how to use the feature "--prompt-cache" in llama-cli

[shenyunjason]

hello
I want to use the feature"--prompt-cache" in llama-cli(version b5593), the model is qwen2.5-0.5b-instruct-q8_0.gguf,
my command is "./llama-cli --no-warmup -m qwen2.5-0.5b-instruct-q8_0.gguf --prompt-cache prompt.bin ",
I do not find prompt.bin is created after some questions are finished, it seems that the feature does not work.
Is what I did right? Pls give me some helps.