import torch

print(f"PyTorch版本: {torch.__version__}")

print(f"DCU可用: {torch.cuda.is_available()}")

if torch.cuda.is_available():

   # 测试基础张量操作

   a = torch.randn(1000, 1000).cuda()

   b = torch.randn(1000, 1000).cuda()

   print("张量创建成功")

   # 逐步骤测试

   c = a + b  # 先测试简单操作

   print("加法操作成功")

   c = torch.matmul(a, b)  # 再测试矩阵乘法

   print("矩阵乘法成功")

   print("✅ 所有测试通过！")

#   python test.py

上述代码报错：

(dcu_env) [root@r5330 workspace]# python test.py

PyTorch版本: 2.1.0a0

DCU可用: True

张量创建成功

Traceback (most recent call last):

 File "test.py", line 12, in <module>

   c = a + b  # 先测试简单操作

RuntimeError: HIP error: invalid device function

HIP kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.

For debugging consider passing HIP_LAUNCH_BLOCKING=1.

Compile with `TORCH_USE_HIP_DSA` to enable device-side assertions.