Jia Wei

@Januszl Thank you very much. After adding `import nvidia.dali.types as types` and your suggested code, my code can run fine. I noticed that the tensor I stored earlier are torch.cuda.FloatTensor...

@JanuszL I use` Inputimages = torch.empty(file[1], dtype=torch_type, device=device)` to replace` Inputimages = torch.zeros(file[1], dtype=torch_type).to(device)`. Unfortunately, Then, I used DALI and torch.load() to read the feature map from SSD to GPU...

Hi@JanuszL We tried using kvikio for data transfer from SSD to GPU and it seems that kvikio.Cufile() also has a large build overhead in write mode ("w"). In read mode...

@JanuszL We are not currently using the Nsight System profiler and it will take us some time to learn and try out the tool, if you could help us with...

Hi@azrael417 @JanuszL Thank you very much. For your first question, the output of `gdschecker -P` are:  I'm not sure if this output means that the system is compatible. I'm...

Hi @azrael417 , Thanks again, Thorsten. We still have some questions that we would like to get your help with. 1. We do have GDS installed, we use `gdscheck `and...

@azrael417 @JanuszL We have done more experiments on DALI, comparing **device='cpu'** and **decive='gpu'** under the same experimental environment and pipeline settings, and the results are shown in the table below....

Hi @azrael417 , Thank you for your reply. We found that the reason for the NVMe unsupported error may be due to the linux kernel version. We found that the...

Hi, @azrael417 After communicating with the GDS engineer via email, our GDS environment now appears to be working properly. This is shown in the figure below.  However, when we...

Hi, @azrael417 The files are all on the Intel Optane SSD which I mentioned before, I think it is a supported NVME drive. I run gdsio to read file by...