finish Phish #38

S1/gsd123_#38/Phish_cuda.py

@@ -0,0 +1,93 @@
+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()
+
+    def _compile_cuda_kernel(self):
+        cpp_source = """
+        torch::Tensor phish_cuda(torch::Tensor x);
+        """
+
+        cuda_source = """
+        #include <torch/extension.h>
+        #include <cuda_runtime.h>
+        #include <math.h>
+
+        // Constant: 1 / sqrt(2)
+        #define M_SQRT1_2_F 0.70710678118654752440f
+
+        __device__ __forceinline__ float gelu_op(float x) {
+            // GELU(x) = 0.5 * x * (1 + erf(x / sqrt(2)))
+            return 0.5f * x * (1.0f + erff(x * M_SQRT1_2_F));
+        }
+
+        __device__ __forceinline__ float phish_op(float x) {
+            // Phish(x) = x * tanh(GELU(x))
+            float g = gelu_op(x);
+            return x * tanhf(g);
+        }
+
+        __global__ void phish_kernel(
+            const float* __restrict__ x,
+            float* __restrict__ output,
+            const int n_elements) 
+        {
+            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 = __ldg(&x_vec[i]);
+                float4 r;
+
+                r.x = phish_op(v.x);
+                r.y = phish_op(v.y);
+                r.z = phish_op(v.z);
+                r.w = phish_op(v.w);
+
+                out_vec[i] = r;
+            }
+
+            const int tail_start = vec_loops << 2;
+            for (int i = tail_start + tid; i < n_elements; i += stride) {
+                output[i] = phish_op(x[i]);
+            }
+        }
+
+        torch::Tensor phish_cuda(torch::Tensor x) {
+            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);
+
+            phish_kernel<<<blocks, threads>>>(
+                x_c.data_ptr<float>(),
+                output.data_ptr<float>(),
+                n_elements
+            );
+
+            return output;
+        }
+        """
+
+        self.op = load_inline(
+            name="phish_v1",
+            cpp_sources=cpp_source,
+            cuda_sources=cuda_source,
+            functions=["phish_cuda"],
+            extra_cuda_cflags=["-O3", "--use_fast_math"],
+            verbose=False
+        )
+
+    def forward(self, x):
+        return self.op.phish_cuda(x)

S1/gsd123_#38/Phish_torch.py

@@ -0,0 +1,25 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class Model(nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        # Phish = x * tanh(GELU(x))
+        return x * torch.tanh(F.gelu(x))
+
+
+batch_size = 1024
+feature_dim = 1024
+
+
+def get_inputs():
+    x = torch.randn(batch_size, feature_dim, dtype=torch.float32)
+    return [x]
+
+
+def get_init_inputs():
+    return []

S1/gsd123_#38/prompt.txt

@@ -0,0 +1,85 @@
+You write custom CUDA kernels to replace the pytorch operators in the given GeGLU architecture to get speedups.
+
+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 Optimization Strategies:
+
+Vectorized Memory Access
+
+Uses float4 for 4-element vector loads/stores
+
+__ldg() for read-only caching through texture memory
+
+Bit shifts for division (>> 2, << 2) for efficiency
+
+Phish Activation Function
+
+Computes Phish(x) = x * tanh(GELU(x))
+
+Combination of GELU and tanh activations
+
+Requires nested function evaluations
+
+Optimized GELU Computation
+
+Uses erff for error function approximation
+
+Predefined constant M_SQRT1_2_F (1/√2)
+
+Avoids repeated sqrtf calls
+
+Memory Access
+
+contiguous() tensors for coalescing
+
+__restrict__ pointers
+
+Grid-stride loop for arbitrary sizes
+
+Performance Optimization
+
+Compiler flags: -O3, --use_fast_math
+
+Efficient kernel launch configuration
+
+Block count limited to 65535
+
+Uses tanhf for fast hyperbolic tangent
+
+Mathematical Efficiency
+
+Inline functions for GELU and Phish
+
+Vectorized operations for 4 elements simultaneously
+
+Single pass through data
+
+Key Innovation: Vectorized Phish activation function combining GELU and tanh, optimized with mathematical constants and fast transcendental functions.
+
+
+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__()
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        # Phish = x * tanh(GELU(x))
+        return x * torch.tanh(F.gelu(x))
+
+
+batch_size = 1024
+feature_dim = 1024
+
+
+def get_inputs():
+    x = torch.randn(batch_size, feature_dim, dtype=torch.float32)
+    return [x]
+
+
+def get_init_inputs():
+    return []

S1/gsd123_#38/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 Phish_torch import Model, get_inputs, get_init_inputs
+from Phish_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()