Replies: 1 comment 2 replies
-
For achieving faster inference with YOLOv8 models, using a single RTX 4090 would generally offer better performance compared to multiple RTX 4060s, especially considering the higher memory bandwidth and compute capabilities of the 4090. This setup minimizes potential bottlenecks and complexities associated with multi-GPU configurations. Here's a simple example to set up your model for inference on a single GPU: from ultralytics import YOLO
# Load your model
model = YOLO('path_to_your_model.pt', device='cuda:0') # Specify GPU device
# Run inference
results = model.predict('path_to_image_or_video') Ensure your system has adequate cooling and power supply to handle the demands of the RTX 4090. If you decide later that you need more power, you can consider scaling up with additional GPUs. |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
If I have 1 4090 can i have multiple 4060(less memory) to run yolov8 m fp 16 model(p2 head) faster.
What setup will be faster.
All i care about is faster inference.
I am asking this because i have to make the purchase i don't have it yet so based on this i will buy.
ignore other bottleneck like CPU and all lets assume i can get the best CPU possible for the task.
please suggest any other solutions also, open to suggestions.
Thank you
Beta Was this translation helpful? Give feedback.
All reactions