nnfp/benchmark/performance.md

Inference Performance

Test Scripts

from towhee import ops
import time

decode = ops.audio_decode.ffmpeg()
audio = [x[0] for x in decode('path/to/test.wav')]

op = ops.audio_embedding.nnfp()
# op = ops.audio_embedding.nnfp(model_path='path/to/torchscript/model.pt')
# op = ops.audio_embedding.nnfp(model_path='path/to/model.onnx')


start = time.time()
for _ in range(100):
	embs = op(audio)
	assert(embs.shape == (10, 128))
end = time.time()

print((end-start) / 100)

Performance (Default model)

• Device: MacOS, 2.3 GHz Quad-Core Intel Core i7, 8 CPUs
• Input: 10s stereo audio, shape (2, 441344), sr 44100, loop for 100 times

inference method	mem usage	avg time
pytorch	0.3G	0.160s
torchscript	0.3G	0.167s
onnx	0.4G	0.093s

• Device: MacOS, 2.3 GHz Quad-Core Intel Core i7, 8 CPUs
• Input: 188s stereo audio, shape (2, 8328408), sr 44100, loop for 100 times

inference method	mem usage	avg time
pytorch	2.7G	2.29s
torchscript	2.5G	2.30s
onnx	1.3G	1.77s

• Device: MacOS, 2.3 GHz Quad-Core Intel Core i7, 8 CPUs
• Input: 600s mono audio, shape (1, 9600000), sr 16000, loop for 20 times

inference method	mem usage	avg time
pytorch	4.9G	9.04s
torchscript	4.9G	9.302s
onnx	2.7G	4.586s