Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.
And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.
High memory usage Environments
See original GitHub issueIm working with an environment that has a very high memory usage. This usually prevents any sort of Async Sampling, since copies of the environment are very expensive. Is there any example of working with a high memory usage environment for ASync Sampling?
My understanding is that each state needs to be hashable, so it could be that because of the large state space - Im running quickly out of memory on instances with 500GB+ of RAM
So, a good question might be - when sample_async or high-throughput architectures are used, what data is duplicated?
Issue Analytics
- State:
- Created 5 years ago
- Reactions:1
- Comments:21 (6 by maintainers)
Top Results From Across the Web
Top Related Medium Post
Top Related StackOverflow Question
Troubleshoot Live Code
Top Related Reddit Thread
Top Related Hackernoon Post
Top Related Tweet
Top Related Dev.to Post
Top Related Hashnode Post
Actually, increasing the plasma store size works for fixing that!
@goodcheer I created a plasma store and indexed out of it. But its still pretty complex - some objects when deserialized create copies, which will blow the memory - https://github.com/ray-project/ray/issues/3881
But if you’re data is in numpy arrays, the plasma store works really well to solve this problem