以下是10個使用Numba CUDA進行編程的例子，這些例子涵蓋了基本的向量加法、矩陣乘法以及其他一些常見操作：

1. 向量加法

from numba import cuda
import numpy as np
@cuda.jit
def vector_add(a, b, c):i = cuda.grid(1)if i < len(a):c[i] = a[i] + b[i]
a = np.array([1, 2, 3], dtype=np.float32)
b = np.array([4, 5, 6], dtype=np.float32)
c = np.zeros_like(a)
threads_per_block = 32
blocks_per_grid = (len(a) + (threads_per_block - 1)) // threads_per_block
vector_add[blocks_per_grid, threads_per_block](a, b, c)
print(c)

2. 矩陣乘法

@cuda.jit
def matmul(A, B, C):i, j = cuda.grid(2)if i < C.shape[0] and j < C.shape[1]:tmp = 0.for k in range(A.shape[1]):tmp += A[i, k] * B[k, j]C[i, j] = tmp
A = np.array([[1, 2], [3, 4]], dtype=np.float32)
B = np.array([[5, 6], [7, 8]], dtype=np.float32)
C = np.zeros((2, 2), dtype=np.float32)
threads_per_block = (16, 16)
blocks_per_grid_x = math.ceil(A.shape[0] / threads_per_block[0])
blocks_per_grid_y = math.ceil(B.shape[1] / threads_per_block[1])
blocks_per_grid = (blocks_per_grid_x, blocks_per_grid_y)
matmul[blocks_per_grid, threads_per_block](A, B, C)
print(C)

3. 向量點乘

@cuda.jit
def dot_product(a, b, result):i = cuda.grid(1)tmp = 0.0if i < a.shape[0]:tmp = a[i] * b[i]cuda.atomic.add(result, 0, tmp)
a = np.array([1, 2, 3], dtype=np.float32)
b = np.array([4, 5, 6], dtype=np.float32)
result = np.zeros(1, dtype=np.float32)
threads_per_block = 32
blocks_per_grid = (len(a) + (threads_per_block - 1)) // threads_per_block
dot_product[blocks_per_grid, threads_per_block](a, b, result)
print(result[0])

4. 向量乘法

@cuda.jit
def vector_multiply(a, b, c):i = cuda.grid(1)if i < len(a):c[i] = a[i] * b[i]
# 同向量加法示例中的初始化和調用方式

5. 向量除法

@cuda.jit
def vector_divide(a, b, c):i = cuda.grid(1)if i < len(a):c[i] = a[i] / b[i]
# 同向量加法示例中的初始化和調用方式

6. 向量求和

@cuda.jit
def vector_sum(a, result):i = cuda.grid(1)tmp = 0.0if i < a.shape[0]:tmp = a[i]cuda.atomic.add(result, 0, tmp)
# 同向量點乘示例中的初始化和調用方式

7. 向量求最大值

@cuda.jit
def vector_max(a, result):i = cuda.grid(1)if i < a.shape[0]:cuda.atomic.max(result, 0, a[i])
# 同向量點乘示例中的初始化和調用方式

8. 向量求最小值

@cuda.jit
def vector_min(a, result):i = cuda.grid(1)if i < a.shape[0]:cuda.atomic.min(result, 0, a[i])
# 同向量點乘示例中的初始化和調用方式

9. 向量排序（冒泡排序）

@cuda.jit
def bubble_sort(arr):i = cuda.grid(1)n = arr.shape[0]if i < n:for j in range(n - 1):if arr[i] > arr[j]:arr[i], arr[j] = arr[j], arr[i]
# 同向量加法示例中的初始化