Description
Recently, while running tasks with DeepAgentCode, I observed a dramatic decrease in cache hit rate. The current ratio of input hit : input miss : output is approximately 500 : 500 : 2. Previously, for the majority of runs, the ratio was about 1180 : 12 : 3. This large number of input cache misses has caused a significant increase in overhead. I would like to ask the maintainers to investigate this regression.
The problem appeared recently; earlier runs with the same workload did not show this behaviour. No code changes were made on my side. It might be related to a recent update of DeepAgentCode or cache invalidation logic.
Plugins
No response
DeepAgent Code version
No response
Steps to reproduce
No response
Screenshot and/or share link
No response
Operating System
No response
Terminal
No response
Description
Recently, while running tasks with DeepAgentCode, I observed a dramatic decrease in cache hit rate. The current ratio of input hit : input miss : output is approximately 500 : 500 : 2. Previously, for the majority of runs, the ratio was about 1180 : 12 : 3. This large number of input cache misses has caused a significant increase in overhead. I would like to ask the maintainers to investigate this regression.
The problem appeared recently; earlier runs with the same workload did not show this behaviour. No code changes were made on my side. It might be related to a recent update of DeepAgentCode or cache invalidation logic.
Plugins
No response
DeepAgent Code version
No response
Steps to reproduce
No response
Screenshot and/or share link
No response
Operating System
No response
Terminal
No response