G1Medium★ Core Essential · Industrial BedrockPart G · Architectures & Parameter-Efficient Fine-Tuning

Conv2d from Scratch with Stride & Padding

⚡Flashcard Drill➔

Industrial-grade implementation and mathematical foundations of Conv2d from Scratch with Stride & Padding.

⏱️ Time Complexity: O(B * C_out * H_out * W_out * C_in * Kh * Kw)
💾 Space Complexity: O(B * C_in * (H+2P) * (W+2P)) 填充后的中间张量
💡

Core Mental Anchor / Mnemonic

Master Conv2d from Scratch with Stride & Padding: enforce numerical stability, check tensor shapes, and eliminate redundant memory allocations.

📐 Mathematical Derivation & Core Formula

Hout=⌊Hin+2P−KhS⌋+1,Ob,cout,h,w=∑cin∑i∑jXb,cin,h⋅S+i,w⋅S+jWcout,cin,i,jH_{\text{out}} = \left\lfloor\frac{H_{\text{in}} + 2P - K_h}{S}\right\rfloor + 1, \quad O_{b, c_{\text{out}}, h, w} = \sum_{c_{\text{in}}} \sum_{i} \sum_{j} X_{b, c_{\text{in}}, h\cdot S + i, w\cdot S + j} W_{c_{\text{out}}, c_{\text{in}}, i, j}
### Mathematical Derivation & Theoretical Principles
Detailed first-principles formulation and architectural mechanics for Conv2d from Scratch with Stride & Padding.

Refer to the LaTeX equation above for the core operator definition. The operator is designed to ensure strict numerical bounds, avoiding floating-point overflows and gradient anomalies.

🔄 Tensor Dimensions & Shape Flow

x: (B, Cin, H, W) -> 零填充 -> (B, Cin, H+2P, W+2P) -> 窗口slice -> 与核相乘累加 -> out: (B, Cout, H_out, W_out)

🛡️ Industrial Numerical Stability & Pitfalls

  • Ensure proper multi-dimensional tensor broadcasting and keepdims retention.
  • Enforce numerical guards (eps clamping and overflow thresholds) during exponentiation and division.
  • Verify train versus eval mode behavioral distinctions (e.g. frozen running statistics and dropout bypass).

💻 Industrial Code Implementation

import numpy as np

def conv2d_forward(
    x: np.ndarray,          # (B, C_in, H, W)
    w: np.ndarray,          # (C_out, C_in, Kh, Kw)
    b: np.ndarray = None,   # (C_out,)
    stride: int = 1,
    padding: int = 0
) -> np.ndarray:
    B, C_in, H, W = x.shape
    C_out, _, Kh, Kw = w.shape
    
    # 1. 计算输出特征图尺寸
    H_out = (H + 2 * padding - Kh) // stride + 1
    W_out = (W + 2 * padding - Kw) // stride + 1
    
    # 2. 空间零填充
    if padding > 0:
        x_pad = np.pad(x, ((0, 0), (0, 0), (padding, padding), (padding, padding)), mode='constant')
    else:
        x_pad = x
        
    out = np.zeros((B, C_out, H_out, W_out), dtype=x.dtype)
    
    # 3. 滑动窗口四层循环计算
    for h in range(H_out):
        h_start = h * stride
        h_end = h_start + Kh
        for w_idx in range(W_out):
            w_start = w_idx * stride
            w_end = w_start + Kw
            
            # 截取感受野切片: (B, C_in, Kh, Kw)
            x_slice = x_pad[:, :, h_start:h_end, w_start:w_end]
            
            # 沿输入通道与核尺寸乘加规约,输出 (B, C_out)
            # x_slice: (B, 1, Cin, Kh, Kw), w: (1, Cout, Cin, Kh, Kw)
            val = np.sum(x_slice[:, np.newaxis, :, :, :] * w[np.newaxis, :, :, :, :], axis=(2, 3, 4))
            
            if b is not None:
                val += b[np.newaxis, :]
                
            out[:, :, h, w_idx] = val
            
    return out

🧪 Runnable Assertions & Validation

Copy and run directly in Python / Jupyter to verify correctness:

import numpy as np
x = np.ones((1, 1, 3, 3))
w = np.ones((1, 1, 2, 2)) # 2x2 全 1 核
out = conv2d_forward(x, w, stride=1, padding=0)
assert out.shape == (1, 1, 2, 2)
assert np.allclose(out, 4.0), "2x2 全 1 卷积结果应全为 4.0"
print("✓ Conv2d 卷积纯手写自测通过")

🎯 Core Architecture Follow-up Q&A

Q1:What are the key trade-offs and memory bottlenecks when deploying Conv2d from Scratch with Stride & Padding in high-throughput inference?
Memory bandwidth (HBM to SRAM I/O) is the primary latency factor. Fusing element-wise operations and avoiding intermediate tensor materialization significantly outperforms naive implementations.
Q2:How does Conv2d from Scratch with Stride & Padding handle extreme numerical boundaries or precision reduction (FP16/BF16/INT8)?
Under low precision, operations must be upcasted to FP32 during accumulation to prevent underflow/overflow, followed by proper scaling and clamping before converting back to the target format.
⚡
Test muscle memory on this topic via 3-Layer Flashcards
Active Recall: Mentally formulate Prompt ➔ Math Derivation ➔ Benchmark Implementation in 3 minutes.
🎯 Drill This Card Now➔
←Prev: Multi-Armed Bandit: epsilon-Greedy & UCB1All 69 KernelsNext: LoRA (Low-Rank Adaptation)→
中文速览与面试要点

Conv2d 卷积算子纯手写(四层循环实现)

计算机视觉底层基石,纯手写前向滑动窗口,清晰展示输出尺寸公式、零填充与通道累加。

核心要点

  • ### 滑动窗口与空间几何投影 二维卷积将输入特征图 X R^B C_ in H W 映射为 O R^B C_ out H_ out W_ out。 - Padding(填充):在输入外围补 P 圈 0,保护边缘空间分辨率; - Stride(步长):卷积核在空间移动的步距 S; 每个输出通道 c_ out 拥有一套独立的 C_ in K_h K_w 权重核,在输入对应的三维感受野区域执行内积并累加求和,最后加上偏置 b。 在硬件底层(cuDNN / TensorRT),通常使用 im2col + GEMM 将滑动窗口展开为二维大矩阵,调用高度优化的矩阵乘法单元计算。
  • x: (B, Cin, H, W) -> 零填充 -> (B, Cin, H+2P, W+2P) -> 窗口切片 -> 与核相乘累加 -> out: (B, Cout, H_out, W_out)
  • 输出尺寸计算必须使用整数向下取整地板除 //,而非浮点除
  • Padding 必须只作用在后两维 (H, W),绝不能在 Batch 或 Channel 维度上补零
  • 工业实现中使用 im2col 将感受野拉平为 (H_out*W_out, Cin*Kh*Kw) 执行矩阵乘,显存翻倍但速度快 10 倍
  • Depthwise Separable Convolution(深度可分离卷积)如何大幅削减 Conv2d 的计算量?:将标准卷积拆分为两阶段:1. Depthwise Convolution:每个输入通道独立进行单通道空间卷积(计算量降为 1/C_ out);2. Pointwise Convolution:使用 1 1 卷积做跨通道线性混合。整体计算量被压缩为标准卷积的 1C_ out + 1K_h K_w(通常节省 80%~90% FLOPS)。

参考实现

import numpy as np

def conv2d_forward(
    x: np.ndarray,          # (B, C_in, H, W)
    w: np.ndarray,          # (C_out, C_in, Kh, Kw)
    b: np.ndarray = None,   # (C_out,)
    stride: int = 1,
    padding: int = 0
) -> np.ndarray:
    B, C_in, H, W = x.shape
    C_out, _, Kh, Kw = w.shape
    
    # 1. 计算输出特征图尺寸
    H_out = (H + 2 * padding - Kh) // stride + 1
    W_out = (W + 2 * padding - Kw) // stride + 1
    
    # 2. 空间零填充
    if padding > 0:
        x_pad = np.pad(x, ((0, 0), (0, 0), (padding, padding), (padding, padding)), mode='constant')
    else:
        x_pad = x
        
    out = np.zeros((B, C_out, H_out, W_out), dtype=x.dtype)
    
    # 3. 滑动窗口四层循环计算
    for h in range(H_out):
        h_start = h * stride
        h_end = h_start + Kh
        for w_idx in range(W_out):
            w_start = w_idx * stride
            w_end = w_start + Kw
            
            # 截取感受野切片: (B, C_in, Kh, Kw)
            x_slice = x_pad[:, :, h_start:h_end, w_start:w_end]
            
            # 沿输入通道与核尺寸乘加规约,输出 (B, C_out)
            # x_slice: (B, 1, Cin, Kh, Kw), w: (1, Cout, Cin, Kh, Kw)
            val = np.sum(x_slice[:, np.newaxis, :, :, :] * w[np.newaxis, :, :, :, :], axis=(2, 3, 4))
            
            if b is not None:
                val += b[np.newaxis, :]
                
            out[:, :, h, w_idx] = val
            
    return out