GPU usage at 0

[brandaobrandisborges]

Specs:

• Jetson Nano
• CUDA 10.2
• GPU: 128-core NVIDIA Maxwell™ architecture-based GPU.
• CPU: Quad-core ARM® A57.

Code implementation:

import spacy
import thinc
gpu_t = spacy.require_gpu(gpu_id=0)
print(gpu_t) #returns True
nlp = spacy.load("en_core_web_md")
main_doc = nlp("lorem ipsum dolor sit amet consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et")
search_doc = nlp("lorem ipsum dolor sit amet consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et")
_similarity = main_doc.similarity(search_doc)
print(_similarity)

The code above takes around 10 seconds to run, and it doesn't change GPU usage, whilist one of my CPU cores stays at 100% during processing.

Am I doing anything wrong?

[brandaobrandisborges]

Everything is fine. I have been able to compare the results and the CPU usage is to load the model. Also, when switching to between GPU/CPU I can notice a performance difference, CPU being faster for this small task.