Cache the Wasmer module in all compute workers. If a compute worker receives a request and doesn't have the corresponding module, it should notify the coordinator which will fetch and provide it from the host.
Also, now that you don't need to send the Module with each run_with command, switch to using channels to send requests to workers instead of post_message. This way the coordinator<>worker communication will generally avoid the JS event loop and be functionally instant.
Cache the Wasmer module in all compute workers. If a compute worker receives a request and doesn't have the corresponding module, it should notify the coordinator which will fetch and provide it from the host.
Also, now that you don't need to send the Module with each
run_withcommand, switch to using channels to send requests to workers instead of post_message. This way the coordinator<>worker communication will generally avoid the JS event loop and be functionally instant.