3 ms·
Maybe one day their inference could be running on photonics hardware! Or imagine for the code analysis they replaced phyton running on kubernetes with wasm eng
by singularity2001 3y ago
Maybe one day their inference could be running on photonics hardware!
Or imagine for the code analysis they replaced phyton running on kubernetes with wasm engines. would need a small wast to wasm compiler and interfaces for simple io and plotting, but many analytics tasks could probably run much faster.
- ilaksh 3y agoI think the inferencing is using many orders of magnitude more compute than the occasional analytics tasks. But I assume the containers do need some significant resources. As far as making analysis lighter weight, I think that something like you say will eventually be the way they go. For my own agent hosting of agent-written code, I moved to Rust and have been incorporating Rhai scripting which is vastly more efficient than containers.