, for example, combining chunk+gelu+elementwise_mul), or algorithmic changes (such as optimized memory access patterns). You are only limited by your imagination.
+
+CUDA kernel for Range Clip Gate activation with vectorized elementwise operations.
+
+Optimizations:
+
+Vectorized Memory Operations: Uses float4 loads/stores to process 4 elements per instruction.
+
+Coalesced Memory Access: Threads access contiguous memory locations via vectorized operations.
+
+Fast Math: Compiler flags enable fast approximate expf and sigmoid.
+
+Elementwise operation:
+
+Sigmoid gate: gate = sigmoid(x)
+
+Gated multiplication: gated = x * gate
+
+Range clipping: output = clamp(gated, clip_min, clip_max)
+
+Mathematically:
+output = clamp(x·sigmoid(x), min, max)
+
+Characteristics:
+
+Sigmoid creates soft gating (0-1 multiplier).
+
+Self-gating: input gates itself.
+
+Hard clipping to predefined range.
+
+Similar to Swish/SiLU but with clipping.
+
+Use cases:
+
+Controlled activation magnitude.
+
+Preventing activation explosion.
+
+Self-attention gating mechanisms.
+
+Default clipping: [0.0, 6.0] (similar to ReLU6 but with gating).
+
+Specialized for stable activation with bounded output range and input-dependent gating.
+
+
+
+Here's an example to show you the syntax of inline embedding custom CUDA operators in torch: The example given architecture is:
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class Model(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.clip_min = 0.0
+        self.clip_max = 6.0
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        gate = torch.sigmoid(x)
+        gated_output = x * gate
+        return torch.clamp(gated_output, min=self.clip_min, max=self.clip_max)
+
+
+batch_size = 128
+feature_dim = 512
+
+
+def get_inputs():
+    x = torch.randn(batch_size, feature_dim, dtype=torch.float32)
+    return [x]
+
+
+def get_init_inputs():
+    return []

S1/uucoco_#67/rangeclipgate_cuda.py

@@ -0,0 +1,95 @@
+import torch
+import torch.nn as nn
+from torch.utils.cpp_extension import load_inline
+
+class ModelNew(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self._compile_cuda_kernel()
+        self.clip_min = 0.0
+        self.clip_max = 6.0
+
+    def _compile_cuda_kernel(self):
+        cpp_source = """
+        torch::Tensor range_clip_gate_cuda(torch::Tensor x, float clip_min, float clip_max);
+        """
+
+        cuda_source = """
+        #include <torch/extension.h>
+        #include <cuda_runtime.h>
+        #include <math.h>
+
+        __device__ __forceinline__ float sigmoid_op(float x) {
+            return 1.0f / (1.0f + expf(-x));
+        }
+
+        __device__ __forceinline__ float range_clip_gate_op(float x, float clip_min, float clip_max) {
+            float gate = sigmoid_op(x);
+            float gated_output = x * gate;
+            return fmaxf(clip_min, fminf(clip_max, gated_output));
+        }
+
+        __global__ void range_clip_gate_kernel(
+            const float* __restrict__ x,
+            float* __restrict__ output,
+            const int n_elements,
+            const float clip_min,
+            const float clip_max) 
+        {
+            const int tid = blockIdx.x * blockDim.x + threadIdx.x;
+            const int stride = blockDim.x * gridDim.x;
+
+            const int vec_loops = n_elements >> 2;
+            const float4* x_vec = reinterpret_cast<const float4*>(x);
+            float4* out_vec = reinterpret_cast<float4*>(output);
+
+            for (int i = tid; i < vec_loops; i += stride) {
+                float4 v = x_vec[i];
+                float4 r;
+
+                r.x = range_clip_gate_op(v.x, clip_min, clip_max);
+                r.y = range_clip_gate_op(v.y, clip_min, clip_max);
+                r.z = range_clip_gate_op(v.z, clip_min, clip_max);
+                r.w = range_clip_gate_op(v.w, clip_min, clip_max);
+
+                out_vec[i] = r;
+            }
+
+            const int tail_start = vec_loops << 2;
+            for (int i = tail_start + tid; i < n_elements; i += stride) {
+                output[i] = range_clip_gate_op(x[i], clip_min, clip_max);
+            }
+        }
+
+        torch::Tensor range_clip_gate_cuda(torch::Tensor x, float clip_min, float clip_max) {
+            auto x_c = x.contiguous();
+            const int n_elements = x_c.numel();
+            auto output = torch::empty_like(x_c);
+
+            const int threads = 256;
+            const int max_blocks = 65535;
+            const int blocks = std::min((n_elements + threads * 4 - 1) / (threads * 4), max_blocks);
+
+            range_clip_gate_kernel<<<blocks, threads>>>(
+                x_c.data_ptr<float>(),
+                output.data_ptr<float>(),
+                n_elements,
+                clip_min,
+                clip_max
+            );
+
+            return output;
+        }
+        """
+
+        self.op = load_inline(
+            name="rcg_v1",
+            cpp_sources=cpp_source,
+            cuda_sources=cuda_source,
+            functions=["range_clip_gate_cuda"],
+            extra_cuda_cflags=["-O3", "--use_fast_math"],
+            verbose=False
+        )
+
+    def forward(self, x):
+        return self.op.range_clip_gate_cuda(x, self.clip_min, self.clip_max)

S1/uucoco_#67/rangeclipgate_torch.py

@@ -0,0 +1,28 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class Model(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.clip_min = 0.0
+        self.clip_max = 6.0
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        gate = torch.sigmoid(x)
+        gated_output = x * gate
+        return torch.clamp(gated_output, min=self.clip_min, max=self.clip_max)
+
+
+batch_size = 128
+feature_dim = 512
+
+
+def get_inputs():
+    x = torch.randn(batch_size, feature_dim, dtype=torch.float32)
+    return [x]
+
+
+def get_init_inputs():
+    return []

S1/uucoco_#67/run_code.py

@@ -0,0 +1,77 @@
+###########################################################
+# 性能和精度验证程序
+###########################################################
+import torch
+import torch.nn as nn
+import time
+from rangeclipgate_torch import Model, get_inputs, get_init_inputs
+from rangeclipgate_cuda  import ModelNew
+
+
+def run_benchmark():
+    # 检查 CUDA 是否可用
+    if not torch.cuda.is_available():
+        print("CUDA 不可用，请确保您有可用的 NVIDIA GPU 并已正确安装 PyTorch CUDA 版本。")
+        return
+    else:
+        device = torch.device("cuda")
+
+    # 初始化模型
+    init_inputs = get_init_inputs()
+    init_inputs = [
+        x.cuda(device=device) if isinstance(x, torch.Tensor) else x for x in init_inputs
+    ]
+    inputs = get_inputs()
+    inputs = [
+        x.cuda(device=device) if isinstance(x, torch.Tensor) else x for x in inputs
+    ]
+
+    torch_model = Model(*init_inputs).cuda()
+    cuda_model = ModelNew(*init_inputs).cuda()
+
+    torch_model.eval()
+    cuda_model.eval()
+
+    print("-------------------- 精度对齐验证 --------------------")
+    with torch.no_grad():
+        output_torch = torch_model(*inputs)
+        output_cuda = cuda_model(*inputs)
+
+    precision_flag = torch.allclose(output_torch, output_cuda, rtol=1e-03)
+    if precision_flag:
+        print("✅ 精度对齐：两个模型的输出结果非常接近。")
+    else:
+        print("❌ 精度不一致！")
+
+    print("\n-------------------- 性能加速比测试 --------------------")
+    num_iterations = 100
+
+    # PyTorch 模型计时
+    torch.cuda.synchronize()
+    start_time = time.time()
+    for _ in range(num_iterations):
+        _ = torch_model(*inputs)
+    torch.cuda.synchronize()
+    torch_time = (time.time() - start_time) / num_iterations
+
+    # 自定义 CUDA 内核计时
+    torch.cuda.synchronize()
+    start_time = time.time()
+    for _ in range(num_iterations):
+        _ = cuda_model(*inputs)
+    torch.cuda.synchronize()
+    cuda_time = (time.time() - start_time) / num_iterations
+
+    print(f"PyTorch torch.relu 平均执行时间: {torch_time:.6f} 秒")
+    print(f"自定义 CUDA 内核 平均执行时间: {cuda_time:.6f} 秒")
+    speedup = 0
+    if cuda_time > 0:
+        speedup = torch_time / cuda_time
+        print(f"加速比 (Speedup): {speedup:.2f}x")
+    else:
+        print("CUDA 内核执行时间为0，无法计算加速比。")
+    return precision_flag, speedup
+
+
+if __name__ == "__main__":
+    precision_flag, speedup = run_benchmark()