访问 Python 中的 GPU 硬件规格？

Question

我想使用 Numba 或类似的 Python CUDA 包访问各种 NVidia GPU 规范。可用设备内存、二级缓存大小、内存时钟频率等信息

通过阅读，我了解到我可以通过 Numba 的 CUDA 设备接口访问部分信息（但不是全部）。

from numba import cuda

device = cuda.get_current_device()
attribs = [s for s in dir(device) if s.isupper()]
for attr in attribs:
    print(attr, '=', getattr(device, attr))

测试机输出：

ASYNC_ENGINE_COUNT = 4
CAN_MAP_HOST_MEMORY = 1
COMPUTE_CAPABILITY = (5, 0)
MAX_BLOCK_DIM_X = 1024
MAX_BLOCK_DIM_Y = 1024
MAX_BLOCK_DIM_Z = 64
MAX_GRID_DIM_X = 2147483647
MAX_GRID_DIM_Y = 65535
MAX_GRID_DIM_Z = 65535
MAX_SHARED_MEMORY_PER_BLOCK = 49152
MAX_THREADS_PER_BLOCK = 1024
MULTIPROCESSOR_COUNT = 3
PCI_BUS_ID = 1
PCI_DEVICE_ID = 0
UNIFIED_ADDRESSING = 1
WARP_SIZE = 32

如您所见，我遗漏了 here 中列出的许多字段，例如 TOTAL_CONSTANT_MEMORY、MAX_SHARED_MEMORY_PER_BLOCK、MEMORY_CLOCK_RATE 和 MAX_THREADS_PER_MULTI_PROCESSOR。

如何在 Python 中查看这些值？

Answer 1

所有这些值都通过 __getattr__ method. You can access them with similar like in this method 延迟设置到设备对象。您需要 dir 不是设备，而是枚举自身：

from numba.cuda.cudadrv import enums
from numba import cuda

device = cuda.get_current_device()
attribs= [name.replace("CU_DEVICE_ATTRIBUTE_", "") for name in dir(enums) if name.startswith("CU_DEVICE_ATTRIBUTE_")]
for attr in attribs:
    print(attr, '=', getattr(device, attr))

访问 Python 中的 GPU 硬件规格？

Access GPU hardware specifications in Python?

python

cuda

gpu

nvidia

numba