DLPack tensor is not contiguous. Only contiguous DLPack tensors that are stored in C-Order are supported. #834
Description
I use
torch==2.5.1+cu121
tritonclient[all]==2.51.0
when I use get error
cudashm.set_shared_memory_region_from_dlpack(region.handle, [data])
error
tritonclient.utils.cuda_shared_memory._utils.CudaSharedMemoryException: DLPack tensor is not contiguous. Only contiguous DLPack tensors that are stored in C-Order are supported.
I print my torch tensor
print(
f"is_contiguous: {data.is_contiguous()}, device: {data.device}, dtype: {data.dtype} shape {data.shape} stride {data.stride()}")
get
is_contiguous: True, device: cuda:0, dtype: torch.float32 shape torch.Size([1, 3, 224, 224]) stride (150528, 50176, 224, 1)
even I update the newest version of tritonclient can't work