Hi,
I couldn’t find the right forum to post this, so please let me know if this post should be moved.
We’re considering purchasing one of two RTX embedded GPUs (NVIDIA RTX Embedded GPU Solutions | NVIDIA):
- NVIDIA RTX 5000 Ada Generation
- NVIDIA RTX 2000 Ada Generation
We would like to use these for object detection (e.g. YOLO models). How can we estimate / calculate the FPS (frames-per-second) for, say, a YOLOv4 object detection model on this GPU without having the hardware?
For example, if we know the image size(s), model size/precision, and other parameters, does NVIDIA provide a tool (maybe an emulator?) to get benchmarks estimates for these types of inference workloads on a particular GPU?
Even a rough benchmark would be very helpful.
This will help us determine which GPU to purchase for our product.
Thank you!