You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thus it appears that the throughput of each task class will drop, the more task classes we pack in an event loop.
Maybe we need to use multiple eventloops; one per task class.(maybe upto a maximum number of eventloops)
If we also go down this road; we also need to benchmark the added memory consumption per eventloop
look at the benchmarks in;
wiji/tests/test_worker.py
Lines 534 to 847 in a315334
When we only have one unique task class in eventloop the processing rate using an InMemoryBroker is
7520 tasks/second
When we have two uniqe task classes the rate drops to;
4237 tasks/second
for three the rate is;
2816 tasks/second
for four;
2157 tasks/second
for eight;
1120 tasks/second
The text was updated successfully, but these errors were encountered: