NVIDIA-AI-IOT / cuDLA-samples

YOLOv5 on Orin DLA
Other
167 stars 17 forks source link

How to run simultaneously on two DLAs #29

Open 2yjia opened 4 months ago

lynettez commented 4 months ago

You can create two cuDLA handle with API cudlaCreateDevice(0, &devHandle, CUDLA_CUDA_DLA); and set the DLA core index with 0 and 1, then load two DLA loadables and execute then on two CUDA Streams in CUDA hyrid mode.