Jetson DLA vs. open-sourced NVDLA

Autonomous Machines Jetson & Embedded Systems Jetson AGX Orin
1

What’s the relationship between those two? Is the DLA in Jetson an internal or second version of the open-sourced NVDLA, or they are two different things? It seems the nvdla · GitHub without updates for several years.

Thanks,
Rex

2

I can’t answer that, but there is something you’d be interested in related to this: The GPU of a Jetson is integrated directly to the memory controller (an iGPU). The GPU of almost all other hardware out there is a discrete GPU on the PCIe bus (dGPU). Software which needs to find the GPU is almost always using nvidia-smi, but this only works on PCI GPUs. None of that software, even if the correct architecture, can work with a Jetson due to detection. So far as I know the Jetson version is the same as the others, except that it is able to detect and find the iGPU.

3

Thanks for your answer but I am not sure we are on the same page. I am asking the Deep Learning Accelerator, and it seems you are talking GPU.

5

Hi,

Yes, it indicates the same DLA accelerator.

There are two generations of DLA, Orin is equipped with DLA v2.
But the GitHub you mentioned is the first version of DLA, which is used on the Xavier.

Thanks.

1 Like
6

The relevance of the iGPU versus dGPU is that GPU software, in general, including DLA, needs to find the GPU to talk to it. If you are speaking of DLA running on CPU, then it won’t matter.

8

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.