ML Service: Performance benchmarks |
|
Issue descriptionLet's get some very basic numbers on ML Service performance: Latency: We can ping/pong repeated calls from Chromium on a tiny test model. Throughput: We can see how many QPS can we shovel e.g. 4K video at. Probably need a larger test model.
,
Dec 10
Latency is usually under 1ms, with occasional spikes into single-digit milliseconds. Per rockot@ this is reasonable for Mojo. |
|
►
Sign in to add a comment |
|
Comment 1 by amoylan@chromium.org
, Oct 9